IESET.
Hypotheses·labour·spain_reported_sexual_assault_rate_definition_controlled

The Spanish police-recorded sexual-offence rate per 100,000 population cannot be interpreted directly across the 7 October 2022 entry into force of Ley Orgánica 10/2022 ("Ley de garantía integral de la libertad sexual", colloquially "solo sí es sí"), because that statute redefined the boundaries of the sexual-offence category: it replaced the prior distinction between "abuso sexual" (without violence/intimidation) and "agresión sexual" (with violence/intimidation) with a single consent-based "agresión sexual" offence category, broadened what conduct is prosecutable, and reorganised statistical reporting categories used by INE and the Sistema Estadístico de Criminalidad of the Ministerio del Interior.

The hypothesis operationalises two separable empirical questions. (Q1, descriptive-plus-comparative): on the 2015-2021 sample window — entirely pre-reform — does the Spanish reported sexual-offence rate trajectory deviate from a European comparator pool (FRA, ITA, PRT, DEU, SWE, NLD, BEL) after absorbing common-Europe reporting-culture shocks? (Q2, within-country definitional-shift): when the 7 October 2022 law takes effect, what is the level-shift in the Spanish reported-offence series, how much of it is attributable to the definitional reclassification, and how much residual (if any) is attributable to change in underlying reporting propensity or underlying offending? Neither question permits a raw pre-vs-post comparison of the Spanish series in levels, because the measurement instrument changed.

SUPPORTEDengine/runs/spain_reported_sexual_assault_rate_definition_controlled

SUPPORTED — coef=+1.2e+04 (sign matches claim +), p=0.0569

confidence cueThis is a clear pass for the claim as written. It still applies only to this sample, period, and method.

policy briefNeeds review

In ordinary language

In plain terms, this asks whether spain post 2022 10 07 is actually linked to better or worse reported sexual offence rate per 100k from 2010 to 2023.

plain answer

The data clearly moved in the predicted direction. coef=+1.2e+04 (sign matches claim +), p=0.0569

why it matters

Labor-market rules often help some workers while risking job loss or slower hiring for others. This test looks for that tradeoff in observable employment or unemployment data.

how the test works

It compares 8 country or place units from 2010 to 2023, using a panel fe design, with fixed effects for country and year.

what was measured
What changed
  • Spain post 2022 10 07
Possible pathