Evidence inventory: entry x configuration x output

Generated by python scripts/build_evidence_inventory.py from statspai.validation_scope.SCOPES. Do not edit by hand; --check is the drift gate.

Each estimator below has a configuration grid: one value per dimension. The table counts the cells in which an output has reference evidence, meaning a known-truth (T1) or same-byte cross-language (T2) artifact ran that configuration and compared that output. A function-level tier says nothing about which cells those are. For one fitted result, call sp.validation_scope(result).

Functions without a scope map are not listed. That is not a claim that they lack evidence: they carry function-level evidence only (sp.parity_status).

Cells with reference evidence, by output

Function	Cells	estimate	se	vcov	joint_test	diagnostic	interval	Fully covered
aipw	512	32 / 512	17 / 512	--	--	--	--	17 / 512
callaway_santanna	69120	1380 / 69120	686 / 69120	1 / 69120	257 / 69120	--	--	686 / 69120
causal_forest	96	0 / 96	0 / 96	--	--	--	--	0 / 96
did_2x2	80	20 / 80	20 / 80	--	--	--	3 / 80	20 / 80
did_imputation	240	36 / 240	13 / 240	1 / 240	--	--	--	12 / 240
did_multiplegt_dyn	18432	136 / 18432	37 / 18432	26 / 18432	36 / 18432	--	--	37 / 18432
dml	9120	33 / 9120	30 / 9120	--	--	--	--	30 / 9120
ebalance	16	10 / 16	3 / 16	--	--	--	--	3 / 16
etwfe	128	16 / 128	16 / 128	3 / 128	--	--	--	16 / 128
etwfe_glm	2880	51 / 2880	36 / 2880	--	--	--	--	36 / 2880
event_study	1536	138 / 1536	138 / 1536	3 / 1536	--	--	3 / 1536	138 / 1536
fast.feols	12	12 / 12	6 / 12	--	--	--	--	6 / 12
feols	144	72 / 144	29 / 144	--	--	--	--	27 / 144
g_computation	48	16 / 48	5 / 48	--	--	--	--	5 / 48
gardner_did	64	24 / 64	7 / 64	1 / 64	--	--	--	7 / 64
glm	320	144 / 320	32 / 320	--	--	--	--	32 / 320
heckman	12	9 / 12	6 / 12	--	--	--	--	6 / 12
honest_did	116640	--	--	--	--	159 / 116640	--	159 / 116640
ipw	480	28 / 480	12 / 480	--	--	--	--	12 / 480
iv	448	224 / 448	63 / 448	--	3 / 448	3 / 448	--	63 / 448
ivreg	448	224 / 448	63 / 448	--	3 / 448	3 / 448	--	63 / 448
logit	20	10 / 20	10 / 20	--	3 / 20	--	--	10 / 20
nbreg	12	9 / 12	7 / 12	--	--	--	--	7 / 12
nnmatch	1920	17 / 1920	17 / 1920	--	--	--	--	17 / 1920
panel	144	99 / 144	46 / 144	--	5 / 144	--	--	46 / 144
poisson	20	15 / 20	12 / 20	--	2 / 20	--	--	12 / 20
ppmlhdfe	72	54 / 72	9 / 72	--	--	--	--	9 / 72
probit	20	10 / 20	6 / 20	--	2 / 20	--	--	6 / 20
psm	3200	100 / 3200	29 / 3200	--	--	--	--	29 / 3200
qreg	14	14 / 14	9 / 14	2 / 14	--	--	--	9 / 14
rddensity	2	1 / 2	1 / 2	--	--	1 / 2	--	1 / 2
rdrobust	25920	159 / 25920	159 / 25920	--	--	--	--	159 / 25920
regress	28	28 / 28	26 / 28	--	5 / 28	1 / 28	--	26 / 28
sdid	288	60 / 288	27 / 288	--	--	--	--	27 / 288
sun_abraham	108	63 / 108	63 / 108	6 / 108	--	--	4 / 108	63 / 108
synth	12	3 / 12	1 / 12	--	--	--	--	3 / 12
tmle	672	41 / 672	37 / 672	--	--	--	--	37 / 672
tobit	36	18 / 36	13 / 36	--	--	--	--	13 / 36
xtabond	88704	74 / 88704	42 / 88704	--	--	20 / 88704	--	42 / 88704
xtdpdsys	72576	30 / 72576	18 / 72576	--	--	8 / 72576	--	18 / 72576

-- means no artifact compares that output for the function at all. Fully covered counts cells whose primary outputs all have reference evidence.

Other evidence statuses

Cells whose strongest evidence for an output is not a reference row. These are reported at their own grade and are not parity.

Function	Output	seed_equivalence (T3)	stochastic_screen (S)	coverage_simulation (B)	disclosure (T4)	not_covered
aipw	estimate	0	0	0	0	480
aipw	se	0	0	0	0	495
callaway_santanna	estimate	0	0	0	0	67740
callaway_santanna	se	0	4	0	0	68430
callaway_santanna	coverage	0	0	1	0	69119
callaway_santanna	vcov	0	0	0	0	69119
callaway_santanna	joint_test	0	0	0	0	68863
causal_forest	estimate	3	0	0	0	93
causal_forest	se	0	1	0	0	95
causal_forest	coverage	0	0	1	0	95
did_2x2	estimate	0	0	0	0	60
did_2x2	se	0	0	0	0	60
did_2x2	interval	0	0	0	0	77
did_imputation	estimate	0	0	0	3	201
did_imputation	se	0	0	0	0	227
did_imputation	vcov	0	0	0	0	239
did_multiplegt_dyn	estimate	0	0	0	0	18296
did_multiplegt_dyn	se	0	1	0	0	18394
did_multiplegt_dyn	vcov	0	0	0	0	18406
did_multiplegt_dyn	joint_test	0	0	0	0	18396
dml	estimate	0	0	0	0	9087
dml	se	0	0	0	3	9087
dml	coverage	0	0	2	0	9118
ebalance	estimate	0	0	0	0	6
ebalance	se	0	0	0	3	10
etwfe	estimate	0	0	0	0	112
etwfe	se	0	0	0	0	112
etwfe	vcov	0	0	0	0	125
etwfe_glm	estimate	0	0	0	0	2829
etwfe_glm	se	0	0	0	4	2840
event_study	estimate	0	0	0	0	1398
event_study	se	0	0	0	0	1398
event_study	vcov	0	0	0	0	1533
event_study	interval	0	0	0	0	1533
fast.feols	estimate	0	0	0	0	0
fast.feols	se	0	0	0	0	6
fast.feols	coverage	0	0	1	0	11
feols	estimate	0	0	0	0	72
feols	se	0	0	0	0	115
g_computation	estimate	0	0	0	0	32
g_computation	se	0	0	0	0	43
gardner_did	estimate	0	0	0	0	40
gardner_did	se	0	0	0	0	57
gardner_did	vcov	0	0	0	0	63
glm	estimate	0	0	0	0	176
glm	se	0	0	0	0	288
heckman	estimate	0	0	0	0	3
heckman	se	0	0	0	0	6
honest_did	diagnostic	0	30	0	105	116346
ipw	estimate	0	0	0	0	452
ipw	se	0	0	0	0	468
iv	estimate	0	0	0	0	224
iv	se	0	0	0	7	378
iv	coverage	0	0	1	0	447
iv	diagnostic	0	0	0	0	445
iv	joint_test	0	0	0	0	445
ivreg	estimate	0	0	0	0	224
ivreg	se	0	0	0	7	378
ivreg	coverage	0	0	1	0	447
ivreg	diagnostic	0	0	0	0	445
ivreg	joint_test	0	0	0	0	445
logit	estimate	0	0	0	0	10
logit	se	0	0	0	0	10
logit	joint_test	0	0	0	0	17
nbreg	estimate	0	0	0	0	3
nbreg	se	0	0	0	0	5
nnmatch	estimate	0	0	0	0	1903
nnmatch	se	0	0	0	0	1903
panel	estimate	0	0	0	0	45
panel	se	0	0	0	0	98
panel	joint_test	0	0	0	1	138
poisson	estimate	0	0	0	0	5
poisson	se	0	0	0	0	8
poisson	joint_test	0	0	0	0	18
ppmlhdfe	estimate	0	0	0	0	18
ppmlhdfe	se	0	0	0	0	63
probit	estimate	0	0	0	0	10
probit	se	0	0	0	4	10
probit	joint_test	0	0	0	0	18
psm	estimate	0	0	0	0	3100
psm	se	0	0	0	2	3169
qreg	estimate	0	0	0	0	0
qreg	se	0	0	0	0	5
qreg	vcov	0	0	0	0	12
rddensity	estimate	0	0	0	0	1
rddensity	se	0	0	0	0	1
rddensity	diagnostic	0	0	0	0	1
rdrobust	estimate	0	0	0	0	25761
rdrobust	se	0	0	0	0	25761
rdrobust	coverage	0	0	1	0	25919
regress	estimate	0	0	0	0	0
regress	se	0	0	0	0	2
regress	coverage	0	0	1	0	27
regress	diagnostic	0	0	0	0	27
regress	joint_test	0	0	0	0	23
sdid	estimate	0	0	0	0	228
sdid	se	0	11	0	0	250
sdid	coverage	0	0	1	0	287
sun_abraham	estimate	0	0	0	0	45
sun_abraham	se	0	0	0	0	45
sun_abraham	coverage	0	0	1	0	107
sun_abraham	vcov	0	0	0	0	102
sun_abraham	interval	0	0	0	0	104
synth	estimate	0	0	0	1	8
synth	se	0	0	0	0	11
tmle	estimate	0	0	0	0	631
tmle	se	0	0	0	2	633
tobit	estimate	0	0	0	0	18
tobit	se	0	0	0	0	23
xtabond	estimate	0	0	0	0	88630
xtabond	se	0	0	0	0	88662
xtabond	diagnostic	0	0	0	0	88684
xtdpdsys	estimate	0	0	0	0	72546
xtdpdsys	se	0	0	0	0	72558
xtdpdsys	diagnostic	0	0	0	0	72568

Artifacts per function

aipw

Dimensions: estimand in {ate, att}; cross_fit in {true, false}; se_method in {influence, sandwich}; outcome_model in {linear, logit, probit, poisson}; propensity in {logit, known}; trim in {inactive, binding}; weights in {none, set}; cluster in {none, set}.

Kind	Outputs	Configuration run	Artifact	Entry point
T2	estimate, se	estimand=ate; cross_fit=false; se_method=sandwich; outcome_model=linear; propensity=logit; trim=inactive; weights=none/set; cluster=none/set	tests/reference_parity/test_teffects_design_stata_parity.py	sp.aipw(...)
T2	estimate, se	estimand=ate; cross_fit=false; se_method=sandwich; outcome_model=logit/poisson/probit; propensity=logit; trim=inactive; weights=none/set; cluster=none	tests/reference_parity/test_aipw_outcome_models_stata_parity.py	sp.aipw(...)
T2	estimate, se	estimand=ate; cross_fit=false; se_method=sandwich; outcome_model=logit/poisson/probit; propensity=logit; trim=inactive; weights=none; cluster=set	tests/reference_parity/test_aipw_outcome_models_stata_parity.py	sp.aipw(...)
T2	estimate, se	estimand=ate; cross_fit=false; se_method=influence; outcome_model=linear; propensity=logit; trim=inactive; weights=none; cluster=none	tests/reference_parity/test_teffects_R_parity.py	sp.aipw(...)
T1	estimate, se	estimand=att; cross_fit=false; se_method=influence; outcome_model=linear; propensity=logit; trim=inactive; weights=none; cluster=none	tests/reference_parity/test_teffects_R_parity.py	sp.aipw(...)
T1	estimate, se	estimand=ate/att; cross_fit=true; se_method=influence; outcome_model=linear; propensity=logit; trim=inactive; weights=none; cluster=none	tests/reference_parity/test_aipw_crossfit_identity.py	sp.aipw(...)
T2	estimate, se	estimand=ate; cross_fit=false; se_method=influence; outcome_model=linear; propensity=logit; trim=inactive; weights=none; cluster=none	tests/reference_parity/test_yuksel_aydede_causal_ml_parity.py	sp.aipw(...)

se_method is ignored for estimate: the point estimate is computed before, and independently of, the variance estimator

callaway_santanna

Dimensions: sample in {panel, unbalanced_panel, rcs, unbalanced_pair}; estimator in {dr, drimp, reg, ipw, ipw_abadie}; control_group in {nevertreated, notyettreated}; notyet_cutoff in {period, asinr, cohort}; weights in {none, weighted}; covariates in {none, set}; inference in {analytic, bootstrap}; base_period in {universal, varying}; anticipation in {0, >0}; clustering in {none, set}; parity in {none, r, stata}; variance_convention in {na, delta, reference}.

Kind	Outputs	Configuration run	Artifact	Entry point
T2	estimate, se	sample=panel; estimator=reg; control_group=nevertreated; notyet_cutoff=period; weights=none; covariates=none; inference=analytic; base_period=universal; anticipation=0; clustering=none; parity=none; variance_convention=na	tests/r_parity/04_csdid.py	sp.callaway_santanna(estimator='reg') + sp.aggte(bstrap=False)
T2	estimate, se	sample=panel; estimator=dr/ipw/reg; control_group=nevertreated/notyettreated; notyet_cutoff=period; weights=none/weighted; covariates=none/set; inference=analytic; base_period=universal; anticipation=0; clustering=none; parity=none; variance_convention=na	tests/reference_parity/test_cs_weighted_parity.py	sp.callaway_santanna(...)
T2	estimate, se	sample=panel; estimator=dr; control_group=nevertreated; notyet_cutoff=period; weights=none; covariates=none; inference=analytic; base_period=varying; anticipation=0; clustering=none; parity=none; variance_convention=na	tests/reference_parity/test_aggte_mpdta_parity.py	sp.aggte(cs_result, type=agg_type, bstrap=False)
T2	vcov	sample=panel; estimator=dr; control_group=nevertreated; notyet_cutoff=period; weights=none; covariates=none; inference=analytic; base_period=universal; anticipation=0; clustering=none; parity=none; variance_convention=na	tests/reference_parity/test_aggte_vcov_r_parity.py	sp.aggte(type='dynamic')
T2	estimate, se	sample=panel; estimator=dr; control_group=nevertreated; notyet_cutoff=period; weights=none; covariates=none; inference=analytic; base_period=universal/varying; anticipation=0/>0; clustering=none; parity=none; variance_convention=na	tests/reference_parity/test_cs_gaps_parity.py	sp.callaway_santanna(panel, ...)
T2	estimate, se	sample=unbalanced_panel; estimator=dr/ipw/reg; control_group=nevertreated; notyet_cutoff=period; weights=none; covariates=none/set; inference=analytic; base_period=varying; anticipation=0; clustering=none; parity=none; variance_convention=na/reference	tests/reference_parity/test_cs_gaps_parity.py	sp.callaway_santanna(unbalanced, ...)
T2	estimate, se	sample=panel/rcs; estimator=dr/reg; control_group=notyettreated; notyet_cutoff=period; weights=none; covariates=none; inference=analytic; base_period=universal/varying; anticipation=0/>0; clustering=none; parity=none; variance_convention=na	tests/reference_parity/test_cs_no_never_unbalanced_parity.py	sp.callaway_santanna(df, **kw)
T2	estimate, se	sample=unbalanced_panel; estimator=dr/reg; control_group=nevertreated/notyettreated; notyet_cutoff=period; weights=none; covariates=none; inference=analytic; base_period=varying; anticipation=0; clustering=none; parity=none; variance_convention=na	tests/reference_parity/test_cs_no_never_unbalanced_parity.py	sp.callaway_santanna(df, **kw)
T2	estimate, se	sample=panel/rcs; estimator=dr; control_group=notyettreated; notyet_cutoff=period; weights=none; covariates=none/set; inference=analytic; base_period=universal/varying; anticipation=0/>0; clustering=none; parity=none; variance_convention=na/reference	tests/reference_parity/test_cs_notyet_cutoff_parity.py	sp.callaway_santanna(
T2	estimate, se	sample=rcs; estimator=dr/ipw/reg; control_group=nevertreated/notyettreated; notyet_cutoff=period; weights=none; covariates=set; inference=analytic; base_period=varying; anticipation=0; clustering=none; parity=none; variance_convention=na/reference	tests/reference_parity/test_cs_rcs_parity.py	sp.callaway_santanna(
T2	estimate	sample=rcs; estimator=dr; control_group=nevertreated/notyettreated; notyet_cutoff=period; weights=none; covariates=set; inference=analytic; base_period=varying; anticipation=0; clustering=none; parity=none; variance_convention=delta	tests/reference_parity/test_cs_rcs_parity.py	sp.callaway_santanna(panel=False, estimator='dr', x=...)
T2	estimate	sample=rcs; estimator=reg; control_group=nevertreated; notyet_cutoff=period; weights=none; covariates=none; inference=analytic; base_period=universal; anticipation=0; clustering=none; parity=none; variance_convention=na	tests/reference_parity/test_rcs_aggregation_conventions.py	sp.callaway_santanna(
T2	estimate	sample=panel; estimator=reg; control_group=notyettreated; notyet_cutoff=asinr/cohort/period; weights=none; covariates=none; inference=analytic; base_period=universal; anticipation=0; clustering=none; parity=none; variance_convention=na	tests/reference_parity/test_csdid_conventions_stata_parity.py	sp.callaway_santanna(
T2	estimate	sample=panel; estimator=ipw/ipw_abadie; control_group=nevertreated; notyet_cutoff=period; weights=none; covariates=set; inference=analytic; base_period=universal; anticipation=0; clustering=none; parity=none; variance_convention=na	tests/reference_parity/test_csdid_conventions_stata_parity.py	sp.callaway_santanna(
T2	estimate, se	sample=panel; estimator=reg; control_group=notyettreated; notyet_cutoff=asinr/cohort; weights=none; covariates=none; inference=analytic; base_period=universal; anticipation=0; clustering=none; parity=none; variance_convention=na	tests/reference_parity/test_csdid_options_stata_parity.py	sp.callaway_santanna(
T2	estimate, se	sample=panel; estimator=ipw_abadie; control_group=nevertreated; notyet_cutoff=period; weights=none; covariates=set; inference=analytic; base_period=universal; anticipation=0; clustering=none; parity=none; variance_convention=na	tests/reference_parity/test_csdid_options_stata_parity.py	sp.callaway_santanna(
T2	estimate, se	sample=unbalanced_pair; estimator=dr; control_group=nevertreated; notyet_cutoff=period; weights=none; covariates=set; inference=analytic; base_period=universal; anticipation=0; clustering=none; parity=none; variance_convention=na	tests/reference_parity/test_csdid_options_stata_parity.py	sp.callaway_santanna(
T2	estimate, se	sample=unbalanced_pair; estimator=reg; control_group=notyettreated; notyet_cutoff=cohort; weights=none; covariates=none; inference=analytic; base_period=universal; anticipation=0; clustering=none; parity=none; variance_convention=na	tests/reference_parity/test_csdid_options_stata_parity.py	sp.callaway_santanna(
T2	estimate, se	sample=panel/rcs; estimator=drimp/ipw_abadie; control_group=nevertreated; notyet_cutoff=period; weights=none; covariates=set; inference=analytic; base_period=universal; anticipation=0; clustering=none; parity=none; variance_convention=na	tests/reference_parity/test_cs_options_R_parity.py	sp.callaway_santanna(
T2	estimate, se	sample=panel; estimator=drimp; control_group=notyettreated; notyet_cutoff=period; weights=none; covariates=set; inference=analytic; base_period=varying; anticipation=0; clustering=none; parity=none; variance_convention=na	tests/reference_parity/test_cs_options_R_parity.py	sp.callaway_santanna(
T2	estimate, se	sample=panel; estimator=ipw; control_group=nevertreated; notyet_cutoff=period; weights=none; covariates=set; inference=analytic; base_period=universal; anticipation=0; clustering=set; parity=none; variance_convention=na	tests/reference_parity/test_cs_options_R_parity.py	sp.callaway_santanna(
T2	estimate, se	sample=rcs; estimator=ipw; control_group=nevertreated; notyet_cutoff=period; weights=none; covariates=set; inference=analytic; base_period=universal; anticipation=0; clustering=none; parity=none; variance_convention=na	tests/reference_parity/test_cs_options_R_parity.py	sp.callaway_santanna(
T2	estimate, se	sample=panel; estimator=dr; control_group=nevertreated; notyet_cutoff=period; weights=weighted; covariates=set; inference=analytic; base_period=universal; anticipation=>0; clustering=none; parity=none; variance_convention=na	tests/reference_parity/test_cs_options_R_parity.py	sp.callaway_santanna(
T2	estimate, se	sample=panel; estimator=reg; control_group=notyettreated; notyet_cutoff=period; weights=none; covariates=none; inference=analytic; base_period=varying; anticipation=>0; clustering=set; parity=none; variance_convention=na	tests/reference_parity/test_cs_options_R_parity.py	sp.callaway_santanna(
T2	estimate, se	sample=panel/rcs; estimator=dr; control_group=nevertreated; notyet_cutoff=period; weights=none; covariates=none; inference=analytic; base_period=universal; anticipation=0; clustering=set; parity=none; variance_convention=na	tests/reference_parity/test_cs_rc_cluster_csdid_parity.py	sp.callaway_santanna(clustervars=...)
T2	estimate, se	sample=rcs; estimator=dr; control_group=nevertreated; notyet_cutoff=period; weights=weighted; covariates=none; inference=analytic; base_period=universal; anticipation=0; clustering=none/set; parity=none; variance_convention=na	tests/reference_parity/test_cs_rc_weighted_aggte_did_parity.py	sp.callaway_santanna(panel=False, weights=...) + sp.aggte
T2	estimate, se	sample=panel; estimator=dr; control_group=nevertreated/notyettreated; notyet_cutoff=period; weights=none; covariates=none; inference=analytic; base_period=universal; anticipation=0; clustering=none; parity=none; variance_convention=na	tests/reference_parity/test_cs_balance_full_did_parity.py	sp.callaway_santanna(balance='full')
T2	estimate, se	sample=panel/unbalanced_panel; estimator=dr; control_group=nevertreated/notyettreated; notyet_cutoff=period; weights=none; covariates=none; inference=analytic; base_period=universal/varying; anticipation=0; clustering=set; parity=none; variance_convention=na	tests/reference_parity/test_cs_cluster_analytic_parity.py	sp.callaway_santanna(clustervars=...)
T2	estimate, se	sample=panel; estimator=dr; control_group=nevertreated/notyettreated; notyet_cutoff=period; weights=none; covariates=set; inference=analytic; base_period=universal/varying; anticipation=0; clustering=none; parity=none; variance_convention=na	tests/reference_parity/test_stata_did_commands_parity.py	sp.callaway_santanna(
T4	estimate, se	sample=panel; estimator=dr/drimp; control_group=nevertreated/notyettreated; notyet_cutoff=period; weights=none; covariates=set; inference=analytic; base_period=universal/varying; anticipation=0; clustering=none; parity=none; variance_convention=na	tests/reference_parity/test_stata_did_commands_parity.py	sp.stata(
T2	estimate, se, joint_test	sample=panel; estimator=dr; control_group=nevertreated/notyettreated; notyet_cutoff=period; weights=none; covariates=none/set; inference=analytic; base_period=varying; anticipation=0; clustering=none/set; parity=r; variance_convention=na	tests/reference_parity/test_cs_parity_switch.py	sp.callaway_santanna(
T2	estimate, se, joint_test	sample=panel; estimator=dr; control_group=nevertreated; notyet_cutoff=period; weights=none; covariates=none; inference=analytic; base_period=varying; anticipation=>0; clustering=none; parity=r; variance_convention=na	tests/reference_parity/test_cs_parity_switch.py	sp.callaway_santanna(
T2	estimate, se, joint_test	sample=panel; estimator=dr; control_group=nevertreated/notyettreated; notyet_cutoff=cohort/period; weights=none; covariates=none/set; inference=analytic; base_period=varying; anticipation=0; clustering=none/set; parity=stata; variance_convention=na	tests/reference_parity/test_cs_parity_switch.py	sp.callaway_santanna(
T2	estimate, se	sample=panel; estimator=dr; control_group=nevertreated; notyet_cutoff=period; weights=none; covariates=none; inference=bootstrap; base_period=varying; anticipation=0; clustering=none; parity=stata; variance_convention=na	tests/reference_parity/test_cs_parity_switch.py	sp.callaway_santanna(
T2	estimate, se, joint_test	sample=panel/rcs/unbalanced_panel; estimator=dr; control_group=nevertreated/notyettreated; notyet_cutoff=period; weights=none/weighted; covariates=none/set; inference=analytic; base_period=varying; anticipation=0; clustering=none/set; parity=r; variance_convention=na/reference	tests/reference_parity/test_cs_parity_rc_unbal.py	sp.callaway_santanna(
T2	estimate, se, joint_test	sample=rcs/unbalanced_pair; estimator=dr; control_group=nevertreated/notyettreated; notyet_cutoff=cohort/period; weights=none/weighted; covariates=none/set; inference=analytic; base_period=varying; anticipation=0; clustering=none/set; parity=stata; variance_convention=na/reference	tests/reference_parity/test_cs_parity_rc_unbal.py	sp.callaway_santanna(
T2	estimate, se, joint_test	sample=panel; estimator=dr; control_group=nevertreated/notyettreated; notyet_cutoff=cohort/period; weights=weighted; covariates=none/set; inference=analytic; base_period=varying; anticipation=0; clustering=none/set; parity=stata; variance_convention=na	tests/reference_parity/test_cs_parity_rc_unbal.py	sp.callaway_santanna(
S	se	sample=panel; estimator=dr; control_group=nevertreated; notyet_cutoff=period; weights=none; covariates=none; inference=bootstrap; base_period=universal; anticipation=0; clustering=none/set; parity=none; variance_convention=na	tests/reference_parity/test_cs_inference_parity.py	sp.callaway_santanna(bstrap=True)
B	coverage	sample=panel; estimator=reg; control_group=nevertreated; notyet_cutoff=period; weights=none; covariates=none; inference=analytic; base_period=universal; anticipation=0; clustering=none; parity=none; variance_convention=na	tests/coverage_monte_carlo/results_b1000/coverage_b1000.json	sp.callaway_santanna(estimator='reg')

inference is ignored for estimate: the multiplier bootstrap resamples influence functions for SEs and bands; the ATT(g,t) and their aggregates are unchanged

base_period is ignored for estimate, se: the headline is the simple aggregate of the post-treatment cells, whose base period is g - 1 - anticipation under both schemes; only the pre-treatment placebos (and so the event-study covariance) differ (asserted equal in tests/test_validation_scope.py)

causal_forest

Dimensions: treatment in {binary, continuous}; trees in {100, 200, 300, 500, 1000, 2000, 4000, 8000}; tuning in {grf_defaults, custom}; design in {iid, clustered, fixed_effects}.

Kind	Outputs	Configuration run	Artifact	Entry point
T3	estimate	treatment=binary; trees=2000/500/8000; tuning=grf_defaults; design=iid	tests/reference_parity/test_grf_seed_mc_equivalence.py	sp.causal_forest(...).average_treatment_effect()
S	estimate, se	treatment=binary; trees=2000; tuning=grf_defaults; design=iid	tests/r_parity/13_causal_forest.py	sp.causal_forest(n_estimators=2000)
B	coverage	treatment=binary; trees=2000; tuning=grf_defaults; design=iid	tests/coverage_monte_carlo/results_b1000/coverage_b1000.json	sp.causal_question(design='causal_forest')

did_2x2

Dimensions: vce in {classical, hc1, cluster, wild, survey}; weights in {none, set}; covariates in {none, set}; crit_dist in {t_resid, t_cluster, t_design, bootstrap}.

Kind	Outputs	Configuration run	Artifact	Entry point
T2	estimate, se	vce=hc1; weights=set; covariates=none; crit_dist=t_resid	tests/reference_parity/test_did2x2_ddd_weighted_robust_parity.py	sp.did_2x2(
T2	estimate, se	vce=survey; weights=none; covariates=none; crit_dist=t_design	tests/reference_parity/test_survey_did_parity.py	sp.did_2x2(
T2	estimate, se	vce=survey; weights=none; covariates=none; crit_dist=t_design	tests/reference_parity/test_survey_replicate_parity.py	sp.did_2x2(
T2	estimate, se, interval	vce=cluster; weights=none/set; covariates=none; crit_dist=t_cluster	tests/reference_parity/test_es_crit_parity.py	sp.did_2x2(
T2	estimate, se, interval	vce=hc1; weights=none; covariates=none; crit_dist=t_resid	tests/reference_parity/test_es_crit_parity.py	sp.did_2x2(

crit_dist is ignored for estimate, se: the critical value turns the SE into a p-value and an interval; the ATT and its SE are computed before it is chosen (asserted equal across parity= / crit_dist= in tests/reference_parity/test_es_crit_default.py; crit_dist='t' there is the same fit as parity='stata')

did_imputation

Dimensions: fe in {unit_time, time, none, custom}; covariates in {none, controls, time_covariates, unit_covariates, mixed}; vce in {analytic, bootstrap, none}; weights in {none, set}; horizon in {overall, horizons}.

Kind	Outputs	Configuration run	Artifact	Entry point
T2	estimate	vce=analytic; weights=none; horizon=overall; fe=unit_time; covariates=none	tests/r_parity/16_bjs.py	sp.did_imputation()
T2	estimate	vce=analytic; weights=none; horizon=overall; fe=none/time/unit_time; covariates=none	tests/reference_parity/test_bjs_fe_covariates_parity.py	sp.did_imputation(fe=...)
T2	estimate	vce=analytic; weights=none; horizon=overall; fe=unit_time; covariates=controls/time_covariates	tests/reference_parity/test_bjs_fe_covariates_parity.py	sp.did_imputation(controls=...|time_covariates=...)
T4	estimate	vce=analytic; weights=none; horizon=overall; fe=unit_time; covariates=unit_covariates	tests/reference_parity/test_bjs_fe_covariates_parity.py	sp.did_imputation(unit_covariates=...)
T2	se	vce=analytic; weights=none; horizon=overall; fe=none/time/unit_time; covariates=none	tests/reference_parity/test_option_fixture_bindings.py	sp.did_imputation(fe=...)
T2	se	vce=analytic; weights=none; horizon=overall; fe=unit_time; covariates=controls/time_covariates/unit_covariates	tests/reference_parity/test_option_fixture_bindings.py	sp.did_imputation(controls=...|time_covariates=...|unit_covariates=...)
T2	estimate, se	fe=unit_time; covariates=none; vce=analytic; weights=none; horizon=horizons	tests/r_parity/84_bjs_pretrends.py	sp.did_imputation(horizon=[...], pretrend_method='bjs')
T2	vcov	fe=unit_time; covariates=none; vce=analytic; weights=none; horizon=horizons	tests/reference_parity/test_event_study_vcov_Stata_parity.py	sp.did_imputation(horizon=[...])
T2	estimate, se	fe=custom; covariates=none; vce=analytic; weights=none; horizon=horizons/overall	tests/reference_parity/test_bjs_options_parity.py	sp.did_imputation(fe=[...])
T2	estimate, se	fe=unit_time; covariates=controls/mixed; vce=analytic; weights=none; horizon=overall	tests/reference_parity/test_bjs_options_parity.py	sp.did_imputation(controls=..., time_covariates=...)
T2	estimate, se	fe=unit_time; covariates=none; vce=analytic; weights=set; horizon=horizons/overall	tests/reference_parity/test_bjs_options_parity.py	sp.did_imputation(weights=...)
T2	estimate, se	fe=custom; covariates=controls; vce=analytic; weights=set; horizon=overall	tests/reference_parity/test_bjs_options_parity.py	sp.did_imputation(weights=..., fe=[...], controls=...)

vce is ignored for estimate: the covariance is computed after the imputation

did_multiplegt_dyn

Dimensions: se_method in {analytic, bootstrap}; aggregation in {simple, switchers}; control in {not_yet_treated, never_treated}; switchers in {both, in, out}; same_switchers in {false, true}; controls in {none, set}; trends_nonparam in {none, set}; normalized in {false, true}; weights in {none, set}; cluster in {group, coarser}; treatment in {binary, discrete, continuous}; panel in {balanced, unbalanced}; placebo in {0, >0}.

Kind	Outputs	Configuration run	Artifact	Entry point
T2	estimate, se, vcov, joint_test	se_method=analytic; aggregation=simple/switchers; control=not_yet_treated; switchers=both; same_switchers=false; controls=none; trends_nonparam=none; normalized=false; weights=none; cluster=group; treatment=binary; panel=balanced; placebo=>0	tests/reference_parity/test_dcdh_option_values_stata_parity.py	sp.did_multiplegt_dyn()
T2	estimate, se, vcov, joint_test	se_method=analytic; aggregation=simple/switchers; control=not_yet_treated; switchers=in; same_switchers=false; controls=none; trends_nonparam=none; normalized=false; weights=none; cluster=group; treatment=binary; panel=balanced; placebo=>0	tests/reference_parity/test_dcdh_option_values_stata_parity.py	sp.did_multiplegt_dyn()
T2	estimate, se, vcov, joint_test	se_method=analytic; aggregation=simple/switchers; control=not_yet_treated; switchers=out; same_switchers=false; controls=none; trends_nonparam=none; normalized=false; weights=none; cluster=group; treatment=binary; panel=balanced; placebo=>0	tests/reference_parity/test_dcdh_option_values_stata_parity.py	sp.did_multiplegt_dyn()
T2	estimate, se, vcov, joint_test	se_method=analytic; aggregation=simple/switchers; control=not_yet_treated; switchers=both; same_switchers=true; controls=none; trends_nonparam=none; normalized=false; weights=none; cluster=group; treatment=binary; panel=balanced; placebo=>0	tests/reference_parity/test_dcdh_option_values_stata_parity.py	sp.did_multiplegt_dyn()
T2	estimate, se, vcov, joint_test	se_method=analytic; aggregation=simple/switchers; control=never_treated; switchers=both; same_switchers=false; controls=none; trends_nonparam=none; normalized=false; weights=none; cluster=group; treatment=binary; panel=balanced; placebo=>0	tests/reference_parity/test_dcdh_option_values_stata_parity.py	sp.did_multiplegt_dyn()
T2	estimate, se, vcov, joint_test	se_method=analytic; aggregation=simple/switchers; control=not_yet_treated; switchers=both; same_switchers=false; controls=none; trends_nonparam=set; normalized=false; weights=none; cluster=group; treatment=binary; panel=balanced; placebo=>0	tests/reference_parity/test_dcdh_option_values_stata_parity.py	sp.did_multiplegt_dyn()
T2	estimate, se, vcov, joint_test	se_method=analytic; aggregation=simple/switchers; control=not_yet_treated; switchers=both; same_switchers=false; controls=none; trends_nonparam=none; normalized=true; weights=none; cluster=group; treatment=binary; panel=balanced; placebo=>0	tests/reference_parity/test_dcdh_option_values_stata_parity.py	sp.did_multiplegt_dyn()
T2	estimate, se, vcov, joint_test	se_method=analytic; aggregation=simple/switchers; control=not_yet_treated; switchers=both; same_switchers=false; controls=none; trends_nonparam=none; normalized=false; weights=set; cluster=group; treatment=binary; panel=balanced; placebo=>0	tests/reference_parity/test_dcdh_option_values_stata_parity.py	sp.did_multiplegt_dyn()
T2	estimate, se, vcov, joint_test	se_method=analytic; aggregation=simple/switchers; control=not_yet_treated; switchers=both; same_switchers=false; controls=none; trends_nonparam=none; normalized=false; weights=none; cluster=coarser; treatment=binary; panel=balanced; placebo=>0	tests/reference_parity/test_dcdh_option_values_stata_parity.py	sp.did_multiplegt_dyn()
T2	estimate, se, vcov, joint_test	se_method=analytic; aggregation=simple/switchers; control=not_yet_treated; switchers=both; same_switchers=false; controls=set; trends_nonparam=none; normalized=false; weights=none; cluster=group; treatment=binary; panel=balanced; placebo=>0	tests/reference_parity/test_dcdh_option_values_stata_parity.py	sp.did_multiplegt_dyn()
T2	estimate, se, vcov	se_method=analytic; aggregation=simple/switchers; control=not_yet_treated; switchers=both; same_switchers=false; controls=none; trends_nonparam=none; normalized=false; weights=none; cluster=group; treatment=binary; panel=balanced; placebo=>0	tests/reference_parity/test_dcdh_option_values_stata_parity.py	sp.did_multiplegt_dyn()
T2	estimate, se, vcov, joint_test	se_method=analytic; aggregation=simple/switchers; control=not_yet_treated; switchers=in; same_switchers=false; controls=none; trends_nonparam=set; normalized=true; weights=set; cluster=coarser; treatment=binary; panel=balanced; placebo=>0	tests/reference_parity/test_dcdh_option_values_stata_parity.py	sp.did_multiplegt_dyn()
T2	estimate, se, vcov, joint_test	se_method=analytic; aggregation=simple/switchers; control=not_yet_treated; switchers=both; same_switchers=false; controls=none; trends_nonparam=none; normalized=false; weights=none; cluster=group; treatment=continuous; panel=balanced; placebo=>0	tests/reference_parity/test_dcdh_option_values_stata_parity.py	sp.did_multiplegt_dyn()
T2	estimate, se, vcov, joint_test	se_method=analytic; aggregation=simple/switchers; control=not_yet_treated; switchers=both; same_switchers=false; controls=none; trends_nonparam=none; normalized=false; weights=none; cluster=group; treatment=binary; panel=unbalanced; placebo=>0	tests/reference_parity/test_dcdh_option_values_stata_parity.py	sp.did_multiplegt_dyn()
S	se	se_method=bootstrap; aggregation=switchers; control=not_yet_treated; switchers=both; same_switchers=false; controls=none; trends_nonparam=none; normalized=false; weights=none; cluster=group; treatment=binary; panel=balanced; placebo=>0	tests/reference_parity/test_dcdh_option_values_stata_parity.py	sp.did_multiplegt_dyn()
T2	estimate	se_method=analytic; aggregation=switchers; control=not_yet_treated; switchers=both; same_switchers=false; controls=none; trends_nonparam=none; normalized=false; weights=none; cluster=group; treatment=binary; panel=balanced; placebo=>0	tests/reference_parity/test_multiplegt_dyn_options_parity.py	sp.did_multiplegt_dyn()
T2	estimate	se_method=analytic; aggregation=switchers; control=not_yet_treated; switchers=in/out; same_switchers=false; controls=none; trends_nonparam=none; normalized=false; weights=none; cluster=group; treatment=binary; panel=balanced; placebo=>0	tests/reference_parity/test_multiplegt_dyn_options_parity.py	sp.did_multiplegt_dyn()
T2	estimate	se_method=analytic; aggregation=switchers; control=not_yet_treated; switchers=both; same_switchers=true; controls=none; trends_nonparam=none; normalized=false; weights=none; cluster=group; treatment=binary; panel=balanced; placebo=>0	tests/reference_parity/test_multiplegt_dyn_options_parity.py	sp.did_multiplegt_dyn()
T2	estimate	se_method=bootstrap; aggregation=simple/switchers; control=not_yet_treated; switchers=both; same_switchers=false; controls=none; trends_nonparam=none; normalized=false; weights=none; cluster=group; treatment=binary; panel=balanced; placebo=>0	tests/r_parity/78_multiplegt_dyn.py	sp.did_multiplegt_dyn()
T2	estimate, se	se_method=analytic; aggregation=switchers; control=not_yet_treated; switchers=both; same_switchers=false; controls=none; trends_nonparam=none; normalized=false; weights=none; cluster=group; treatment=binary; panel=balanced; placebo=>0	tests/reference_parity/test_multiplegt_dyn_parity.py	sp.did_multiplegt_dyn()
T2	estimate	se_method=analytic; aggregation=simple; control=not_yet_treated; switchers=both; same_switchers=false; controls=none; trends_nonparam=none; normalized=false; weights=none; cluster=group; treatment=binary; panel=balanced; placebo=>0	tests/reference_parity/test_multiplegt_dyn_parity.py	sp.did_multiplegt_dyn()
T2	estimate	se_method=analytic; aggregation=simple; control=not_yet_treated; switchers=both; same_switchers=false; controls=none; trends_nonparam=none; normalized=false; weights=none; cluster=group; treatment=binary; panel=balanced; placebo=>0	tests/reference_parity/test_dcdh_options_parity.py	sp.did_multiplegt_dyn()
T2	estimate	se_method=analytic; aggregation=simple; control=not_yet_treated; switchers=both; same_switchers=false; controls=set; trends_nonparam=none; normalized=false; weights=none; cluster=group; treatment=binary; panel=balanced; placebo=>0	tests/reference_parity/test_dcdh_options_parity.py	sp.did_multiplegt_dyn()
T2	estimate	se_method=analytic; aggregation=simple; control=not_yet_treated; switchers=both; same_switchers=false; controls=none; trends_nonparam=set; normalized=false; weights=none; cluster=group; treatment=binary; panel=balanced; placebo=>0	tests/reference_parity/test_dcdh_options_parity.py	sp.did_multiplegt_dyn()
T2	estimate	se_method=analytic; aggregation=simple; control=not_yet_treated; switchers=both; same_switchers=false; controls=none; trends_nonparam=none; normalized=true; weights=none; cluster=group; treatment=binary; panel=balanced; placebo=>0	tests/reference_parity/test_dcdh_options_parity.py	sp.did_multiplegt_dyn()
T2	estimate	se_method=analytic; aggregation=simple; control=not_yet_treated; switchers=both; same_switchers=false; controls=none; trends_nonparam=none; normalized=false; weights=none; cluster=group; treatment=continuous; panel=balanced; placebo=>0	tests/reference_parity/test_dcdh_options_parity.py	sp.did_multiplegt_dyn()
T2	estimate, se, joint_test	se_method=analytic; aggregation=switchers; control=not_yet_treated; switchers=both; same_switchers=false; controls=none; trends_nonparam=none; normalized=false; weights=none; cluster=group; treatment=discrete; panel=balanced; placebo=>0	tests/reference_parity/test_dcdh_textbook_stata_parity.py	sp.did_multiplegt_dyn()
T2	estimate, se, joint_test	se_method=analytic; aggregation=switchers; control=not_yet_treated; switchers=both; same_switchers=false; controls=none; trends_nonparam=none; normalized=true; weights=none; cluster=group; treatment=discrete; panel=balanced; placebo=>0	tests/reference_parity/test_dcdh_textbook_stata_parity.py	sp.did_multiplegt_dyn()
T2	estimate, se, joint_test	se_method=analytic; aggregation=switchers; control=not_yet_treated; switchers=both; same_switchers=true; controls=none; trends_nonparam=none; normalized=false; weights=none; cluster=group; treatment=discrete; panel=balanced; placebo=0	tests/reference_parity/test_dcdh_textbook_stata_parity.py	sp.did_multiplegt_dyn()
T2	estimate, se, joint_test	se_method=analytic; aggregation=switchers; control=not_yet_treated; switchers=both; same_switchers=false; controls=none; trends_nonparam=none; normalized=false; weights=set; cluster=coarser; treatment=discrete; panel=balanced; placebo=>0	tests/reference_parity/test_dcdh_textbook_stata_parity.py	sp.did_multiplegt_dyn()
T2	estimate, se, joint_test	se_method=analytic; aggregation=switchers; control=not_yet_treated; switchers=both; same_switchers=false; controls=set; trends_nonparam=none; normalized=false; weights=none; cluster=group; treatment=discrete; panel=balanced; placebo=>0	tests/reference_parity/test_dcdh_textbook_stata_parity.py	sp.did_multiplegt_dyn()
T2	estimate, se, joint_test	se_method=analytic; aggregation=switchers; control=not_yet_treated; switchers=both; same_switchers=false; controls=set; trends_nonparam=none; normalized=false; weights=set; cluster=coarser; treatment=discrete; panel=balanced; placebo=>0	tests/reference_parity/test_dcdh_textbook_stata_parity.py	sp.did_multiplegt_dyn()
T2	estimate, se, joint_test	se_method=analytic; aggregation=switchers; control=not_yet_treated; switchers=both; same_switchers=false; controls=set; trends_nonparam=none; normalized=true; weights=none; cluster=group; treatment=discrete; panel=balanced; placebo=0	tests/reference_parity/test_dcdh_textbook_stata_parity.py	sp.did_multiplegt_dyn()
T2	estimate, se, joint_test	se_method=analytic; aggregation=switchers; control=not_yet_treated; switchers=both; same_switchers=false; controls=set; trends_nonparam=set; normalized=false; weights=none; cluster=group; treatment=discrete; panel=balanced; placebo=0	tests/reference_parity/test_dcdh_textbook_stata_parity.py	sp.did_multiplegt_dyn()
T2	estimate, se, joint_test	se_method=analytic; aggregation=switchers; control=not_yet_treated; switchers=in; same_switchers=false; controls=set; trends_nonparam=none; normalized=false; weights=none; cluster=group; treatment=discrete; panel=balanced; placebo=0	tests/reference_parity/test_dcdh_textbook_stata_parity.py	sp.did_multiplegt_dyn()
T2	estimate, se, joint_test	se_method=analytic; aggregation=switchers; control=not_yet_treated; switchers=both; same_switchers=false; controls=set; trends_nonparam=none; normalized=false; weights=none; cluster=group; treatment=binary; panel=unbalanced; placebo=>0	tests/reference_parity/test_dcdh_textbook_stata_parity.py	sp.did_multiplegt_dyn()
T2	estimate, se	se_method=analytic; aggregation=switchers; control=not_yet_treated; switchers=both; same_switchers=false; controls=none; trends_nonparam=none; normalized=false; weights=none/set; cluster=coarser/group; treatment=binary; panel=balanced; placebo=>0	tests/test_did_multiplegt_dyn_castle_reference.py	sp.did_multiplegt_dyn()

se_method is ignored for estimate: the effects are computed before, and independently of, the variance; the bootstrap resamples afterwards

cluster is ignored for estimate: the cluster enters the variance only

dml

Dimensions: model in {plr, irm, pliv, iivm}; score in {partialling out, iv-type, ate, atte, late}; learners in {linear, default, other}; n_folds in {2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20}; n_rep in {1, >1}; ipw in {n/a, trim0.01, trim1e-12, other}.

Kind	Outputs	Configuration run	Artifact	Entry point
T2	estimate, se	model=plr; score=partialling out; learners=linear; n_folds=5; n_rep=1; ipw=n/a	tests/r_parity/08_dml.py	sp.dml(model='plr')
T2	estimate, se	model=irm; score=ate; learners=linear; n_folds=5; n_rep=1; ipw=trim1e-12	tests/r_parity/71_dml_family.py	sp.dml(model='irm')
T2	estimate, se	model=pliv; score=partialling out; learners=linear; n_folds=5; n_rep=1; ipw=n/a	tests/r_parity/71_dml_family.py	sp.dml(model='pliv')
T2	estimate, se	model=iivm; score=late; learners=linear; n_folds=5; n_rep=1; ipw=trim1e-12	tests/r_parity/71_dml_family.py	sp.dml(model='iivm')
T2	estimate, se	model=plr; score=partialling out; learners=other; n_folds=4; n_rep=1; ipw=n/a	tests/external_parity/test_dml_classifier_nuisance_and_irm_sensitivity.py	sp.dml(model='plr')
T2	estimate, se	model=irm; score=ate/atte; learners=linear; n_folds=4; n_rep=1; ipw=trim0.01	tests/external_parity/test_dml_classifier_nuisance_and_irm_sensitivity.py	sp.dml(model='irm')
B	coverage	model=plr; score=partialling out; learners=default; n_folds=5; n_rep=1; ipw=n/a	tests/coverage_monte_carlo/results_b1000/coverage_b1000.json	sp.dml(model='plr', n_folds=5)
B	coverage	model=irm; score=ate; learners=default; n_folds=5; n_rep=1; ipw=trim0.01	tests/coverage_monte_carlo/results_b1000/coverage_b1000.json	sp.causal_question(design='dml')
T2	estimate, se	model=plr; score=iv-type; learners=linear; n_folds=5; n_rep=1; ipw=n/a	tests/test_dml_ivtype_reference.py	sp.dml(model='plr', score='IV-type')
T2	estimate, se	model=plr; score=iv-type; learners=other; n_folds=5; n_rep=1; ipw=n/a	tests/test_dml_ivtype_reference.py	sp.dml(model='plr', score='IV-type')
T2	estimate, se	model=irm; score=ate/atte; learners=linear; n_folds=5; n_rep=1; ipw=trim0.01	tests/test_dml_ivtype_reference.py	sp.dml(model='irm')
T2	estimate, se	model=plr; score=iv-type/partialling out; learners=linear; n_folds=10/2/3; n_rep=1; ipw=n/a	tests/reference_parity/test_dml_options_doubleml_parity.py	sp.dml(df, model='plr')
T2	estimate, se	model=irm; score=ate; learners=linear; n_folds=10/2/3; n_rep=1; ipw=trim0.01	tests/reference_parity/test_dml_options_doubleml_parity.py	sp.dml(df, model='irm')
T2	estimate, se	model=irm; score=atte; learners=linear; n_folds=3; n_rep=1; ipw=trim0.01	tests/reference_parity/test_dml_options_doubleml_parity.py	sp.dml(df, model='irm')
T2	estimate, se	model=irm; score=ate/atte; learners=linear; n_folds=3; n_rep=1; ipw=other	tests/reference_parity/test_dml_options_doubleml_parity.py	sp.dml(df, model='irm')
T2	estimate, se	model=pliv; score=partialling out; learners=linear; n_folds=10/2/3; n_rep=1; ipw=n/a	tests/reference_parity/test_dml_options_doubleml_parity.py	sp.dml(df, model='pliv')
T2	estimate, se	model=iivm; score=late; learners=linear; n_folds=10/2/3; n_rep=1; ipw=trim0.01	tests/reference_parity/test_dml_options_doubleml_parity.py	sp.dml(df, model='iivm')
T2	estimate, se	model=iivm; score=late; learners=linear; n_folds=3; n_rep=1; ipw=other	tests/reference_parity/test_dml_options_doubleml_parity.py	sp.dml(df, model='iivm')
T2	estimate	model=plr; score=iv-type/partialling out; learners=linear; n_folds=5; n_rep=>1; ipw=n/a	tests/reference_parity/test_dml_options_doubleml_parity.py	sp.dml(df, model='plr', n_rep=3)
T2	estimate	model=irm; score=ate; learners=linear; n_folds=5; n_rep=>1; ipw=trim0.01	tests/reference_parity/test_dml_options_doubleml_parity.py	sp.dml(df, model='irm', n_rep=3)
T4	se	model=plr; score=iv-type/partialling out; learners=linear; n_folds=5; n_rep=>1; ipw=n/a	tests/reference_parity/test_dml_options_doubleml_parity.py	sp.dml(df, model='plr', n_rep=3)
T4	se	model=irm; score=ate; learners=linear; n_folds=5; n_rep=>1; ipw=trim0.01	tests/reference_parity/test_dml_options_doubleml_parity.py	sp.dml(df, model='irm', n_rep=3)

ebalance

Dimensions: moments in {1, 2, 3, mixed}; dof_adjust in {true, false}; vce in {mestimation, naive}.

Kind	Outputs	Configuration run	Artifact	Entry point
T2	estimate, se	moments=1/2/3; dof_adjust=false; vce=mestimation	tests/reference_parity/test_ebalance_weightit_parity.py	sp.ebalance(...)
T4	se	moments=1/2/3; dof_adjust=false; vce=naive	tests/reference_parity/test_ebalance_weightit_parity.py	sp.ebalance(...)
T2	estimate	moments=1/mixed; dof_adjust=true; vce=mestimation	tests/reference_parity/test_the_effect_stata_parity.py	sp.ebalance(...)

vce is ignored for estimate: the point estimate is computed before, and independently of, the variance estimator

etwfe

Dimensions: family in {linear, glm}; panel in {true, false}; cgroup in {notyet, nevertreated}; controls in {none, set}; xvar in {none, set}; weights in {none, set}; agg_weights in {estimation, unit}.

Kind	Outputs	Configuration run	Artifact	Entry point
T2	estimate	family=linear; controls=none; xvar=none; weights=none; agg_weights=estimation; panel=false; cgroup=notyet	tests/r_parity/17_etwfe.py	sp.etwfe(panel=False)
T2	estimate	family=linear; controls=none; xvar=none; weights=none; agg_weights=estimation; panel=true; cgroup=nevertreated	tests/r_parity/17_etwfe.py	sp.etwfe(cgroup='nevertreated')
T4	se	family=linear; controls=none; xvar=none; weights=none; agg_weights=estimation; panel=false; cgroup=notyet	tests/r_parity/17_etwfe.py	sp.etwfe(panel=False)
T4	se	family=linear; controls=none; xvar=none; weights=none; agg_weights=estimation; panel=true; cgroup=nevertreated	tests/r_parity/17_etwfe.py	sp.etwfe(cgroup='nevertreated')
T2	vcov	family=linear; controls=none; xvar=none; weights=none; agg_weights=estimation; panel=false; cgroup=notyet	tests/reference_parity/test_event_study_vcov_R_parity.py	sp.etwfe(panel=False)
T2	vcov	family=linear; controls=none; xvar=none; weights=none; agg_weights=estimation; panel=true; cgroup=nevertreated	tests/reference_parity/test_event_study_vcov_R_parity.py	sp.etwfe(cgroup='nevertreated')
T2	estimate, vcov	family=linear; controls=none; xvar=none; weights=none; agg_weights=estimation; panel=true; cgroup=notyet	tests/reference_parity/test_validation_entry_points.py	sp.etwfe()
T4	se	family=linear; controls=none; xvar=none; weights=none; agg_weights=estimation; panel=true; cgroup=notyet	tests/reference_parity/test_validation_entry_points.py	sp.etwfe()
T2	estimate	family=linear; panel=true; cgroup=nevertreated/notyet; controls=none; xvar=none; weights=set; agg_weights=estimation	tests/test_etwfe_weights_reference.py	sp.etwfe(panel, ..., weights=...)
T4	se	family=linear; panel=true; cgroup=nevertreated/notyet; controls=none; xvar=none; weights=set; agg_weights=estimation	tests/test_etwfe_weights_reference.py	sp.etwfe(panel, ..., weights=...)
T2	estimate	family=linear; panel=true; cgroup=nevertreated/notyet; controls=none; xvar=none; weights=set; agg_weights=unit	tests/test_etwfe_weights_reference.py	sp.etwfe(panel, ..., weights=...)
T4	se	family=linear; panel=true; cgroup=nevertreated/notyet; controls=none; xvar=none; weights=set; agg_weights=unit	tests/test_etwfe_weights_reference.py	sp.etwfe(panel, ..., weights=...)
T2	estimate, se	family=linear; controls=none; xvar=none; weights=none; agg_weights=estimation; panel=true; cgroup=nevertreated/notyet	tests/test_etwfe_emfx_reference.py	sp.etwfe(castle, ...)
T2	estimate, se	family=linear; controls=none; xvar=none; weights=none; agg_weights=estimation/unit; panel=true; cgroup=nevertreated/notyet	tests/reference_parity/test_etwfe_options_parity.py	sp.etwfe(panel, **args)
T2	estimate, se	family=linear; controls=none; xvar=none; weights=none; agg_weights=estimation; panel=false; cgroup=notyet	tests/reference_parity/test_etwfe_options_parity.py	sp.etwfe(panel, **args)
T2	estimate, se	family=linear; controls=none; xvar=none; weights=set; agg_weights=estimation/unit; panel=true; cgroup=nevertreated/notyet	tests/reference_parity/test_etwfe_options_parity.py	sp.etwfe(panel, **args)
T2	estimate, se	family=linear; controls=none; xvar=set; weights=none; agg_weights=estimation; panel=true; cgroup=nevertreated/notyet	tests/reference_parity/test_etwfe_options_parity.py	sp.etwfe(panel, **args)
T2	estimate, se	family=linear; controls=set; xvar=none; weights=none; agg_weights=estimation; panel=true; cgroup=nevertreated/notyet	tests/reference_parity/test_etwfe_options_parity.py	sp.etwfe(panel, **args)
T2	estimate, se	family=glm; controls=none; xvar=none; weights=none; agg_weights=estimation; panel=true; cgroup=nevertreated/notyet	tests/reference_parity/test_etwfe_options_parity.py	sp.etwfe(panel, **args)
T2	estimate, se	family=glm; controls=none; xvar=set; weights=none; agg_weights=estimation; panel=true; cgroup=notyet	tests/reference_parity/test_etwfe_options_parity.py	sp.etwfe(panel, **args)

etwfe_glm

Dimensions: family in {poisson, logit, gaussian}; cgroup in {notyet, nevertreated}; fe in {cohort, unit}; hettype in {timecohort, time, cohort, event, twfe}; scale in {response, link}; response_se in {profile, margins, unconditional}; controls in {none, set}; xvar in {none, set}; weights in {none, set}.

Kind	Outputs	Configuration run	Artifact	Entry point
T2	estimate	family=logit/poisson; cgroup=notyet; fe=cohort; hettype=timecohort; scale=response; response_se=profile; controls=none; xvar=none; weights=none	tests/reference_parity/test_etwfe_glm_parity.py	sp.etwfe(family='poisson'|'logit')
T4	se	family=logit/poisson; cgroup=notyet; fe=cohort; hettype=timecohort; scale=response; response_se=profile; controls=none; xvar=none; weights=none	tests/reference_parity/test_etwfe_glm_parity.py	sp.etwfe(family='poisson'|'logit')
T2	estimate, se	family=logit/poisson; cgroup=nevertreated/notyet; fe=cohort; hettype=timecohort; scale=link/response; response_se=profile; controls=none; xvar=none; weights=none	tests/reference_parity/test_etwfe_options_parity.py	sp.etwfe(panel, **args)
T2	estimate, se	family=poisson; cgroup=notyet; fe=cohort; hettype=timecohort; scale=link/response; response_se=profile; controls=none; xvar=set; weights=none	tests/reference_parity/test_etwfe_options_parity.py	sp.etwfe(panel, **args)
T2	estimate	family=logit/poisson; cgroup=nevertreated/notyet; fe=cohort; hettype=timecohort; scale=response; response_se=unconditional; controls=none; xvar=none; weights=none	tests/reference_parity/test_etwfe_options_parity.py	sp.etwfe(panel, **args)
T4	se	family=logit/poisson; cgroup=nevertreated/notyet; fe=cohort; hettype=timecohort; scale=response; response_se=unconditional; controls=none; xvar=none; weights=none	tests/reference_parity/test_etwfe_options_parity.py	sp.etwfe(panel, **args)
T2	estimate, se	family=gaussian; cgroup=notyet; fe=unit; hettype=cohort/event/time/timecohort/twfe; scale=link; response_se=profile; controls=none; xvar=none; weights=none	tests/reference_parity/test_etwfe_linear_jwdid_parity.py	sp.etwfe(df, y, group, time, first_treat, fe='unit' | hettype=...)
T2	estimate, se	family=gaussian; cgroup=nevertreated; fe=unit; hettype=cohort/event/timecohort; scale=link; response_se=profile; controls=none; xvar=none; weights=none	tests/reference_parity/test_etwfe_linear_jwdid_parity.py	sp.etwfe(df, y, group, time, first_treat, fe='unit' | hettype=...)
T2	estimate, se	family=gaussian; cgroup=notyet; fe=unit; hettype=event/timecohort; scale=link; response_se=profile; controls=none; xvar=set; weights=none	tests/reference_parity/test_etwfe_linear_jwdid_parity.py	sp.etwfe(df, y, group, time, first_treat, fe='unit' | hettype=...)
T2	estimate, se	family=gaussian; cgroup=nevertreated; fe=unit; hettype=timecohort; scale=link; response_se=profile; controls=none; xvar=set; weights=none	tests/reference_parity/test_etwfe_linear_jwdid_parity.py	sp.etwfe(df, y, group, time, first_treat, fe='unit' | hettype=...)
T2	estimate	family=poisson; cgroup=notyet; fe=unit; hettype=cohort/event/time/timecohort/twfe; scale=response; response_se=profile; controls=none; xvar=none; weights=none	tests/reference_parity/test_etwfe_poisson_jwdid_parity.py	sp.etwfe(panel, family='poisson', fe='unit', ...)
T2	estimate	family=poisson; cgroup=nevertreated; fe=unit; hettype=cohort/event/timecohort; scale=response; response_se=profile; controls=none; xvar=none; weights=none	tests/reference_parity/test_etwfe_poisson_jwdid_parity.py	sp.etwfe(panel, family='poisson', fe='unit', ...)
T2	estimate	family=poisson; cgroup=notyet; fe=unit; hettype=event/timecohort; scale=response; response_se=profile; controls=none; xvar=set; weights=none	tests/reference_parity/test_etwfe_poisson_jwdid_parity.py	sp.etwfe(panel, family='poisson', fe='unit', ...)
T2	estimate	family=poisson; cgroup=nevertreated; fe=unit; hettype=timecohort; scale=response; response_se=profile; controls=none; xvar=set; weights=none	tests/reference_parity/test_etwfe_poisson_jwdid_parity.py	sp.etwfe(panel, family='poisson', fe='unit', ...)
T2	estimate, se	family=poisson; cgroup=notyet; fe=unit; hettype=timecohort; scale=response; response_se=margins; controls=none; xvar=none/set; weights=none	tests/reference_parity/test_etwfe_poisson_jwdid_parity.py	sp.etwfe(panel, family='poisson', fe='unit', response_se='margins')
T2	estimate, se	family=poisson; cgroup=notyet; fe=unit; hettype=cohort/event/time/timecohort/twfe; scale=link; response_se=profile; controls=none; xvar=none; weights=none	tests/reference_parity/test_etwfe_poisson_jwdid_parity.py	sp.etwfe(df, family='poisson', fe='unit', scale='link', ...)
T2	estimate, se	family=poisson; cgroup=nevertreated; fe=unit; hettype=cohort/event/timecohort; scale=link; response_se=profile; controls=none; xvar=none; weights=none	tests/reference_parity/test_etwfe_poisson_jwdid_parity.py	sp.etwfe(df, family='poisson', fe='unit', scale='link', ...)
T2	estimate, se	family=poisson; cgroup=notyet; fe=unit; hettype=event/timecohort; scale=link; response_se=profile; controls=none; xvar=set; weights=none	tests/reference_parity/test_etwfe_poisson_jwdid_parity.py	sp.etwfe(df, family='poisson', fe='unit', scale='link', ...)
T2	estimate, se	family=poisson; cgroup=nevertreated; fe=unit; hettype=timecohort; scale=link; response_se=profile; controls=none; xvar=set; weights=none	tests/reference_parity/test_etwfe_poisson_jwdid_parity.py	sp.etwfe(df, family='poisson', fe='unit', scale='link', ...)
T2	estimate, se	family=gaussian; cgroup=nevertreated/notyet; fe=unit; hettype=timecohort; scale=link; response_se=profile; controls=set; xvar=none; weights=none	tests/reference_parity/test_etwfe_linear_jwdid_parity.py	sp.etwfe(df, y, group, time, first_treat, fe='unit' | hettype=...)

event_study

Dimensions: adoption in {single_date, staggered}; window in {-4_4, other}; ref_period in {-1, other}; covariates in {none, set}; weights in {none, set}; cluster in {unit, other}; bins in {none, set}; intensity in {none, set}; absorb in {none, set}; crit_dist in {normal, t_resid, t_cluster}.

Kind	Outputs	Configuration run	Artifact	Entry point
T2	estimate, se	adoption=single_date; window=-4_4; ref_period=-1; covariates=none; weights=none; cluster=unit; bins=none; intensity=none; absorb=none; crit_dist=normal	tests/r_parity/85_twfe_event_study.py	sp.event_study(window=(-4, 4), cluster='unit')
T2	vcov	adoption=single_date; window=-4_4; ref_period=-1; covariates=none; weights=none; cluster=unit; bins=none; intensity=none; absorb=none; crit_dist=normal	tests/reference_parity/test_event_study_vcov_R_parity.py	sp.event_study(window=(-4, 4), cluster='unit')
T2	estimate, se	adoption=single_date/staggered; window=-4_4/other; ref_period=-1/other; covariates=none/set; weights=none/set; cluster=unit; bins=none; intensity=none; absorb=none; crit_dist=normal	tests/reference_parity/test_event_study_grid_R_parity.py	sp.event_study(covariates=..., weights=...)
T2	estimate, se	adoption=staggered; window=-4_4/other; ref_period=-1/other; covariates=none; weights=none; cluster=unit; bins=none; intensity=none; absorb=none; crit_dist=normal	tests/reference_parity/test_event_study_options_R_parity.py	sp.event_study(...)
T2	estimate, se	adoption=staggered; window=-4_4; ref_period=-1; covariates=none; weights=none/set; cluster=unit; bins=none; intensity=none; absorb=none; crit_dist=normal	tests/reference_parity/test_event_study_options_R_parity.py	sp.event_study(...)
T2	estimate, se	adoption=staggered; window=-4_4; ref_period=-1; covariates=set; weights=set; cluster=unit; bins=none; intensity=none; absorb=none; crit_dist=normal	tests/reference_parity/test_event_study_options_R_parity.py	sp.event_study(...)
T2	estimate, se	adoption=staggered; window=-4_4; ref_period=-1; covariates=none/set; weights=none; cluster=other; bins=none; intensity=none; absorb=none; crit_dist=normal	tests/reference_parity/test_event_study_options_R_parity.py	sp.event_study(...)
T2	estimate, se	adoption=single_date; window=-4_4; ref_period=-1; covariates=none; weights=set; cluster=other; bins=none; intensity=none; absorb=none; crit_dist=normal	tests/reference_parity/test_event_study_options_R_parity.py	sp.event_study(...)
T2	estimate, se	adoption=staggered; window=other; ref_period=other; covariates=none; weights=none; cluster=unit; bins=set; intensity=none; absorb=none/set; crit_dist=normal	tests/reference_parity/test_event_study_options_R_parity.py	sp.event_study(...)
T2	estimate, se	adoption=single_date; window=other; ref_period=other; covariates=set; weights=none; cluster=unit; bins=set; intensity=none; absorb=none; crit_dist=normal	tests/reference_parity/test_event_study_options_R_parity.py	sp.event_study(...)
T2	estimate, se	adoption=single_date; window=-4_4; ref_period=-1; covariates=none; weights=none/set; cluster=unit; bins=none; intensity=set; absorb=none; crit_dist=normal	tests/reference_parity/test_event_study_options_R_parity.py	sp.event_study(...)
T2	estimate, se	adoption=single_date; window=other; ref_period=-1; covariates=none; weights=none; cluster=other; bins=none; intensity=set; absorb=none; crit_dist=normal	tests/reference_parity/test_event_study_options_R_parity.py	sp.event_study(...)
T2	estimate, se	adoption=staggered; window=-4_4; ref_period=-1; covariates=none; weights=none; cluster=unit; bins=none; intensity=set; absorb=none; crit_dist=normal	tests/reference_parity/test_event_study_options_R_parity.py	sp.event_study(...)
T2	estimate, se	adoption=staggered; window=-4_4; ref_period=-1; covariates=none/set; weights=none; cluster=unit; bins=none; intensity=none; absorb=set; crit_dist=normal	tests/reference_parity/test_event_study_options_R_parity.py	sp.event_study(...)
T2	estimate, se	adoption=single_date; window=-4_4; ref_period=-1; covariates=none; weights=set; cluster=unit; bins=none; intensity=none; absorb=set; crit_dist=normal	tests/reference_parity/test_event_study_options_R_parity.py	sp.event_study(...)
T2	estimate, se, interval	adoption=staggered; window=other; ref_period=-1; covariates=none; weights=set; cluster=other/unit; bins=none; intensity=none; absorb=none; crit_dist=t_cluster	tests/reference_parity/test_es_crit_parity.py	sp.event_study(
T2	estimate, se, interval	adoption=staggered; window=other; ref_period=-1; covariates=none; weights=none; cluster=unit; bins=none; intensity=none; absorb=none; crit_dist=t_cluster	tests/reference_parity/test_es_crit_parity.py	sp.event_study(

crit_dist is ignored for estimate, se, vcov, joint_test: the critical value turns an SE into a p-value and an interval; coefficients, SEs and the pre-trend test are computed before it is chosen (asserted equal across parity= in tests/reference_parity/test_es_crit_parity.py)

fast.feols

Dimensions: vcov in {iid, hc1, cr1}; ssc in {fixest, statspai}; weights in {none, set}.

Kind	Outputs	Configuration run	Artifact	Entry point
T2	estimate, se	vcov=iid; ssc=fixest; weights=none	tests/r_parity/03_hdfe.py	sp.fast.feols(vcov='iid')
T2	estimate, se	vcov=cr1; ssc=fixest; weights=none	tests/r_parity/15_hdfe_cluster.py	sp.fast.feols(vcov='cr1')
B	coverage	vcov=cr1; ssc=fixest; weights=none	tests/coverage_monte_carlo/results_b1000/coverage_b1000.json	sp.fast.feols(vcov='cr1')
T2	estimate, se	vcov=cr1/hc1/iid; ssc=fixest; weights=none/set	tests/reference_parity/test_fast_feols_weights_fixest_parity.py	sp.fast.feols(formula, data, weights=..., vcov=...)

vcov is ignored for estimate: the least-squares / k-class point estimate is computed before, and independently of, the covariance estimator

ssc is ignored for estimate: the small-sample factor scales the covariance only

feols

Dimensions: fe in {none, one, multi}; iv in {false, true}; vcov in {iid, hetero, cr1, cr3, cluster_multiway, other}; weights in {none, set}; ssc in {default, custom}.

Kind	Outputs	Configuration run	Artifact	Entry point
T2	estimate, se	fe=multi/none/one; iv=false; vcov=cluster_multiway/cr1/hetero/iid; weights=none/set; ssc=default	tests/reference_parity/test_feols_grid_fixest_parity.py	sp.feols(...)
T2	estimate, se	fe=multi; iv=false; vcov=cluster_multiway/iid; weights=none; ssc=default	tests/external_parity/test_gow_ding_accounting.py	sp.feols(...)
T2	estimate, se	fe=one; iv=false; vcov=cr1; weights=none; ssc=default	tests/reference_parity/test_sasp_within_parity.py	sp.feols(...)
T2	estimate, se	fe=multi; iv=false; vcov=cr1; weights=none/set; ssc=default	tests/reference_parity/test_castle_stata_parity.py	sp.feols(...)
T2	estimate, se	fe=multi; iv=false; vcov=cr1; weights=none; ssc=default	tests/reference_parity/test_feols_multiref_R_parity.py	sp.feols(...)
T2	estimate, se	fe=multi; iv=false; vcov=cluster_multiway; weights=none; ssc=default	tests/reference_parity/test_accounting_research_parity.py	sp.feols(...)
T2	se	fe=none; iv=false/true; vcov=cr1/hetero; weights=none; ssc=custom	tests/reference_parity/test_ssc_presets_Stata_parity.py	sp.feols(...)
T2	se	fe=one; iv=false; vcov=cr1; weights=none; ssc=custom	tests/reference_parity/test_ssc_presets_Stata_parity.py	sp.feols(...)

vcov is ignored for estimate: the point estimate is computed before, and independently of, the covariance estimator

ssc is ignored for estimate: the point estimate is computed before, and independently of, the covariance estimator

g_computation

Dimensions: estimand in {ate, att, atc}; by_arm in {true, false}; ps in {none, set}; se_method in {bootstrap, analytic}; ml_q in {ols, other}.

Kind	Outputs	Configuration run	Artifact	Entry point
T1	estimate	estimand=atc/ate/att; by_arm=false; ps=none; se_method=bootstrap; ml_q=ols	tests/test_inference_scan3_weighting.py	sp.g_computation(...)
T1	se	estimand=ate; by_arm=false; ps=none; se_method=bootstrap; ml_q=ols	tests/test_inference_scan3_weighting.py	sp.g_computation(...)
T1	estimate, se	estimand=ate/att; by_arm=true; ps=none/set; se_method=analytic; ml_q=ols	tests/test_inference_scan3_weighting.py	sp.g_computation(...)
T1	estimate	estimand=ate; by_arm=false; ps=none; se_method=bootstrap; ml_q=other	tests/reference_parity/test_gformula_parity.py	sp.g_computation(...)

se_method is ignored for estimate: the point estimate is computed before, and independently of, the variance estimator

gardner_did

Dimensions: vce in {analytic, stage2, bootstrap, none}; weights in {none, set}; covariates in {none, set}; event_study in {off, on}; form in {first_treat, general}.

Kind	Outputs	Configuration run	Artifact	Entry point
T2	estimate, se	vce=analytic; weights=none/set; covariates=none/set; event_study=off; form=general	tests/reference_parity/test_stata_did_commands_parity.py	sp.gardner_did(treat=, fe=)
T2	estimate, se	vce=analytic; weights=none; covariates=none; form=first_treat; event_study=off	tests/r_parity/73_did2s.py	sp.gardner_did()
T2	estimate, se, vcov	vce=analytic; weights=none; covariates=none; form=first_treat; event_study=on	tests/reference_parity/test_event_study_vcov_R_parity.py	sp.gardner_did(event_study=True)
T2	estimate, se	vce=stage2; weights=none; covariates=none; form=first_treat; event_study=off	tests/reference_parity/test_gardner_stage2_R_parity.py	sp.gardner_did(vce='stage2')

vce is ignored for estimate: the covariance is computed after the two stages

glm

Dimensions: family_link in {gaussian_identity, gaussian_log, binomial_logit, binomial_probit, binomial_cloglog, binomial_cauchit, poisson_log, gamma_log, inverse_gaussian_log, other}; vce in {classical, robust, hc0, hc1, hc2, hc3, hac, cluster}; weights in {none, set}; information in {observed, expected}.

Kind	Outputs	Configuration run	Artifact	Entry point
T2	estimate, se	family_link=binomial_cloglog/binomial_logit/binomial_probit/gaussian_identity/gaussian_log/poisson_log; vce=classical/cluster/robust; weights=none; information=observed	tests/reference_parity/test_vce_grammar_stata_parity.py	sp.glm(...)
T2	estimate, se	family_link=binomial_logit/poisson_log; vce=classical; weights=none; information=observed	tests/reference_parity/test_glm_parity.py	sp.glm(...)
T2	estimate, se	family_link=gamma_log/inverse_gaussian_log; vce=classical; weights=none; information=expected/observed	tests/reference_parity/test_glm_dispersion_stata_r_parity.py	sp.glm(...)
T2	estimate, se	family_link=binomial_cauchit; vce=classical; weights=none; information=expected	tests/reference_parity/test_linear_model_extensions_parity.py	sp.glm(...)
T2	estimate, se	family_link=binomial_logit/poisson_log; vce=hc0/hc1/hc2/hc3; weights=none; information=observed	tests/reference_parity/test_ml_sandwich_r_parity.py	sp.glm(...)
T2	se	family_link=binomial_probit; vce=hc0; weights=none; information=expected	tests/reference_parity/test_ml_sandwich_r_parity.py	sp.glm(...)

vce is ignored for estimate: the maximum-likelihood point estimate is computed before, and independently of, the covariance estimator

information is ignored for estimate: the maximum-likelihood point estimate is computed before, and independently of, the covariance estimator

heckman

Dimensions: method in {twostep, ml}; vce in {classical, robust, cluster}; weights in {none, set}.

Kind	Outputs	Configuration run	Artifact	Entry point
T2	estimate, se	method=twostep; vce=classical; weights=none	tests/r_parity/43_heckman.py	sp.heckman(...)
T1	se	method=twostep; vce=classical; weights=none	tests/reference_parity/test_heckman_se_parity.py	sp.heckman(...)
T2	estimate, se	method=ml; vce=classical/cluster/robust; weights=none	tests/reference_parity/test_ldv_design_stata_parity.py	sp.heckman(...)
T2	estimate, se	method=ml; vce=cluster/robust; weights=set	tests/reference_parity/test_ldv_design_stata_parity.py	sp.heckman(...)

vce is ignored for estimate: the maximum-likelihood point estimate is computed before, and independently of, the covariance estimator

honest_did

Dimensions: restriction in {smoothness, relative_magnitude, smoothness_relative_magnitude}; shape in {none, sign, monotone}; interval in {flci, arp_conditional, arp_c_lf, arp_c_f, worst_case_bias}; code_path in {native, r}; target in {single, average, weights}; window in {full, restricted}; alpha in {0.05, other}; m_grid in {zero_only, zero, positive}; rm_grid in {honestdid, extended, user, n/a}; source in {callaway_santanna, sun_abraham, event_study, etwfe, gardner_did, did_imputation, moments, supplied, other}.

Kind	Outputs	Configuration run	Artifact	Entry point
T2	diagnostic	restriction=smoothness; shape=none; interval=flci; code_path=native; target=single; window=full; alpha=0.05; m_grid=positive/zero/zero_only; rm_grid=n/a; source=supplied	tests/r_parity/10_honest_did.py	sp.honest_did(res, e=0, m_grid=M_GRID, method='smoothness')
T2	diagnostic	restriction=relative_magnitude; shape=none; interval=arp_conditional; code_path=native; target=single; window=full; alpha=0.05; m_grid=positive/zero/zero_only; rm_grid=honestdid; source=supplied	tests/r_parity/21_honest_relmags.py	sp.honest_did(res, method='relative_magnitude')
T2	diagnostic	restriction=relative_magnitude; shape=none; interval=arp_conditional; code_path=native; target=single; window=full; alpha=0.05; m_grid=positive/zero/zero_only; rm_grid=honestdid; source=supplied	tests/reference_parity/test_honest_rm_R_parity.py	sp.honest_did(res, method='relative_magnitude')
S	diagnostic	restriction=relative_magnitude; shape=none; interval=arp_c_lf; code_path=native; target=average/single; window=full; alpha=0.05; m_grid=positive/zero/zero_only; rm_grid=honestdid; source=supplied	tests/reference_parity/test_honest_rm_R_parity.py	sp.honest_did(res, method='relative_magnitude')
T2	diagnostic	restriction=relative_magnitude; shape=none; interval=arp_conditional; code_path=native; target=average/single; window=full; alpha=0.05; m_grid=positive/zero/zero_only; rm_grid=honestdid; source=moments	tests/reference_parity/test_honest_did_moments_R_parity.py	sp.honest_did_from_moments(
T4	diagnostic	restriction=relative_magnitude; shape=none; interval=arp_conditional; code_path=native; target=average/single; window=full; alpha=0.05; m_grid=positive/zero/zero_only; rm_grid=extended; source=moments	tests/reference_parity/test_honest_did_moments_R_parity.py	sp.honest_did_from_moments(
T2	diagnostic	restriction=relative_magnitude; shape=none; interval=arp_conditional; code_path=native; target=average/single/weights; window=full; alpha=0.05/other; m_grid=positive/zero/zero_only; rm_grid=honestdid/user; source=moments	tests/reference_parity/test_honest_did_options_R_parity.py	sp.honest_did_from_moments(
T2	diagnostic	restriction=relative_magnitude; shape=none; interval=arp_conditional; code_path=native; target=average/single/weights; window=full/restricted; alpha=0.05/other; m_grid=positive/zero/zero_only; rm_grid=honestdid; source=callaway_santanna	tests/reference_parity/test_honest_did_options_R_parity.py	sp.honest_did(cs, **kw)
S	diagnostic	restriction=relative_magnitude; shape=none; interval=arp_c_lf; code_path=native; target=average/single; window=full; alpha=0.05; m_grid=positive/zero/zero_only; rm_grid=honestdid; source=callaway_santanna	tests/reference_parity/test_honest_did_options_R_parity.py	sp.honest_did(cs, **kw)
T1	diagnostic	restriction=smoothness; shape=none; interval=flci; code_path=native; target=average/single/weights; window=full/restricted; alpha=0.05/other; m_grid=zero_only; rm_grid=n/a; source=callaway_santanna/moments	tests/reference_parity/test_honest_did_options_R_parity.py	sp.honest_did(cs, **kw)
T4	diagnostic	restriction=smoothness; shape=none; interval=flci; code_path=native; target=average/single/weights; window=full/restricted; alpha=0.05/other; m_grid=positive/zero/zero_only; rm_grid=n/a; source=callaway_santanna/moments	tests/reference_parity/test_honest_did_options_R_parity.py	sp.honest_did(cs, **kw)
T4	diagnostic	restriction=smoothness; shape=none; interval=flci; code_path=native; target=average/single; window=full; alpha=0.05; m_grid=positive; rm_grid=n/a; source=moments	tests/reference_parity/test_honest_did_moments_R_parity.py	sp.honest_did_from_moments(
T2	diagnostic	restriction=relative_magnitude; shape=none; interval=arp_conditional; code_path=native; target=average/single/weights; window=restricted; alpha=0.05; m_grid=positive/zero/zero_only; rm_grid=honestdid; source=etwfe	tests/reference_parity/test_honest_did_lvec_window_Stata_parity.py	sp.honest_did(fit, backend='r', ...) vs sp.honest_did(fit, ...)
T4	diagnostic	restriction=relative_magnitude/smoothness; shape=none; interval=arp_conditional/flci; code_path=native; target=average/single; window=restricted; alpha=0.05; m_grid=positive/zero/zero_only; rm_grid=honestdid/n/a; source=etwfe	tests/reference_parity/test_honest_did_lvec_window_Stata_parity.py	sp.honest_did(fit, method=...)
T2	diagnostic	restriction=relative_magnitude; shape=none; interval=arp_conditional; code_path=native; target=single; window=full; alpha=0.05; m_grid=positive/zero/zero_only; rm_grid=honestdid; source=callaway_santanna	tests/reference_parity/test_honest_did_backend_parity.py	sp.honest_did(cs_result, method='relative_magnitude')
T2	diagnostic	restriction=smoothness; shape=monotone/none/sign; interval=arp_conditional; code_path=native; target=average/single; window=full; alpha=0.05; m_grid=positive/zero/zero_only; rm_grid=honestdid; source=moments	tests/reference_parity/test_honest_restrictions_R_parity.py	sp.honest_did_from_moments(
T2	diagnostic	restriction=relative_magnitude/smoothness_relative_magnitude; shape=monotone/none/sign; interval=arp_conditional; code_path=native; target=average/single; window=full; alpha=0.05; m_grid=positive/zero/zero_only; rm_grid=honestdid; source=moments	tests/reference_parity/test_honest_restrictions_R_parity.py	sp.honest_did_from_moments(
T4	diagnostic	restriction=smoothness; shape=monotone/none/sign; interval=arp_c_f; code_path=native; target=single; window=full; alpha=0.05; m_grid=positive/zero/zero_only; rm_grid=honestdid; source=moments	tests/reference_parity/test_honest_restrictions_R_parity.py	sp.honest_did_from_moments(
S	diagnostic	restriction=relative_magnitude/smoothness/smoothness_relative_magnitude; shape=none/sign; interval=arp_c_lf; code_path=native; target=single; window=full; alpha=0.05; m_grid=positive/zero/zero_only; rm_grid=honestdid; source=moments	tests/reference_parity/test_honest_restrictions_R_parity.py	sp.honest_did_from_moments(

ipw

Dimensions: estimand in {ate, att, atc, ato, atm}; se_method in {sandwich, bootstrap}; ps_model in {logit, probit, known}; normalized in {true, false}; trim in {zero, set}; weights in {none, set}; cluster in {none, set}.

Kind	Outputs	Configuration run	Artifact	Entry point
T2	estimate, se	estimand=ate/att; se_method=sandwich; ps_model=logit; normalized=true; trim=zero; weights=none/set; cluster=none/set	tests/reference_parity/test_teffects_design_stata_parity.py	sp.ipw(...)
T2	estimate, se	estimand=ate/att; se_method=sandwich; ps_model=logit/probit; normalized=true; trim=zero; weights=none; cluster=none	tests/reference_parity/test_teffects_probit_stata.py	sp.ipw(...)
T2	estimate, se	estimand=ate/atm/ato/att; se_method=sandwich; ps_model=logit; normalized=true; trim=zero; weights=none; cluster=none	tests/reference_parity/test_barrett_causal_inference_in_r_parity.py	sp.ipw(...)
T2	estimate	estimand=atc; se_method=sandwich; ps_model=logit; normalized=true; trim=zero; weights=none; cluster=none	tests/reference_parity/test_barrett_causal_inference_in_r_parity.py	sp.ipw(...)
T1	estimate	estimand=atc/ate/att; se_method=bootstrap; ps_model=logit; normalized=true; trim=zero; weights=none; cluster=none	tests/reference_parity/test_ipw_parity.py	sp.ipw(...)
T1	estimate	estimand=ate; se_method=bootstrap; ps_model=logit; normalized=false; trim=zero; weights=none; cluster=none	tests/reference_parity/test_ipw_parity.py	sp.ipw(...)

se_method is ignored for estimate: the point estimate is computed before, and independently of, the variance estimator

iv

Dimensions: estimator in {2sls, liml, fuller, gmm}; vce in {classical, hc0, hc1, hc2, hc3, hc4, cr1, cr2, cr3, cluster_multiway, hac, wild, conley, jackknife}; identification in {just, over}; absorb in {none, set}; small in {true, false}.

Kind	Outputs	Configuration run	Artifact	Entry point
T2	joint_test	estimator=2sls; vce=classical/cr1/hc1; identification=over; absorb=none; small=true	tests/reference_parity/test_joint_wald_stata_parity.py	sp.ivreg(formula, data, robust=..., cluster=...)
T2	estimate, se	estimator=2sls; vce=classical/cr1/hc1; identification=just/over; absorb=none; small=true	tests/reference_parity/test_iv_card_aer_parity.py	sp.iv(formula, data, robust=..., cluster=...)
T2	diagnostic	estimator=2sls; vce=classical; identification=over; absorb=none; small=true	tests/reference_parity/test_iv_card_aer_parity.py	sp.iv(formula, data)
T2	estimate, se	estimator=2sls; vce=hc1; identification=just; absorb=none; small=true	tests/r_parity/02_iv.py	sp.ivreg(robust='hc1')
T2	estimate, se	estimator=liml; vce=classical; identification=over; absorb=none; small=true	tests/reference_parity/test_validation_entry_points.py	sp.iv(formula, data, method='liml')
B	coverage	estimator=2sls; vce=hc1; identification=just; absorb=none; small=true	tests/coverage_monte_carlo/results_b1000/coverage_b1000.json	sp.ivreg(robust='hc1')
T2	estimate, se	estimator=2sls/liml; vce=classical/cr1/hc1; identification=just; absorb=none; small=true	tests/reference_parity/test_iv_small_Stata_parity.py	sp.iv(formula, data, method=..., small=True)
T2	estimate, se	estimator=2sls/liml; vce=classical/cr1/hc0; identification=just; absorb=none; small=false	tests/reference_parity/test_iv_small_Stata_parity.py	sp.iv(formula, data, method=..., small=False)
T2	estimate, se	estimator=gmm; vce=cr1/hc1; identification=over; absorb=none; small=true	tests/reference_parity/test_iv_gmm_stata_parity.py	sp.iv(formula, data, method='gmm')
T2	estimate, se	estimator=gmm; vce=cr1/hc0; identification=over; absorb=none; small=false	tests/reference_parity/test_iv_gmm_stata_parity.py	sp.iv(formula, data, method='gmm', small=False)
T2	estimate, se	estimator=gmm; vce=classical; identification=over; absorb=none; small=true	tests/reference_parity/test_iv_gmm_stata_parity.py	sp.iv(formula, data, method='gmm')
T2	estimate, se	estimator=2sls; vce=classical/cluster_multiway/cr1/hc1; identification=just; absorb=set; small=true	tests/reference_parity/test_iv_hdfe_stata_parity.py	sp.iv(formula, data, absorb=[...])
T2	estimate, se	estimator=2sls; vce=cr1/hc1; identification=over; absorb=set; small=true	tests/reference_parity/test_iv_hdfe_stata_parity.py	sp.iv(formula, data, absorb=[...])
T2	estimate	estimator=liml; vce=cr1/hc1; identification=over; absorb=set; small=true	tests/reference_parity/test_iv_hdfe_stata_parity.py	sp.iv(formula, data, absorb=[...], method='liml')
T4	se	estimator=liml; vce=cr1/hc1; identification=over; absorb=set; small=true	tests/reference_parity/test_iv_hdfe_stata_parity.py	sp.iv(formula, data, absorb=[...], method='liml')
T2	estimate	estimator=fuller; vce=hc1; identification=over; absorb=set; small=true	tests/reference_parity/test_iv_hdfe_stata_parity.py	sp.iv(formula, data, absorb=[...], method='fuller')
T4	se	estimator=fuller; vce=hc1; identification=over; absorb=set; small=true	tests/reference_parity/test_iv_hdfe_stata_parity.py	sp.iv(formula, data, absorb=[...], method='fuller')
T2	estimate, se	estimator=2sls; vce=classical/cluster_multiway/cr1/hc0/hc1/hc2/hc3; identification=just/over; absorb=none; small=true	tests/reference_parity/test_vce_menu_parity.py	sp.iv(IV[ident], data, **opts)
T2	estimate, se	estimator=2sls; vce=cr2/cr3/jackknife; identification=just/over; absorb=none; small=true	tests/reference_parity/test_vce_menu_parity.py	sp.ivreg(IV[ident], data, vce=..., cluster='g')
T2	estimate, se	estimator=2sls/liml; vce=classical/cr1/hc1; identification=just/over; absorb=none; small=true	tests/reference_parity/test_vce_menu_parity.py	sp.iv(IV[ident], data, method=..., small=True)
T2	estimate, se	estimator=2sls/liml; vce=classical/cr1/hc0; identification=just/over; absorb=none; small=false	tests/reference_parity/test_vce_menu_parity.py	sp.iv(IV[ident], data, method=..., small=False)
T2	estimate, se	estimator=liml; vce=hc0; identification=just/over; absorb=none; small=true	tests/reference_parity/test_vce_menu_parity.py	sp.iv(IV[ident], data, method='liml', robust='hc0')
T2	estimate, se	estimator=fuller; vce=classical/cr1/hc0/hc1; identification=just/over; absorb=none; small=true	tests/reference_parity/test_vce_menu_parity.py	sp.iv(IV[ident], data, method='fuller', ...)
T2	estimate, se	estimator=2sls; vce=cluster_multiway; identification=just/over; absorb=none; small=true	tests/reference_parity/test_vce_menu_parity.py	sp.ivreg(IV[ident], data, cluster=['g', 'yr'])
T4	se	estimator=2sls/fuller/liml; vce=cluster_multiway; identification=just/over; absorb=none; small=true	tests/reference_parity/test_vce_menu_parity.py	sp.iv(IV[ident], data, method=..., cluster=['g', 'yr'])
T2	estimate, se	estimator=2sls; vce=conley; identification=just/over; absorb=none; small=true	tests/reference_parity/test_vce_menu_parity.py	sp.ivreg(IV[ident], data, vce='conley', ...)
T2	estimate, se	estimator=2sls; vce=wild; identification=just/over; absorb=none; small=true	tests/reference_parity/test_vce_menu_parity.py	sp.ivreg(IV[ident], data, vce='wild', cluster='g')
T2	diagnostic	estimator=2sls; vce=wild; identification=just/over; absorb=none; small=true	tests/reference_parity/test_vce_menu_parity.py	sp.ivreg(IV[ident], data, vce='wild', cluster='g')

vce is ignored for estimate: the least-squares / k-class point estimate is computed before, and independently of, the covariance estimator

ivreg

Dimensions: estimator in {2sls, liml, fuller, gmm}; vce in {classical, hc0, hc1, hc2, hc3, hc4, cr1, cr2, cr3, cluster_multiway, hac, wild, conley, jackknife}; identification in {just, over}; absorb in {none, set}; small in {true, false}.

Kind	Outputs	Configuration run	Artifact	Entry point
T2	joint_test	estimator=2sls; vce=classical/cr1/hc1; identification=over; absorb=none; small=true	tests/reference_parity/test_joint_wald_stata_parity.py	sp.ivreg(formula, data, robust=..., cluster=...)
T2	estimate, se	estimator=2sls; vce=classical/cr1/hc1; identification=just/over; absorb=none; small=true	tests/reference_parity/test_iv_card_aer_parity.py	sp.iv(formula, data, robust=..., cluster=...)
T2	diagnostic	estimator=2sls; vce=classical; identification=over; absorb=none; small=true	tests/reference_parity/test_iv_card_aer_parity.py	sp.iv(formula, data)
T2	estimate, se	estimator=2sls; vce=hc1; identification=just; absorb=none; small=true	tests/r_parity/02_iv.py	sp.ivreg(robust='hc1')
T2	estimate, se	estimator=liml; vce=classical; identification=over; absorb=none; small=true	tests/reference_parity/test_validation_entry_points.py	sp.iv(formula, data, method='liml')
B	coverage	estimator=2sls; vce=hc1; identification=just; absorb=none; small=true	tests/coverage_monte_carlo/results_b1000/coverage_b1000.json	sp.ivreg(robust='hc1')
T2	estimate, se	estimator=2sls/liml; vce=classical/cr1/hc1; identification=just; absorb=none; small=true	tests/reference_parity/test_iv_small_Stata_parity.py	sp.iv(formula, data, method=..., small=True)
T2	estimate, se	estimator=2sls/liml; vce=classical/cr1/hc0; identification=just; absorb=none; small=false	tests/reference_parity/test_iv_small_Stata_parity.py	sp.iv(formula, data, method=..., small=False)
T2	estimate, se	estimator=gmm; vce=cr1/hc1; identification=over; absorb=none; small=true	tests/reference_parity/test_iv_gmm_stata_parity.py	sp.iv(formula, data, method='gmm')
T2	estimate, se	estimator=gmm; vce=cr1/hc0; identification=over; absorb=none; small=false	tests/reference_parity/test_iv_gmm_stata_parity.py	sp.iv(formula, data, method='gmm', small=False)
T2	estimate, se	estimator=gmm; vce=classical; identification=over; absorb=none; small=true	tests/reference_parity/test_iv_gmm_stata_parity.py	sp.iv(formula, data, method='gmm')
T2	estimate, se	estimator=2sls; vce=classical/cluster_multiway/cr1/hc1; identification=just; absorb=set; small=true	tests/reference_parity/test_iv_hdfe_stata_parity.py	sp.iv(formula, data, absorb=[...])
T2	estimate, se	estimator=2sls; vce=cr1/hc1; identification=over; absorb=set; small=true	tests/reference_parity/test_iv_hdfe_stata_parity.py	sp.iv(formula, data, absorb=[...])
T2	estimate	estimator=liml; vce=cr1/hc1; identification=over; absorb=set; small=true	tests/reference_parity/test_iv_hdfe_stata_parity.py	sp.iv(formula, data, absorb=[...], method='liml')
T4	se	estimator=liml; vce=cr1/hc1; identification=over; absorb=set; small=true	tests/reference_parity/test_iv_hdfe_stata_parity.py	sp.iv(formula, data, absorb=[...], method='liml')
T2	estimate	estimator=fuller; vce=hc1; identification=over; absorb=set; small=true	tests/reference_parity/test_iv_hdfe_stata_parity.py	sp.iv(formula, data, absorb=[...], method='fuller')
T4	se	estimator=fuller; vce=hc1; identification=over; absorb=set; small=true	tests/reference_parity/test_iv_hdfe_stata_parity.py	sp.iv(formula, data, absorb=[...], method='fuller')
T2	estimate, se	estimator=2sls; vce=classical/cluster_multiway/cr1/hc0/hc1/hc2/hc3; identification=just/over; absorb=none; small=true	tests/reference_parity/test_vce_menu_parity.py	sp.iv(IV[ident], data, **opts)
T2	estimate, se	estimator=2sls; vce=cr2/cr3/jackknife; identification=just/over; absorb=none; small=true	tests/reference_parity/test_vce_menu_parity.py	sp.ivreg(IV[ident], data, vce=..., cluster='g')
T2	estimate, se	estimator=2sls/liml; vce=classical/cr1/hc1; identification=just/over; absorb=none; small=true	tests/reference_parity/test_vce_menu_parity.py	sp.iv(IV[ident], data, method=..., small=True)
T2	estimate, se	estimator=2sls/liml; vce=classical/cr1/hc0; identification=just/over; absorb=none; small=false	tests/reference_parity/test_vce_menu_parity.py	sp.iv(IV[ident], data, method=..., small=False)
T2	estimate, se	estimator=liml; vce=hc0; identification=just/over; absorb=none; small=true	tests/reference_parity/test_vce_menu_parity.py	sp.iv(IV[ident], data, method='liml', robust='hc0')
T2	estimate, se	estimator=fuller; vce=classical/cr1/hc0/hc1; identification=just/over; absorb=none; small=true	tests/reference_parity/test_vce_menu_parity.py	sp.iv(IV[ident], data, method='fuller', ...)
T2	estimate, se	estimator=2sls; vce=cluster_multiway; identification=just/over; absorb=none; small=true	tests/reference_parity/test_vce_menu_parity.py	sp.ivreg(IV[ident], data, cluster=['g', 'yr'])
T4	se	estimator=2sls/fuller/liml; vce=cluster_multiway; identification=just/over; absorb=none; small=true	tests/reference_parity/test_vce_menu_parity.py	sp.iv(IV[ident], data, method=..., cluster=['g', 'yr'])
T2	estimate, se	estimator=2sls; vce=conley; identification=just/over; absorb=none; small=true	tests/reference_parity/test_vce_menu_parity.py	sp.ivreg(IV[ident], data, vce='conley', ...)
T2	estimate, se	estimator=2sls; vce=wild; identification=just/over; absorb=none; small=true	tests/reference_parity/test_vce_menu_parity.py	sp.ivreg(IV[ident], data, vce='wild', cluster='g')
T2	diagnostic	estimator=2sls; vce=wild; identification=just/over; absorb=none; small=true	tests/reference_parity/test_vce_menu_parity.py	sp.ivreg(IV[ident], data, vce='wild', cluster='g')

vce is ignored for estimate: the least-squares / k-class point estimate is computed before, and independently of, the covariance estimator

logit

Dimensions: vce in {classical, robust, hc0, hc1, cluster}; weights in {none, set}; perfect_prediction in {drop, keep}.

Kind	Outputs	Configuration run	Artifact	Entry point
T2	estimate, se	vce=classical; weights=none; perfect_prediction=drop	tests/r_parity/57_logit.py	sp.logit(formula=...)
T2	estimate, se	vce=classical/cluster/robust; weights=none; perfect_prediction=drop	tests/reference_parity/test_vce_grammar_stata_parity.py	sp.logit(...)
T2	estimate, se	vce=classical/cluster/robust; weights=set; perfect_prediction=drop	tests/reference_parity/test_count_weights_stata_parity.py	sp.logit(...)
T1	estimate, se	vce=classical; weights=set; perfect_prediction=drop	tests/reference_parity/test_count_weights_stata_parity.py	sp.logit(...)
T2	estimate, se	vce=hc0/hc1; weights=none/set; perfect_prediction=drop	tests/reference_parity/test_ml_sandwich_r_parity.py	sp.logit(...)
T2	joint_test	vce=classical/robust; weights=none; perfect_prediction=drop	tests/reference_parity/test_joint_wald_stata_parity.py	sp.logit(...)
T2	joint_test	vce=classical/cluster; weights=none; perfect_prediction=drop	tests/reference_parity/test_postestimation_stata_parity.py	sp.logit(...)
T2	estimate, se	vce=classical; weights=none; perfect_prediction=drop	tests/reference_parity/test_logit_perfect_prediction_Stata_parity.py	sp.logit(...)
T2	estimate, se	vce=classical; weights=none; perfect_prediction=drop	tests/reference_parity/test_regression_stories_r_parity.py	sp.logit(...)

vce is ignored for estimate: the maximum-likelihood point estimate is computed before, and independently of, the covariance estimator

nbreg

Dimensions: vce in {classical, robust, cluster}; weights in {none, set}; dispersion in {nb2, nb1}.

Kind	Outputs	Configuration run	Artifact	Entry point
T2	estimate, se	vce=classical; weights=none; dispersion=nb2	tests/r_parity/42_nbreg.py	sp.nbreg(formula=...)
T4	se	vce=classical; weights=none; dispersion=nb2	tests/r_parity/42_nbreg.py	sp.nbreg(formula=...)
T2	estimate, se	vce=classical/cluster/robust; weights=none; dispersion=nb1/nb2	tests/reference_parity/test_vce_grammar_stata_parity.py	sp.nbreg(...)
T1	estimate, se	vce=classical; weights=set; dispersion=nb2	tests/reference_parity/test_count_weights_stata_parity.py	sp.nbreg(...)

vce is ignored for estimate: the maximum-likelihood point estimate is computed before, and independently of, the covariance estimator

nnmatch

Dimensions: estimand in {att, ate}; metric in {mahalanobis, ivariance, euclidean}; n_matches in {1, 2, 3, 4, 5}; exact in {none, set}; bias_adjust in {none, set}; vce in {robust, iid}; vce_nn in {2, 3, 4, na}; caliper in {none, set}.

Kind	Outputs	Configuration run	Artifact	Entry point
T2	estimate, se	estimand=att; metric=mahalanobis; n_matches=1; exact=none; bias_adjust=none; vce=robust; vce_nn=2; caliper=none	tests/reference_parity/test_nnmatch_teffects_parity.py	sp.match(df, y, treat, covariates, method='nnmatch', ...)
T2	estimate, se	estimand=ate; metric=mahalanobis; n_matches=1; exact=none; bias_adjust=none; vce=robust; vce_nn=2; caliper=none	tests/reference_parity/test_nnmatch_teffects_parity.py	sp.match(df, y, treat, covariates, method='nnmatch', ...)
T2	estimate, se	estimand=att; metric=mahalanobis; n_matches=1; exact=none; bias_adjust=set; vce=robust; vce_nn=2; caliper=none	tests/reference_parity/test_nnmatch_teffects_parity.py	sp.match(df, y, treat, covariates, method='nnmatch', ...)
T2	estimate, se	estimand=att; metric=ivariance; n_matches=3; exact=none; bias_adjust=none; vce=robust; vce_nn=2; caliper=none	tests/reference_parity/test_nnmatch_teffects_parity.py	sp.match(df, y, treat, covariates, method='nnmatch', ...)
T2	estimate, se	estimand=att; metric=euclidean; n_matches=1; exact=none; bias_adjust=none; vce=robust; vce_nn=2; caliper=none	tests/reference_parity/test_nnmatch_teffects_parity.py	sp.match(df, y, treat, covariates, method='nnmatch', ...)
T2	estimate, se	estimand=att; metric=mahalanobis; n_matches=1; exact=none; bias_adjust=none; vce=iid; vce_nn=na; caliper=none	tests/reference_parity/test_nnmatch_teffects_parity.py	sp.match(df, y, treat, covariates, method='nnmatch', ...)
T2	estimate, se	estimand=att; metric=mahalanobis; n_matches=1; exact=none; bias_adjust=none; vce=robust; vce_nn=4; caliper=none	tests/reference_parity/test_nnmatch_teffects_parity.py	sp.match(df, y, treat, covariates, method='nnmatch', ...)
T2	estimate, se	estimand=att; metric=mahalanobis; n_matches=1; exact=set; bias_adjust=set; vce=robust; vce_nn=2; caliper=none	tests/reference_parity/test_nnmatch_teffects_parity.py	sp.match(df, y, treat, covariates, method='nnmatch', ...)
T2	estimate, se	estimand=ate; metric=mahalanobis; n_matches=2; exact=none; bias_adjust=set; vce=robust; vce_nn=2; caliper=none	tests/reference_parity/test_nnmatch_teffects_parity.py	sp.match(df, y, treat, covariates, method='nnmatch', ...)
T2	estimate, se	estimand=att; metric=mahalanobis; n_matches=1/4/5; exact=none; bias_adjust=none; vce=robust; vce_nn=2; caliper=none	tests/reference_parity/test_psm_options_parity.py	sp.match(df, ..., method='nnmatch', n_matches=4)
T2	estimate, se	estimand=ate; metric=mahalanobis; n_matches=4/5; exact=none; bias_adjust=none; vce=robust; vce_nn=2; caliper=none	tests/reference_parity/test_psm_options_parity.py	sp.match(df, ..., method='nnmatch', n_matches=4)
T2	estimate, se	estimand=att; metric=mahalanobis; n_matches=1; exact=none; bias_adjust=none; vce=robust; vce_nn=3; caliper=none	tests/reference_parity/test_psm_options_parity.py	sp.match(df, ..., method='nnmatch', vce_nn=3)
T2	estimate, se	estimand=ate; metric=mahalanobis; n_matches=4; exact=none; bias_adjust=none; vce=robust; vce_nn=3; caliper=none	tests/reference_parity/test_psm_options_parity.py	sp.match(df, ..., method='nnmatch', vce_nn=3)
T2	estimate, se	estimand=att; metric=mahalanobis; n_matches=4; exact=none; bias_adjust=none; vce=robust; vce_nn=2; caliper=set	tests/reference_parity/test_psm_options_parity.py	sp.match(df, ..., method='nnmatch', caliper=3.0)
T2	estimate, se	estimand=ate; metric=mahalanobis; n_matches=1; exact=none; bias_adjust=none; vce=robust; vce_nn=2; caliper=set	tests/reference_parity/test_psm_options_parity.py	sp.match(df, ..., method='nnmatch', caliper=3.0)

panel

Dimensions: method in {fe, twoway, pooled, fd, re, be, mundlak, chamberlain}; vce in {unadjusted, robust, cluster}; ssc in {linearmodels, stata, fixest}; weights in {none, set}.

Kind	Outputs	Configuration run	Artifact	Entry point
T2	estimate	method=fe/re; vce=unadjusted; ssc=linearmodels; weights=none	tests/r_parity/35_panel.py	sp.panel(data=df, ...)
T2	joint_test	method=fe; vce=unadjusted; ssc=linearmodels; weights=none	tests/reference_parity/test_joint_wald_stata_parity.py	sp.panel(data, formula, method='fe')
T2	joint_test	method=fe; vce=unadjusted; ssc=stata; weights=none	tests/reference_parity/test_joint_wald_stata_parity.py	sp.panel(data, formula, method='fe')
T2	joint_test	method=fe; vce=unadjusted; ssc=fixest; weights=none	tests/reference_parity/test_joint_wald_stata_parity.py	sp.panel(data, formula, method='fe')
T2	joint_test	method=fe; vce=cluster; ssc=stata; weights=none	tests/reference_parity/test_joint_wald_stata_parity.py	sp.panel(data, formula, method='fe', cluster=..., ssc=...)
T2	joint_test	method=fe; vce=cluster; ssc=fixest; weights=none	tests/reference_parity/test_joint_wald_stata_parity.py	sp.panel(data, formula, method='fe', cluster=..., ssc=...)
T4	joint_test	method=fe; vce=cluster; ssc=linearmodels; weights=none	tests/reference_parity/test_joint_wald_stata_parity.py	sp.panel(data, formula, method='fe', cluster=...)
T2	estimate, se	method=fd/fe/pooled/re/twoway; vce=cluster/robust/unadjusted; ssc=stata; weights=none	tests/reference_parity/test_panel_ssc_stata_parity.py	sp.panel(data, ..., ssc=ssc)
T2	estimate, se	method=be; vce=unadjusted; ssc=stata; weights=none	tests/reference_parity/test_panel_ssc_stata_parity.py	sp.panel(data, ..., ssc=ssc)
T2	estimate, se	method=mundlak; vce=cluster/unadjusted; ssc=stata; weights=none	tests/reference_parity/test_panel_ssc_stata_parity.py	sp.panel(data, ..., ssc=ssc)
T2	estimate, se	method=chamberlain; vce=cluster; ssc=stata; weights=none	tests/reference_parity/test_panel_ssc_stata_parity.py	sp.panel(data, ..., ssc=ssc)
T2	estimate, se	method=fe/pooled/twoway; vce=cluster/unadjusted; ssc=stata; weights=set	tests/reference_parity/test_panel_ssc_stata_parity.py	sp.panel(data, ..., ssc=ssc)
T2	estimate, se	method=fd/fe/pooled/twoway; vce=cluster/robust/unadjusted; ssc=fixest; weights=none	tests/reference_parity/test_panel_ssc_stata_parity.py	sp.panel(data, ..., ssc=ssc)
T2	estimate, se	method=fe/pooled/twoway; vce=cluster/robust/unadjusted; ssc=fixest; weights=set	tests/reference_parity/test_panel_ssc_stata_parity.py	sp.panel(data, ..., ssc=ssc)

vce is ignored for estimate: the covariance is computed after the coefficients

ssc is ignored for estimate: the small-sample convention scales the covariance only

poisson

Dimensions: vce in {classical, robust, hc0, hc1, cluster}; weights in {none, set}; exposure in {none, set}.

Kind	Outputs	Configuration run	Artifact	Entry point
T2	estimate, se	vce=classical; weights=none; exposure=none	tests/r_parity/58_poisson.py	sp.poisson(formula=...)
T2	estimate, se	vce=classical/cluster/robust; weights=none; exposure=none	tests/reference_parity/test_vce_grammar_stata_parity.py	sp.poisson(...)
T2	estimate, se	vce=classical/cluster/robust; weights=set; exposure=none	tests/reference_parity/test_count_weights_stata_parity.py	sp.poisson(...)
T1	estimate, se	vce=classical; weights=set; exposure=none	tests/reference_parity/test_count_weights_stata_parity.py	sp.poisson(...)
T2	estimate, se	vce=classical/hc1; weights=none; exposure=none	tests/reference_parity/test_accounting_research_parity.py	sp.poisson(...)
T2	estimate, se	vce=hc0/hc1; weights=none/set; exposure=none	tests/reference_parity/test_ml_sandwich_r_parity.py	sp.poisson(...)
T2	estimate, se	vce=classical/hc0; weights=none; exposure=set	tests/reference_parity/test_ml_sandwich_r_parity.py	sp.poisson(...)
T2	joint_test	vce=classical/robust; weights=none; exposure=none	tests/reference_parity/test_joint_wald_stata_parity.py	sp.poisson(...)

vce is ignored for estimate: the maximum-likelihood point estimate is computed before, and independently of, the covariance estimator

ppmlhdfe

Dimensions: fe in {one, multi}; vce in {classical, robust, hc0, hc1, cr1, cluster_multiway}; weights in {none, set}; ssc in {stata, fixest, none}.

Kind	Outputs	Configuration run	Artifact	Entry point
T2	estimate, se	fe=one; vce=robust; weights=none; ssc=fixest/stata	tests/r_parity/37_ppmlhdfe.py	sp.ppmlhdfe(formula=...)
T2	estimate, se	fe=multi; vce=robust; weights=none; ssc=fixest/stata	tests/r_parity/47_ppmlhdfe_3fe.py	sp.ppmlhdfe(formula=...)
T2	estimate, se	fe=one; vce=robust; weights=set; ssc=stata	tests/reference_parity/test_count_weights_stata_parity.py	sp.ppmlhdfe(...)
T2	estimate, se	fe=one; vce=cr1; weights=none/set; ssc=stata	tests/reference_parity/test_count_weights_stata_parity.py	sp.ppmlhdfe(...)
T2	estimate, se	fe=multi; vce=cr1; weights=none; ssc=stata	tests/reference_parity/test_ppmlhdfe_singletons_Stata_parity.py	sp.ppmlhdfe(...)
T2	se	fe=multi; vce=cluster_multiway/cr1; weights=none; ssc=stata	tests/reference_parity/test_ppmlhdfe_twoway_parity.py	sp.ppmlhdfe(...)

vce is ignored for estimate: the point estimate is computed before, and independently of, the covariance estimator

ssc is ignored for estimate: the point estimate is computed before, and independently of, the covariance estimator

probit

Dimensions: vce in {classical, robust, hc0, hc1, cluster}; weights in {none, set}; perfect_prediction in {drop, keep}.

Kind	Outputs	Configuration run	Artifact	Entry point
T2	estimate, se	vce=classical; weights=none; perfect_prediction=drop	tests/r_parity/48_probit.py	sp.probit(formula=...)
T4	se	vce=classical; weights=none; perfect_prediction=drop	tests/r_parity/48_probit.py	sp.probit(formula=...)
T2	estimate, se	vce=classical/cluster/robust; weights=none; perfect_prediction=drop	tests/reference_parity/test_vce_grammar_stata_parity.py	sp.probit(...)
T2	estimate, se	vce=classical/cluster/robust; weights=set; perfect_prediction=drop	tests/reference_parity/test_count_weights_stata_parity.py	sp.probit(...)
T4	se	vce=hc0/hc1; weights=none/set; perfect_prediction=drop	tests/reference_parity/test_ml_sandwich_r_parity.py	sp.probit(...)
T2	joint_test	vce=classical/cluster; weights=none; perfect_prediction=drop	tests/reference_parity/test_postestimation_stata_parity.py	sp.probit(...)

vce is ignored for estimate: the maximum-likelihood point estimate is computed before, and independently of, the covariance estimator

psm

Dimensions: method in {nearest, kernel, radius, stratify}; distance in {propensity, mahalanobis}; n_matches in {1, 2, 3, 4, 5, 6, 7, 8, 9, 10}; replace in {true, false}; caliper in {none, set}; bias_correction in {true, false}; se_method in {abadie_imbens, abadie_imbens_2016, psmatch2, analytic, bootstrap}.

Kind	Outputs	Configuration run	Artifact	Entry point
T2	estimate	method=nearest; distance=propensity; n_matches=1; replace=true; caliper=none; bias_correction=false; se_method=abadie_imbens/abadie_imbens_2016	tests/r_parity/11_psm.py	sp.psm(method='nn')
T2	se	method=nearest; distance=propensity; n_matches=1; replace=true; caliper=none; bias_correction=false; se_method=abadie_imbens_2016	tests/r_parity/11_psm.py	sp.psm(se_method='abadie_imbens_2016')
T2	estimate	method=nearest; distance=propensity; n_matches=1/2; replace=false; caliper=none; bias_correction=false; se_method=abadie_imbens/abadie_imbens_2016	tests/reference_parity/test_matching_r_parity.py	sp.match(lalonde, ..., method='nearest', replace=False)
T2	estimate	method=nearest; distance=mahalanobis; n_matches=1; replace=false; caliper=none; bias_correction=false; se_method=abadie_imbens/abadie_imbens_2016	tests/reference_parity/test_matching_r_parity.py	sp.match(lalonde, ..., distance='mahalanobis', replace=False)
T2	estimate, se	method=nearest; distance=propensity; n_matches=1/10/2/3/4/5/6/7/8/9; replace=true; caliper=none; bias_correction=false; se_method=abadie_imbens	tests/reference_parity/test_psm_options_parity.py	sp.match(df, y='y', treat='d', covariates=X, n_matches=k)
T2	estimate, se	method=nearest; distance=propensity; n_matches=1/10/2/3/4/5/6/7/8/9; replace=true; caliper=none; bias_correction=false; se_method=abadie_imbens_2016	tests/reference_parity/test_psm_options_parity.py	sp.match(df, ..., se_method='abadie_imbens_2016')
T2	estimate, se	method=nearest; distance=propensity; n_matches=1; replace=true; caliper=set; bias_correction=false; se_method=abadie_imbens_2016	tests/reference_parity/test_psm_options_parity.py	sp.match(df, ..., caliper=0.2, se_method='abadie_imbens_2016')
T2	estimate, se	method=nearest; distance=propensity; n_matches=1; replace=true; caliper=none/set; bias_correction=false; se_method=psmatch2	tests/reference_parity/test_psm_options_parity.py	sp.match(df, ..., caliper=0.01, se_method='psmatch2')
T2	estimate	method=nearest; distance=propensity; n_matches=1; replace=true; caliper=set; bias_correction=false; se_method=abadie_imbens	tests/reference_parity/test_psm_options_parity.py	sp.match(df, ..., caliper=0.01)
T4	se	method=nearest; distance=propensity; n_matches=1; replace=true; caliper=set; bias_correction=false; se_method=abadie_imbens	tests/reference_parity/test_psm_options_parity.py	sp.match(df, ..., caliper=0.01)
T4	se	method=nearest; distance=propensity; n_matches=1; replace=false; caliper=none; bias_correction=false; se_method=abadie_imbens	tests/reference_parity/test_psm_options_parity.py	sp.match(df, ..., replace=False, m_order='smallest')
T2	estimate, se	method=nearest; distance=mahalanobis; n_matches=1/3; replace=true; caliper=none; bias_correction=false; se_method=abadie_imbens	tests/reference_parity/test_psm_options_parity.py	sp.match(df, ..., distance='mahalanobis')
T2	estimate, se	method=kernel; distance=propensity; n_matches=1; replace=true; caliper=none; bias_correction=false; se_method=psmatch2	tests/reference_parity/test_psm_options_parity.py	sp.match(df, ..., method='kernel', kernel=k)
T2	estimate, se	method=radius; distance=propensity; n_matches=1; replace=true; caliper=set; bias_correction=false; se_method=psmatch2	tests/reference_parity/test_psm_options_parity.py	sp.match(df, ..., method='radius', caliper=0.02)
T2	estimate, se	method=stratify; distance=propensity; n_matches=1; replace=true; caliper=none; bias_correction=false; se_method=analytic	tests/reference_parity/test_psm_options_parity.py	sp.match(df, ..., method='stratify')
T2	estimate, se	method=stratify; distance=propensity; n_matches=1; replace=true; caliper=none; bias_correction=false; se_method=analytic	tests/reference_parity/test_qiu_jiaping_methods_stata.py	sp.match(data, ..., method='stratify', strata='b1')
T2	estimate, se	method=nearest; distance=propensity; n_matches=1; replace=true; caliper=none; bias_correction=true; se_method=abadie_imbens	tests/reference_parity/test_psm_options_parity.py	sp.match(df, ..., bias_correction=True)

se_method is ignored for estimate: the variance estimator does not change the matches

qreg

Dimensions: vce in {iid, robust, nid, kernel, ker, cluster, powell}; weights in {none, set}.

Kind	Outputs	Configuration run	Artifact	Entry point
T2	estimate, se	vce=nid; weights=none	tests/r_parity/40_qreg.py	sp.qreg(...)
T2	estimate, se	vce=iid/robust; weights=none	tests/reference_parity/test_ldv_design_stata_parity.py	sp.qreg(...)
T2	estimate, se, vcov	vce=cluster/kernel; weights=none	tests/reference_parity/test_qreg_cluster_stata_parity.py	sp.qreg(...)
T2	estimate, se	vce=robust; weights=set	tests/reference_parity/test_ldv_gaps_stata_parity.py	sp.qreg(...)
T2	estimate, se	vce=ker/nid; weights=none/set	tests/reference_parity/test_linear_model_extensions_parity.py	sp.qreg(...)

vce is ignored for estimate: the maximum-likelihood point estimate is computed before, and independently of, the covariance estimator

rddensity

Dimensions: code_path in {native, rddensity}.

Kind	Outputs	Configuration run	Artifact	Entry point
T2	diagnostic	code_path=native	tests/r_parity/09_rddensity.py	sp.rddensity(backend='native')
T2	diagnostic, estimate, se	code_path=native	tests/reference_parity/test_rddensity_parity.py	sp.rddensity(df, x, c)
T2	diagnostic, estimate	code_path=native	tests/reference_parity/test_rddensity_all_stata.py	sp.rddensity(df, x)

rdrobust

Dimensions: design in {sharp, fuzzy, kink}; bwselect in {mserd, msetwo, msesum, msecomb1, msecomb2, cerrd, certwo, cersum, cercomb1, cercomb2, cct, manual}; kernel in {triangular, epanechnikov, uniform}; p in {0, 1, 2, 3, 4}; vce in {nn, hc0, hc1, hc2, hc3, cluster}; covariates in {none, set}; code_path in {native, port}; weights in {none, set}.

Kind	Outputs	Configuration run	Artifact	Entry point
T2	estimate, se	design=sharp; bwselect=mserd; kernel=triangular; p=1; vce=nn; covariates=none; code_path=native; weights=none	tests/r_parity/06_rd.py	sp.rdrobust(df, y, x, c)
T2	estimate, se	design=sharp; bwselect=manual; kernel=triangular; p=1; vce=nn; covariates=none; code_path=native; weights=none	tests/r_parity/06_rd.py	sp.rdrobust(h=15)
B	coverage	design=sharp; bwselect=mserd; kernel=triangular; p=1; vce=nn; covariates=none; code_path=native; weights=none	tests/coverage_monte_carlo/results_b1000/coverage_b1000.json	sp.rdrobust(df, y, x, c)
T2	estimate, se	design=sharp; bwselect=mserd; kernel=triangular; p=1; vce=nn; covariates=none; code_path=native; weights=set	tests/reference_parity/test_rd_weights_parity.py	sp.rdrobust(weights=)
T2	estimate, se	design=sharp; bwselect=msetwo; kernel=triangular; p=1; vce=hc1; covariates=none; code_path=native; weights=set	tests/reference_parity/test_rd_weights_parity.py	sp.rdrobust(weights=)
T2	estimate, se	design=sharp; bwselect=cerrd; kernel=triangular; p=2; vce=nn; covariates=none; code_path=native; weights=set	tests/reference_parity/test_rd_weights_parity.py	sp.rdrobust(weights=)
T2	estimate, se	design=sharp; bwselect=mserd; kernel=triangular; p=1; vce=nn; covariates=set; code_path=native; weights=set	tests/reference_parity/test_rd_weights_parity.py	sp.rdrobust(weights=)
T2	estimate, se	design=sharp; bwselect=mserd; kernel=triangular; p=1; vce=cluster; covariates=none; code_path=native; weights=set	tests/reference_parity/test_rd_weights_parity.py	sp.rdrobust(weights=)
T2	estimate, se	design=sharp; bwselect=mserd; kernel=uniform; p=1; vce=hc3; covariates=none; code_path=native; weights=set	tests/reference_parity/test_rd_weights_parity.py	sp.rdrobust(weights=)
T2	estimate, se	design=sharp; bwselect=manual; kernel=triangular; p=1; vce=nn; covariates=none; code_path=native; weights=set	tests/reference_parity/test_rd_weights_parity.py	sp.rdrobust(weights=)
T2	estimate, se	design=fuzzy; bwselect=mserd; kernel=triangular; p=1; vce=nn; covariates=none; code_path=native; weights=set	tests/reference_parity/test_rd_weights_parity.py	sp.rdrobust(weights=)
T2	estimate, se	design=fuzzy; bwselect=msecomb2; kernel=triangular; p=1; vce=nn; covariates=none; code_path=native; weights=set	tests/reference_parity/test_rd_weights_parity.py	sp.rdrobust(weights=)
T2	estimate, se	design=fuzzy; bwselect=msecomb2; kernel=triangular; p=1; vce=nn; covariates=none; code_path=native; weights=none	tests/reference_parity/test_rd_weights_parity.py	sp.rdrobust(weights=)
T2	estimate, se	design=sharp; bwselect=cerrd/cersum/certwo/mserd/msesum/msetwo; kernel=epanechnikov/triangular/uniform; p=1/2; vce=nn; covariates=none; code_path=native; weights=none	tests/reference_parity/test_rdrobust_parity.py	sp.rdrobust(df, y, x, c)
T2	estimate, se	design=sharp; bwselect=cerrd/cersum/certwo/mserd/msesum/msetwo; kernel=epanechnikov/triangular/uniform; p=0/3/4; vce=nn; covariates=none; code_path=native; weights=none	tests/reference_parity/test_rdrobust_parity.py	sp.rdrobust(df, y, x, c)
T2	estimate, se	design=sharp; bwselect=mserd; kernel=triangular; p=1/2; vce=hc0/hc1/hc2/hc3/nn; covariates=none; code_path=native; weights=none	tests/reference_parity/test_rd_vce_parity.py	sp.rdrobust(df, y, x, c)
T2	estimate, se	design=sharp; bwselect=mserd; kernel=triangular; p=1; vce=hc0/hc2/hc3/nn; covariates=set; code_path=native; weights=none	tests/reference_parity/test_rd_vce_parity.py	sp.rdrobust(df, y, x, c)
T2	estimate, se	design=sharp; bwselect=mserd; kernel=triangular; p=1/2; vce=cluster; covariates=none; code_path=native; weights=none	tests/reference_parity/test_rd_vce_parity.py	sp.rdrobust(df, y, x, c)
T2	estimate, se	design=sharp; bwselect=mserd; kernel=triangular; p=1; vce=cluster; covariates=set; code_path=native; weights=none	tests/reference_parity/test_rd_vce_parity.py	sp.rdrobust(df, y, x, c)
T2	estimate, se	design=sharp; bwselect=cerrd; kernel=triangular; p=1; vce=cluster; covariates=set; code_path=native; weights=none	tests/reference_parity/test_rd_cluster_cer_parity.py	sp.rdrobust(df, y, x, c)
T2	estimate, se	design=fuzzy; bwselect=mserd; kernel=triangular; p=1/2; vce=nn; covariates=none/set; code_path=native; weights=none	tests/reference_parity/test_rdrobust_fuzzy_parity.py	sp.rdrobust(df, y, x, c)
T2	estimate, se	design=fuzzy; bwselect=cerrd/msetwo; kernel=triangular; p=1; vce=nn; covariates=none; code_path=native; weights=none	tests/reference_parity/test_rdrobust_fuzzy_parity.py	sp.rdrobust(df, y, x, c)
T2	estimate, se	design=fuzzy; bwselect=mserd; kernel=uniform; p=1; vce=nn; covariates=none; code_path=native; weights=none	tests/reference_parity/test_rdrobust_fuzzy_parity.py	sp.rdrobust(df, y, x, c)
T2	estimate, se	design=fuzzy; bwselect=mserd; kernel=triangular; p=1; vce=hc0/hc1/hc2/hc3; covariates=none; code_path=native; weights=none	tests/reference_parity/test_rdrobust_fuzzy_parity.py	sp.rdrobust(df, y, x, c)
T2	estimate, se	design=kink; bwselect=mserd; kernel=triangular; p=2; vce=nn; covariates=none; code_path=native; weights=none	tests/reference_parity/test_rdrobust_params_parity.py	sp.rdrobust(df, y, x, c)
T2	estimate, se	design=sharp; bwselect=cercomb1/cercomb2/msecomb1/msecomb2; kernel=epanechnikov/triangular/uniform; p=1/2; vce=nn; covariates=none; code_path=native; weights=none	tests/reference_parity/test_rdrobust_comb_parity.py	sp.rdrobust(df, y, x, c)
T2	estimate, se	design=fuzzy; bwselect=cercomb1/cercomb2/msecomb1/msecomb2; kernel=triangular; p=1; vce=nn; covariates=none; code_path=native; weights=none	tests/reference_parity/test_rdrobust_comb_parity.py	sp.rdrobust(df, y, x, c)
T2	estimate, se	design=kink; bwselect=mserd; kernel=triangular; p=1/2; vce=nn; covariates=none; code_path=native; weights=none	tests/reference_parity/test_rd_iv_rd_R_parity.py	sp.rdrobust(df, y, x, c)
T2	estimate, se	design=kink; bwselect=manual/mserd; kernel=triangular; p=1; vce=hc1; covariates=none; code_path=native; weights=none	tests/reference_parity/test_rd_iv_rd_R_parity.py	sp.rdrobust(df, y, x, c)

regress

Dimensions: vce in {classical, hc0, hc1, hc2, hc3, hc4, cr1, cr2, cr3, cluster_multiway, hac, wild, conley, jackknife}; weights in {none, set}.

Kind	Outputs	Configuration run	Artifact	Entry point
T2	estimate, se	vce=hc1; weights=none	tests/r_parity/01_ols.py	sp.regress(robust='hc1')
T2	estimate, se	vce=classical/cr1/hc1/hc3; weights=none	tests/reference_parity/test_vce_grammar_stata_parity.py	sp.regress(vce=...)
T2	joint_test	vce=classical/cr1/hc1; weights=none	tests/reference_parity/test_r2_postest_parity.py	sp.test(result, 'C(g)[T.2] = ... = 0')
T2	joint_test	vce=classical/cr1/hc1/hc2/hc3; weights=none	tests/reference_parity/test_joint_wald_stata_parity.py	sp.regress(formula, data, robust=..., cluster=...)
T2	estimate, se	vce=hc2/hc3; weights=none	tests/r_parity/55_hc2_hc3.py	sp.regress(robust='hc2'|'hc3')
T2	estimate, se	vce=hc4; weights=none/set	tests/reference_parity/test_linear_model_extensions_parity.py	sp.regress(F_ALL, df, robust=...)
T2	estimate, se	vce=cr1; weights=none	tests/r_parity/14_ols_cluster.py	sp.regress(cluster=...)
T2	estimate, se	vce=hac; weights=none	tests/r_parity/51_newey.py	sp.regress(robust='hac')
T2	estimate, se	vce=classical/cr1/cr2/hc1/hc2/hc3; weights=set	tests/reference_parity/test_regress_vce_weights_stata_parity.py	sp.regress(FORMULA, data, weights=..., vce=...)
T2	estimate, se	vce=cr2; weights=none	tests/reference_parity/test_regress_vce_weights_stata_parity.py	sp.regress(FORMULA, data, vce='cr2', cluster=...)
T1	se	vce=cr3; weights=none/set	tests/reference_parity/test_regress_vce_weights_stata_parity.py	sp.regress(FORMULA, data, vce='cr3', cluster=...)
T4	se	vce=cr3; weights=none	tests/reference_parity/test_regress_vce_weights_stata_parity.py	sp.regress(FORMULA, data, vce='cr3', cluster=...)
T4	se	vce=cluster_multiway; weights=none/set	tests/reference_parity/test_regress_vce_weights_stata_parity.py	sp.regress(FORMULA, data, cluster=[a, b])
B	coverage	vce=hc1; weights=none	tests/coverage_monte_carlo/results_b1000/coverage_b1000.json	sp.regress(robust='hc1')
T2	estimate, se	vce=classical/cluster_multiway/cr1/cr3/hac/hc0/hc1/hc2/hc3/hc4; weights=none/set	tests/reference_parity/test_vce_menu_parity.py	sp.regress(F, data, weights=..., **opts)
T2	estimate, se	vce=cr2; weights=none	tests/reference_parity/test_vce_menu_parity.py	sp.regress(F, data, vce='cr2', cluster='g')
T2	estimate, se	vce=jackknife; weights=none/set	tests/reference_parity/test_vce_menu_parity.py	sp.regress(F, data, vce='jackknife', cluster='g')
T4	se	vce=cr2; weights=set	tests/reference_parity/test_vce_menu_parity.py	sp.regress(F, data, vce='cr2', cluster='g', weights='w')
T2	estimate, se	vce=wild; weights=none	tests/reference_parity/test_vce_menu_parity.py	sp.regress(F, data, vce='wild', cluster='g')
T2	diagnostic	vce=wild; weights=none	tests/reference_parity/test_vce_menu_parity.py	sp.regress(F, data, vce='wild', cluster='g')
T2	estimate, se	vce=conley; weights=none	tests/reference_parity/test_vce_menu_parity.py	sp.regress(F, data, vce='conley', ...)
T2	estimate, se	vce=hac; weights=none/set	tests/reference_parity/test_vce_menu_parity.py	sp.regress(F, data, robust='hac', hac_small=True, ...)

vce is ignored for estimate: the least-squares / k-class point estimate is computed before, and independently of, the covariance estimator

sdid

Dimensions: method in {sdid, sc, did}; se_method in {placebo, bootstrap, jackknife, noinference}; code_path in {native, synthdid, r}; interface in {treated_unit, treat}; design in {block, staggered}; covariates in {none, projected}.

Kind	Outputs	Configuration run	Artifact	Entry point
T2	estimate	method=sdid; se_method=placebo; code_path=native; interface=treated_unit; design=block; covariates=none	tests/r_parity/12_sdid.py	sp.sdid(backend='native')
T2	estimate, se	method=did/sc/sdid; se_method=jackknife/noinference; code_path=native; interface=treat; design=block/staggered; covariates=none/projected	tests/reference_parity/test_sdid_staggered_parity.py	sp.sdid(treat=...)
B	coverage	method=sdid; se_method=placebo; code_path=native; interface=treated_unit; design=block; covariates=none	tests/coverage_monte_carlo/results_b1000/coverage_b1000.json	sp.sdid()
T2	estimate, se	method=did/sc/sdid; se_method=jackknife; code_path=native; interface=treated_unit; design=block; covariates=none	tests/reference_parity/test_did_synth_R_parity.py	sp.sdid(treated_unit=...)
T2	estimate	method=sdid; se_method=placebo; code_path=native; interface=treated_unit; design=block; covariates=none	tests/reference_parity/test_sdid_parity.py	sp.sdid()
S	se	method=did/sc/sdid; se_method=bootstrap/placebo; code_path=native; interface=treated_unit; design=block; covariates=none	tests/reference_parity/test_sdid_resampling_screen.py	sp.sdid(treated_unit=...)
S	se	method=did/sc/sdid; se_method=placebo; code_path=native; interface=treat; design=block; covariates=none	tests/reference_parity/test_sdid_resampling_screen.py	sp.sdid(treat=...)
S	se	method=sdid; se_method=bootstrap/placebo; code_path=native; interface=treat; design=staggered; covariates=none	tests/reference_parity/test_sdid_resampling_screen.py	sp.sdid(treat=...)

se_method is ignored for estimate: the SE method resamples after the unit and time weights are solved; the ATT is unchanged

sun_abraham

Dimensions: control_group in {nevertreated, lastcohort, control_cohort}; aggregation in {fixest_att, event_time}; share_variance in {estimated, fixed}; window_rule in {report, bin, reference}; crit_dist in {normal, t_resid, t_cluster}.

Kind	Outputs	Configuration run	Artifact	Entry point
T2	estimate, se	control_group=nevertreated; aggregation=fixest_att; share_variance=estimated/fixed; window_rule=report; crit_dist=normal	tests/r_parity/05_sunab.py	sp.sun_abraham(aggregation='fixest_att')
T2	estimate, se	control_group=nevertreated; aggregation=event_time; share_variance=estimated/fixed; window_rule=report; crit_dist=normal	tests/reference_parity/test_sunab_event_time_aggregate_parity.py	sp.sun_abraham()
B	coverage	control_group=nevertreated; aggregation=event_time; share_variance=estimated; window_rule=report; crit_dist=normal	tests/coverage_monte_carlo/results_b1000/coverage_b1000.json	sp.sun_abraham()
T2	vcov	control_group=nevertreated; aggregation=event_time; share_variance=fixed; window_rule=report; crit_dist=normal	tests/reference_parity/test_event_study_vcov_R_parity.py	sp.sun_abraham(share_variance=False)
T2	estimate, se	control_group=nevertreated; aggregation=event_time; share_variance=estimated; window_rule=bin; crit_dist=normal	tests/reference_parity/test_dcdh_textbook_stata_parity.py	sp.sun_abraham(window_rule='bin')
T2	estimate, se	control_group=control_cohort/lastcohort; aggregation=event_time/fixest_att; share_variance=estimated/fixed; window_rule=report; crit_dist=normal	tests/reference_parity/test_sunab_options_parity.py	sp.sun_abraham(control_group='lastcohort')
T2	estimate, se	control_group=control_cohort/nevertreated; aggregation=event_time; share_variance=estimated; window_rule=report; crit_dist=normal	tests/reference_parity/test_sunab_control_cohort_parity.py	sp.sun_abraham(control_cohort=2007)
T2	estimate, se	control_group=lastcohort/nevertreated; aggregation=event_time/fixest_att; share_variance=estimated/fixed; window_rule=reference; crit_dist=normal	tests/reference_parity/test_sunab_options_parity.py	sp.sun_abraham(window_rule='reference')
T2	estimate, se, interval	control_group=nevertreated; aggregation=event_time; share_variance=estimated/fixed; window_rule=report; crit_dist=t_cluster/t_resid	tests/reference_parity/test_es_crit_parity.py	sp.sun_abraham(

aggregation is ignored for vcov: the event-time covariance is formed before the summary aggregate is chosen

crit_dist is ignored for estimate, se, vcov: the critical value turns an SE into a p-value and an interval; coefficients and SEs are computed before it is chosen (asserted equal across parity= in tests/reference_parity/test_es_crit_parity.py)

synth

Dimensions: v_method in {equal, nested, regression}; weights in {unique, nonunique}; code_path in {native, synth}.

Kind	Outputs	Configuration run	Artifact	Entry point
T2	estimate	v_method=equal; weights=unique; code_path=native	tests/r_parity/52_scm_unique.py	sp.synth(method='classic')
T4	estimate	v_method=nested; weights=nonunique/unique; code_path=native	tests/r_parity/07_scm.py	sp.synth(method='classic', special predictors)
T1	estimate	v_method=nested/regression; weights=unique; code_path=native	tests/reference_parity/test_synth_unique_options_parity.py	sp.synth(method='classic', special_predictors=...)
T2	se	v_method=equal; weights=unique; code_path=native	tests/reference_parity/test_synth_unique_options_parity.py	sp.synth(method='classic')
T4	estimate	v_method=regression; weights=unique; code_path=native	tests/reference_parity/test_synth_regression_v_stata.py	sp.synth(v_method='regression')
T4	estimate	v_method=regression; weights=unique; code_path=native	tests/reference_parity/test_synth2_stata_parity.py	sp.synth(method='classic', v_method='regression')

tmle

Dimensions: estimand in {ate, att, atc, ey1, ey0, rr, or}; fluctuation in {single, per_arm}; nuisance in {supplied, user_library, default_library}; se_method in {eif, bootstrap}; outcome in {binary, continuous}; weights in {none, set}; cluster in {none, set}.

Kind	Outputs	Configuration run	Artifact	Entry point
T2	estimate, se	estimand=ate; fluctuation=per_arm; nuisance=user_library; se_method=eif; outcome=binary/continuous; weights=none/set; cluster=none/set	tests/reference_parity/test_tmle_design_R_parity.py	sp.tmle(...)
T2	estimate, se	estimand=ate/ey0/ey1; fluctuation=per_arm; nuisance=supplied; se_method=eif; outcome=binary/continuous; weights=none/set; cluster=none/set	tests/reference_parity/test_tmle_parameters_R_parity.py	sp.tmle(...)
T2	estimate, se	estimand=or/rr; fluctuation=per_arm; nuisance=supplied; se_method=eif; outcome=binary; weights=none; cluster=none/set	tests/reference_parity/test_tmle_parameters_R_parity.py	sp.tmle(...)
T2	estimate	estimand=or/rr; fluctuation=per_arm; nuisance=supplied; se_method=eif; outcome=binary; weights=set; cluster=none/set	tests/reference_parity/test_tmle_parameters_R_parity.py	sp.tmle(...)
T4	se	estimand=or; fluctuation=per_arm; nuisance=supplied; se_method=eif; outcome=binary; weights=set; cluster=none/set	tests/reference_parity/test_tmle_parameters_R_parity.py	sp.tmle(...)
T2	estimate, se	estimand=ate; fluctuation=per_arm; nuisance=supplied; se_method=eif; outcome=binary; weights=none; cluster=none	tests/r_parity/72_tmle.py	sp.tmle(fluctuation=...)
T2	estimate, se	estimand=ate; fluctuation=single; nuisance=user_library; se_method=eif; outcome=binary; weights=none; cluster=none	tests/reference_parity/test_tmle_parity.py	sp.tmle(...)

tobit

Dimensions: method in {mle, scls}; vce in {classical, robust, cluster}; weights in {none, set}; censoring in {left, right, two_sided}.

Kind	Outputs	Configuration run	Artifact	Entry point
T2	estimate, se	method=mle; vce=classical; weights=none; censoring=left	tests/r_parity/41_tobit.py	sp.tobit(data=...)
T2	estimate, se	method=mle; vce=classical; weights=none; censoring=left	tests/reference_parity/test_count_quantile_parity.py	sp.tobit(...)
T2	estimate, se	method=mle; vce=classical/cluster/robust; weights=none; censoring=left	tests/reference_parity/test_ldv_design_stata_parity.py	sp.tobit(...)
T2	estimate, se	method=mle; vce=cluster/robust; weights=set; censoring=left	tests/reference_parity/test_ldv_design_stata_parity.py	sp.tobit(...)
T2	estimate, se	method=mle; vce=classical; weights=none; censoring=two_sided	tests/reference_parity/test_ldv_design_stata_parity.py	sp.tobit(...)
T2	estimate, se	method=mle; vce=cluster; weights=set; censoring=two_sided	tests/reference_parity/test_ldv_design_stata_parity.py	sp.tobit(...)
T2	estimate, se	method=mle; vce=cluster/robust; weights=none/set; censoring=two_sided	tests/reference_parity/test_ldv_gaps_stata_parity.py	sp.tobit(...)
T2	estimate, se	method=mle; vce=classical/cluster/robust; weights=none; censoring=right	tests/reference_parity/test_ldv_gaps_stata_parity.py	sp.tobit(...)
T2	estimate	method=scls; vce=classical; weights=none; censoring=left	tests/reference_parity/test_binary_r2_and_scls_parity.py	sp.tobit(...)

vce is ignored for estimate: the maximum-likelihood point estimate is computed before, and independently of, the covariance estimator

xtabond

Dimensions: model in {difference, ah, system_h3_iv_both, system_h3_iv_diff, system_h3_no_iv, system_h2_iv_both, system_h2_iv_diff, system_h2_no_iv, system_iv_level, system_no_constant, h1}; steps in {onestep, twostep, beyond_two}; robust in {true, false}; cluster in {unit, coarser}; transform in {fd, fod}; collapse in {false, true}; gmm_window in {default, max_capped, min_raised, min_and_max}; regressors in {none, x, x_pre_window, x_endo_window, dummies, x_endo_window_dummies, other}; lags in {1, 2, 3plus}; panel in {gap_free, gapped}.

Kind	Outputs	Configuration run	Artifact	Entry point
T2	estimate, se, diagnostic	model=difference; steps=onestep/twostep; robust=false/true; cluster=unit; transform=fd; collapse=false; gmm_window=default; regressors=none; lags=1; panel=gap_free	tests/reference_parity/test_dynpanel_abdata_parity.py	sp.xtabond(...)
T2	estimate, se, diagnostic	model=difference; steps=onestep; robust=true; cluster=unit; transform=fd; collapse=false; gmm_window=default; regressors=none; lags=2; panel=gap_free	tests/reference_parity/test_dynpanel_abdata_parity.py	sp.xtabond(...)
T2	estimate, se	model=difference; steps=onestep; robust=true; cluster=unit; transform=fd; collapse=false; gmm_window=max_capped; regressors=none; lags=1; panel=gap_free	tests/reference_parity/test_dynpanel_abdata_parity.py	sp.xtabond(...)
T2	estimate, se	model=difference; steps=onestep; robust=true; cluster=unit; transform=fd; collapse=false; gmm_window=min_raised; regressors=none; lags=1; panel=gap_free	tests/reference_parity/test_dynpanel_abdata_parity.py	sp.xtabond(...)
T2	estimate, se, diagnostic	model=difference; steps=onestep; robust=true; cluster=unit; transform=fd; collapse=true; gmm_window=default; regressors=none; lags=1; panel=gap_free	tests/reference_parity/test_dynpanel_abdata_parity.py	sp.xtabond(...)
T2	estimate, se, diagnostic	model=difference; steps=onestep/twostep; robust=true; cluster=unit; transform=fd; collapse=false; gmm_window=default; regressors=x; lags=2; panel=gap_free	tests/reference_parity/test_dynpanel_abdata_parity.py	sp.xtabond(...)
T2	estimate	model=difference; steps=onestep; robust=true; cluster=unit; transform=fd; collapse=false; gmm_window=default; regressors=x; lags=2; panel=gap_free	tests/reference_parity/test_dynpanel_abdata_parity.py	sp.xtabond(...)
T2	estimate, se	model=difference; steps=onestep; robust=true; cluster=unit; transform=fd; collapse=false; gmm_window=default; regressors=none; lags=1; panel=gap_free	tests/reference_parity/test_dynpanel_abdata_parity.py	sp.xtabond(...)
T2	estimate, se	model=difference; steps=onestep; robust=true; cluster=unit; transform=fd; collapse=false; gmm_window=default; regressors=x; lags=1; panel=gap_free	tests/r_parity/50_xtabond.py	sp.xtabond(...)
T2	estimate, se	model=difference; steps=onestep; robust=true; cluster=unit; transform=fd; collapse=false; gmm_window=default; regressors=x_endo_window/x_pre_window; lags=2; panel=gap_free	tests/reference_parity/test_dynpanel_abdata_parity.py	sp.xtabond(...)
T2	diagnostic	model=difference; steps=onestep; robust=true; cluster=unit; transform=fd; collapse=false; gmm_window=default; regressors=x; lags=1; panel=gap_free	tests/reference_parity/test_dynpanel_abdata_parity.py	sp.xtabond(...)
T4	diagnostic	model=difference; steps=onestep; robust=true; cluster=unit; transform=fd; collapse=false; gmm_window=default; regressors=none/x; lags=1; panel=gap_free	tests/reference_parity/test_dynpanel_abdata_parity.py	sp.xtabond(...)
T2	estimate, se, diagnostic	model=difference; steps=onestep/twostep; robust=true; cluster=unit; transform=fod; collapse=false; gmm_window=default; regressors=x; lags=1; panel=gap_free	tests/reference_parity/test_dynpanel_abdata_parity.py	sp.xtabond(...)
T2	estimate, se, diagnostic	model=difference; steps=onestep; robust=true; cluster=coarser; transform=fd; collapse=false; gmm_window=default; regressors=x; lags=1; panel=gap_free	tests/reference_parity/test_dynpanel_abdata_parity.py	sp.xtabond(...)
T2	estimate, se	model=ah; steps=onestep; robust=true; cluster=unit; transform=fd; collapse=false; gmm_window=default; regressors=none/x; lags=1; panel=gap_free	tests/reference_parity/test_dynpanel_abdata_parity.py	sp.xtabond(...)
T2	estimate, se	model=difference; steps=onestep/twostep; robust=true; cluster=unit; transform=fd; collapse=false; gmm_window=default; regressors=none; lags=1; panel=gapped	tests/reference_parity/test_dynpanel_abdata_parity.py	sp.xtabond(...)
T2	estimate, se	model=difference; steps=onestep; robust=true; cluster=unit; transform=fod; collapse=false; gmm_window=default; regressors=none; lags=1; panel=gapped	tests/reference_parity/test_dynpanel_abdata_parity.py	sp.xtabond(...)
T2	estimate, se	model=difference; steps=onestep; robust=true; cluster=unit; transform=fd; collapse=false; gmm_window=default; regressors=x; lags=1; panel=gapped	tests/reference_parity/test_dynpanel_abdata_parity.py	sp.xtabond(...)
T2	estimate, se	model=difference; steps=onestep; robust=true; cluster=unit; transform=fd; collapse=true; gmm_window=max_capped/min_and_max; regressors=none; lags=1; panel=gapped	tests/reference_parity/test_dynpanel_abdata_parity.py	sp.xtabond(...)
T1	estimate	model=difference; steps=onestep; robust=true; cluster=unit; transform=fd; collapse=false; gmm_window=default; regressors=x; lags=1; panel=gap_free	tests/reference_parity/test_gmm_dynamic_panel_parity.py	sp.xtabond(...)
S	estimate	model=difference; steps=onestep; robust=true; cluster=unit; transform=fd; collapse=false; gmm_window=default; regressors=x; lags=1; panel=gap_free	tests/reference_parity/test_gmm_dynamic_panel_parity.py	sp.xtabond(...)
T2	estimate, se, diagnostic	model=system_h3_iv_both; steps=onestep/twostep; robust=true; cluster=unit; transform=fd; collapse=false; gmm_window=default; regressors=x; lags=1; panel=gap_free	tests/reference_parity/test_dynpanel_abdata_parity.py	sp.xtabond(method='system')
T2	estimate, se, diagnostic	model=system_h3_iv_both; steps=twostep; robust=true; cluster=unit; transform=fd; collapse=true; gmm_window=default; regressors=x; lags=1; panel=gap_free	tests/reference_parity/test_dynpanel_abdata_parity.py	sp.xtabond(method='system')
T2	estimate, se, diagnostic	model=system_h3_iv_both; steps=onestep/twostep; robust=true; cluster=unit; transform=fod; collapse=false; gmm_window=default; regressors=x; lags=1; panel=gap_free	tests/reference_parity/test_dynpanel_abdata_parity.py	sp.xtabond(method='system')
T2	estimate, se, diagnostic	model=system_h3_iv_both; steps=onestep; robust=true; cluster=coarser; transform=fd; collapse=false; gmm_window=default; regressors=x; lags=1; panel=gap_free	tests/reference_parity/test_dynpanel_abdata_parity.py	sp.xtabond(method='system')
T2	estimate, se	model=system_h3_no_iv; steps=onestep; robust=true; cluster=unit; transform=fd; collapse=false; gmm_window=default; regressors=none; lags=1; panel=gapped	tests/reference_parity/test_dynpanel_abdata_parity.py	sp.xtabond(method='system')
T4	diagnostic	model=system_h3_iv_both; steps=onestep; robust=true; cluster=unit; transform=fd; collapse=false; gmm_window=default; regressors=x; lags=1; panel=gap_free	tests/reference_parity/test_dynpanel_abdata_parity.py	sp.xtabond(method='system')
T2	estimate, se	model=system_h2_no_iv; steps=onestep/twostep; robust=true; cluster=unit; transform=fd; collapse=false; gmm_window=default; regressors=none; lags=1; panel=gap_free	tests/reference_parity/test_dynpanel_abdata_parity.py	sp.xtdpdsys(...)
T2	estimate, se	model=system_h2_iv_diff; steps=onestep/twostep; robust=true; cluster=unit; transform=fd; collapse=false; gmm_window=default; regressors=x; lags=1; panel=gap_free	tests/reference_parity/test_dynpanel_abdata_parity.py	sp.xtdpdsys(...)
T2	estimate, se	model=system_h2_iv_diff; steps=onestep; robust=false; cluster=unit; transform=fd; collapse=false; gmm_window=default; regressors=x; lags=1; panel=gap_free	tests/reference_parity/test_dynpanel_abdata_parity.py	sp.xtdpdsys(...)
T2	estimate, se, diagnostic	model=system_h2_iv_both; steps=onestep; robust=false; cluster=unit; transform=fd; collapse=false; gmm_window=default; regressors=dummies; lags=2; panel=gap_free	tests/reference_parity/test_hansen_methods_stata_parity.py	sp.xtdpdsys(...)
T2	se	model=system_h2_iv_both; steps=onestep; robust=true; cluster=unit; transform=fd; collapse=false; gmm_window=default; regressors=dummies; lags=2; panel=gap_free	tests/reference_parity/test_hansen_methods_stata_parity.py	sp.xtdpdsys(...)
T2	estimate, se	model=system_h2_iv_both; steps=twostep; robust=true; cluster=unit; transform=fd; collapse=false; gmm_window=default; regressors=dummies; lags=2; panel=gap_free	tests/reference_parity/test_hansen_methods_stata_parity.py	sp.xtdpdsys(...)
T2	estimate, se, diagnostic	model=system_h2_no_iv; steps=twostep; robust=false; cluster=unit; transform=fd; collapse=false; gmm_window=default; regressors=none; lags=2; panel=gap_free	tests/reference_parity/test_hansen_methods_stata_parity.py	sp.xtdpdsys(...)
T2	se	model=system_h2_no_iv; steps=twostep; robust=true; cluster=unit; transform=fd; collapse=false; gmm_window=default; regressors=none; lags=2; panel=gap_free	tests/reference_parity/test_hansen_methods_stata_parity.py	sp.xtdpdsys(...)
T2	estimate, se	model=system_h2_iv_both; steps=twostep; robust=true; cluster=unit; transform=fd; collapse=false; gmm_window=max_capped; regressors=x_endo_window_dummies; lags=2; panel=gap_free	tests/reference_parity/test_hansen_methods_stata_parity.py	sp.stata(...)

robust is ignored for estimate: the coefficients are solved before, and independently of, the choice between the classical and the robust / Windmeijer variance (robust= only switches the reported VCE)

xtdpdsys

Dimensions: model in {system_h3_iv_both, system_h3_iv_diff, system_h3_no_iv, system_h2_iv_both, system_h2_iv_diff, system_h2_no_iv, system_iv_level, system_no_constant, h1}; steps in {onestep, twostep, beyond_two}; robust in {true, false}; cluster in {unit, coarser}; transform in {fd, fod}; collapse in {false, true}; gmm_window in {default, max_capped, min_raised, min_and_max}; regressors in {none, x, x_pre_window, x_endo_window, dummies, x_endo_window_dummies, other}; lags in {1, 2, 3plus}; panel in {gap_free, gapped}.

Kind	Outputs	Configuration run	Artifact	Entry point
T2	estimate, se, diagnostic	model=system_h3_iv_both; steps=onestep/twostep; robust=true; cluster=unit; transform=fd; collapse=false; gmm_window=default; regressors=x; lags=1; panel=gap_free	tests/reference_parity/test_dynpanel_abdata_parity.py	sp.xtabond(method='system')
T2	estimate, se, diagnostic	model=system_h3_iv_both; steps=twostep; robust=true; cluster=unit; transform=fd; collapse=true; gmm_window=default; regressors=x; lags=1; panel=gap_free	tests/reference_parity/test_dynpanel_abdata_parity.py	sp.xtabond(method='system')
T2	estimate, se, diagnostic	model=system_h3_iv_both; steps=onestep/twostep; robust=true; cluster=unit; transform=fod; collapse=false; gmm_window=default; regressors=x; lags=1; panel=gap_free	tests/reference_parity/test_dynpanel_abdata_parity.py	sp.xtabond(method='system')
T2	estimate, se, diagnostic	model=system_h3_iv_both; steps=onestep; robust=true; cluster=coarser; transform=fd; collapse=false; gmm_window=default; regressors=x; lags=1; panel=gap_free	tests/reference_parity/test_dynpanel_abdata_parity.py	sp.xtabond(method='system')
T2	estimate, se	model=system_h3_no_iv; steps=onestep; robust=true; cluster=unit; transform=fd; collapse=false; gmm_window=default; regressors=none; lags=1; panel=gapped	tests/reference_parity/test_dynpanel_abdata_parity.py	sp.xtabond(method='system')
T4	diagnostic	model=system_h3_iv_both; steps=onestep; robust=true; cluster=unit; transform=fd; collapse=false; gmm_window=default; regressors=x; lags=1; panel=gap_free	tests/reference_parity/test_dynpanel_abdata_parity.py	sp.xtabond(method='system')
T2	estimate, se	model=system_h2_no_iv; steps=onestep/twostep; robust=true; cluster=unit; transform=fd; collapse=false; gmm_window=default; regressors=none; lags=1; panel=gap_free	tests/reference_parity/test_dynpanel_abdata_parity.py	sp.xtdpdsys(...)
T2	estimate, se	model=system_h2_iv_diff; steps=onestep/twostep; robust=true; cluster=unit; transform=fd; collapse=false; gmm_window=default; regressors=x; lags=1; panel=gap_free	tests/reference_parity/test_dynpanel_abdata_parity.py	sp.xtdpdsys(...)
T2	estimate, se	model=system_h2_iv_diff; steps=onestep; robust=false; cluster=unit; transform=fd; collapse=false; gmm_window=default; regressors=x; lags=1; panel=gap_free	tests/reference_parity/test_dynpanel_abdata_parity.py	sp.xtdpdsys(...)
T2	estimate, se, diagnostic	model=system_h2_iv_both; steps=onestep; robust=false; cluster=unit; transform=fd; collapse=false; gmm_window=default; regressors=dummies; lags=2; panel=gap_free	tests/reference_parity/test_hansen_methods_stata_parity.py	sp.xtdpdsys(...)
T2	se	model=system_h2_iv_both; steps=onestep; robust=true; cluster=unit; transform=fd; collapse=false; gmm_window=default; regressors=dummies; lags=2; panel=gap_free	tests/reference_parity/test_hansen_methods_stata_parity.py	sp.xtdpdsys(...)
T2	estimate, se	model=system_h2_iv_both; steps=twostep; robust=true; cluster=unit; transform=fd; collapse=false; gmm_window=default; regressors=dummies; lags=2; panel=gap_free	tests/reference_parity/test_hansen_methods_stata_parity.py	sp.xtdpdsys(...)
T2	estimate, se, diagnostic	model=system_h2_no_iv; steps=twostep; robust=false; cluster=unit; transform=fd; collapse=false; gmm_window=default; regressors=none; lags=2; panel=gap_free	tests/reference_parity/test_hansen_methods_stata_parity.py	sp.xtdpdsys(...)
T2	se	model=system_h2_no_iv; steps=twostep; robust=true; cluster=unit; transform=fd; collapse=false; gmm_window=default; regressors=none; lags=2; panel=gap_free	tests/reference_parity/test_hansen_methods_stata_parity.py	sp.xtdpdsys(...)
T2	estimate, se	model=system_h2_iv_both; steps=twostep; robust=true; cluster=unit; transform=fd; collapse=false; gmm_window=max_capped; regressors=x_endo_window_dummies; lags=2; panel=gap_free	tests/reference_parity/test_hansen_methods_stata_parity.py	sp.stata(...)
T1	estimate	model=system_h3_iv_both; steps=onestep; robust=true; cluster=unit; transform=fd; collapse=false; gmm_window=default; regressors=x; lags=1; panel=gap_free	tests/reference_parity/test_dynpanel_abdata_parity.py	sp.xtdpdsys(...)

robust is ignored for estimate: the coefficients are solved before, and independently of, the choice between the classical and the robust / Windmeijer variance (robust= only switches the reported VCE)

Option-level Stata fixtures

Fixtures under tests/stata_parity/option_parity/ pin option switches within an estimator and sit outside the Track A enumeration on purpose, so TIER_A_FIXTURE_LOCK.json does not hash them. The SHA-256 below is their lock: an edited fixture fails --check until this file is regenerated. A fixture read by no test fails the build.

Fixture	Read by	SHA-256 (first 16)
tests/stata_parity/option_parity/results/82_csdid_conventions_Stata.json	tests/reference_parity/test_csdid_conventions_stata_parity.py	c988134a90bb0dec
tests/stata_parity/option_parity/results/83_sunab_control_cohort_Stata.json	tests/reference_parity/test_sunab_control_cohort_parity.py tests/reference_parity/test_sunab_options_parity.py	fd4d59285d8606fb
tests/stata_parity/option_parity/results/84_bjs_fe_covariates_Stata.json	tests/reference_parity/test_bjs_fe_covariates_parity.py tests/reference_parity/test_option_fixture_bindings.py	eead3fb4a701b770
tests/stata_parity/option_parity/results/85_multiplegt_dyn_options_Stata.json	tests/reference_parity/test_multiplegt_dyn_options_parity.py	a6b6c27ad860d8a7
tests/stata_parity/option_parity/results/86_lprobust_Stata.json	tests/reference_parity/test_lprobust_parity.py	5dc6478d44596acd
tests/stata_parity/option_parity/results/87_did_had_Stata.json	tests/reference_parity/test_did_had_parity.py	0f158aa295005c95
tests/stata_parity/option_parity/results/92_vce_menu_Stata.json	tests/reference_parity/test_vce_menu_parity.py	320c9f3720262c4f
tests/stata_parity/option_parity/results/93_sunab_options_Stata.json	tests/reference_parity/test_sunab_options_parity.py	696642f7337dbc76
tests/stata_parity/option_parity/results/94_bjs_options_Stata.json	tests/reference_parity/test_bjs_options_parity.py	8f7d765f8a78ad27
tests/stata_parity/option_parity/results/95_rdrobust_comb_Stata.json	tests/reference_parity/test_rdrobust_comb_parity.py	a640f1fd03604042
tests/stata_parity/option_parity/results/96_csdid_options_Stata.json	tests/reference_parity/test_csdid_options_stata_parity.py	f54b8457546789eb
tests/stata_parity/option_parity/results/97_dcdh_options_Stata.json	tests/reference_parity/test_dcdh_option_values_stata_parity.py	f26a2a96ef115c11

Stata translation holdout

A frozen corpus of 42 Stata commands and the numbers Stata 18 MP gives for the 36 that run, scored on five layers by tests/test_stata_translation_holdout.py. Hashed here so the corpus cannot drift towards what the translator already handles.

File	SHA-256 (first 16)
tests/stata_translation_holdout/corpus.json	88e8f103d0d65361
tests/stata_translation_holdout/holdout_Stata.json	4023297b3cd0b260
tests/stata_translation_holdout/holdout_cross.csv	5eca4a3640048a41
tests/stata_translation_holdout/holdout_panel.csv	f5f2b30439c32471

Inference reliability (Track B)

Reference evidence says an estimator reproduces another implementation. Whether its interval covers is a separate question, answered here from the committed Monte Carlo runs under tests/coverage_monte_carlo/results_b1000/. Every replication is in the denominator, failed ones included. MC SE is the Monte Carlo standard error of the rate; a rate within two of them of 0.95 is not distinguishable from nominal at this B.

Design	B	Failed	Coverage	MC SE	Mean SE / MC SD	Within 2 MC SE of 0.95
sp.regress (HC1) on RCT	1000	0	0.952	0.0068	1.001	yes
sp.regress 2x2 DiD	1000	0	0.955	0.0066	1.026	yes
sp.ivreg (HC1) on strong-Z IV	1000	0	0.962	0.0060	0.984	yes
sp.callaway_santanna simple ATT (staggered)	1000	0	0.947	0.0071	1.009	yes
sp.sun_abraham overall ATT (staggered)	1000	0	0.950	0.0069	1.017	yes
sp.panel two-way FE	1000	0	0.948	0.0070	0.990	yes
sp.fast.feols two-way FE (CR1 by unit)	1000	0	0.955	0.0066	0.987	yes
sp.rdrobust sharp (robust CI)	1000	0	0.934	0.0079	0.982	no
sp.sdid placebo (1 treated)	1000	0	0.928	0.0082	0.992	no
sp.ebalance (CIA, ATT)	1000	0	0.945	0.0072	0.969	yes
sp.causal_question(design='dml') IRM ATE	1000	0	0.968	0.0056	1.045	no
sp.dml(model='plr') theta	1000	0	0.883	0.0102	0.885	no
sp.causal_question(design='causal_forest') AIPW ATE	1000	0	0.959	0.0063	1.036	yes

Designs built to break an assumption. The band is the range documented in advance for that design, not a pass mark for nominal coverage.

Stress design	B	Coverage	MC SE	Documented band	Note
sp.ivreg (HC1) weak instrument (pi=0.10)	1000	0.882	0.0102	0.85 to 0.95
sp.callaway_santanna heterogeneous timing+magnitude	1000	0.954	0.0066	0.92 to 0.96
sp.causal_forest AIPW under overlap loss	300	0.900	0.0173	0.85 to 0.99	Capped at B=300 (wall-clock cap; AIPW IF dominates).

Design	B	Size at 5%	MC SE	Power by effect size
sp.regress (HC1) RCT	1000	0.043	0.0064	0.0: 0.043, 0.1: 0.208, 0.2: 0.596, 0.3: 0.903
sp.did 2x2	1000	0.024	0.0048	0.0: 0.024, 0.2: 0.291, 0.4: 0.871, 0.6: 0.996
sp.ivreg strong-Z	1000	0.046	0.0066	0.0: 0.046, 0.2: 0.453, 0.4: 0.935, 0.6: 0.996
sp.rdrobust sharp	500	0.076	0.0119	0.0: 0.076, 0.2: 0.404, 0.4: 0.888, 0.6: 1.000
sp.panel two-way FE	1000	0.052	0.0070	0.0: 0.052, 0.15: 0.231, 0.3: 0.652, 0.45: 0.954
sp.callaway_santanna staggered	300	0.050	0.0126	0.0: 0.050, 0.3: 0.777, 0.6: 1.000, 0.9: 1.000
sp.ebalance (M-estimation SE)	1000	0.055	0.0072	0.0: 0.055, 0.4: 1.000, 0.7: 1.000, 1.0: 1.000

Reliability studies

Simulation studies under tests/reliability/, each with its design fixed in the script before the first run. A row of a study is one method on one design; a fit that was refused counts in the denominator. The table gives, per study and method, how many designs land within two Monte Carlo standard errors of the nominal rate, and the design furthest from it. Every row is in docs/evidence_inventory.json; the reading of each study is in tests/reliability/README.md.

Study	Method	Quantity	B	Designs	At nominal	Furthest from nominal
dml_learners	ols	coverage	300	4	2	0.000 (shape=nonlinear, n=500)
dml_learners	lasso	coverage	300	4	2	0.000 (shape=nonlinear, n=500)
dml_learners	rf	coverage	300	4	1	0.433 (shape=nonlinear, n=500)
dml_learners	gbm	coverage	300	4	0	0.867 (shape=nonlinear, n=500)
dml_learners	stacking	coverage	300	4	2	0.727 (shape=nonlinear, n=500)
extreme_weights_binary_iv	classical	coverage	2000	24	8	0.146 (model=glm, n=1000, sigma=2.0)
extreme_weights_binary_iv	robust	coverage	2000	24	9	0.814 (model=probit, n=200, sigma=2.0)
extreme_weights_fast	iid	coverage	2000	12	2	0.223 (model=fast_fepois, size=200, sigma=2.0)
extreme_weights_fast	hc1	coverage	2000	12	6	0.870 (model=fast_feols, size=50, sigma=2.0)
extreme_weights_fast	cr1	coverage	2000	12	5	0.793 (model=fast_feols, size=50, sigma=2.0)
extreme_weights_fast	classical	coverage	2000	6	2	0.140 (model=nbreg, size=1000, sigma=2.0)
extreme_weights_fast	robust	coverage	2000	6	2	0.841 (model=nbreg, size=200, sigma=2.0)
extreme_weights_models	classical	coverage	2000	12	4	0.143 (model=poisson, size=1000, sigma=2.0)
extreme_weights_models	robust	coverage	2000	12	6	0.840 (model=poisson, size=200, sigma=2.0)
extreme_weights_models	cluster	coverage	2000	6	2	0.777 (model=panel_fe, size=50, sigma=2.0)
extreme_weights	classical	coverage	2000	12	8	0.340 (errors=sampling, n=1000, sigma=2.0)
extreme_weights	hc1	coverage	2000	12	5	0.842 (errors=sampling, n=200, sigma=2.0)
extreme_weights	hc2	coverage	2000	12	6	0.895 (errors=sampling, n=200, sigma=2.0)
extreme_weights	hc3	coverage	2000	12	10	0.932 (errors=sampling, n=1000, sigma=2.0)
few_clusters	cr1	rejection	2000	29	7	0.749 (treated=1, G=40)
few_clusters	cr2	rejection	2000	16	4	0.234 (treated=two, sizes=balanced, G=40)
few_clusters	cr3	rejection	2000	22	12	0.170 (treated=two, sizes=balanced, G=40)
few_clusters	wild	rejection	2000	16	3	0.126 (treated=half, sizes=unbalanced, G=40)
few_clusters	placebo	rejection	2000	5	0	0.139 (treated=10, G=40)
forest_split_evaluation	own/AUTOC	rejection	400	1	0	0.160 (model=tau = 0.3 + 0.0 z, N=150)
forest_split_evaluation	split_bjs/AUTOC	rejection	400	1	0	0.005 (model=tau = 0.3 + 0.0 z, N=150)
forest_split_evaluation	split_forest/AUTOC	rejection	400	1	0	0.005 (model=tau = 0.3 + 0.0 z, N=150)
forest_split_evaluation	own/QINI	rejection	400	1	0	0.122 (model=tau = 0.3 + 0.0 z, N=150)
forest_split_evaluation	split_bjs/QINI	rejection	400	1	0	0.000 (model=tau = 0.3 + 0.0 z, N=150)
forest_split_evaluation	split_forest/QINI	rejection	400	1	0	0.000 (model=tau = 0.3 + 0.0 z, N=150)
forest_split_evaluation	split/gain	rejection	400	1	0	0.000 (model=tau = 0.3 + 0.0 z, N=200)
rate_cross_section	rate_own_ranking	rejection	400	1	0	0.360 (n=800)
rate_cross_section	rate_split	rejection	100	1	0	0.000 (n=800)
rd_mass_points	adjust	coverage	1000	10	6	0.021 (n=4000, support_per_side=5, 976 refused)
rd_mass_points	off	coverage	1000	10	5	0.021 (n=4000, support_per_side=5, 976 refused)
rd_mass_points	cluster	coverage	1000	8	0	0.001 (n=4000, support_per_side=5, 976 refused)
unbalanced_panel	cs	coverage	1000	8	5	0.023 (pattern=attrit_outcome, N=400)
unbalanced_panel	cs_rcs	coverage	1000	8	4	0.000 (pattern=attrit_outcome, N=400)
unbalanced_panel	bjs	coverage	1000	8	5	0.000 (pattern=attrit_outcome, N=400)
unbalanced_panel	twfe	coverage	1000	8	2	0.000 (pattern=attrit_outcome, N=400)

Inputs those do-files read (the other fixtures use tests/orig_parity/data/02_mpdta_original.csv):

Input	SHA-256 (first 16)
tests/stata_parity/option_parity/data_85_dcdh_switch.csv	776bafc6c4c8e2f4
tests/stata_parity/option_parity/data_86_lprobust.csv	895c4205c6d1012b
tests/stata_parity/option_parity/data_87_did_had.csv	7609daa994c6aca4
tests/stata_parity/option_parity/data_94_bjs_options.csv	38192ebdc07262fc
tests/stata_parity/option_parity/data_96_csdid_unbalanced.csv	e63036296fd3fcb1
tests/stata_parity/option_parity/data_97_dcdh_options.csv	dc79c9f35475ab39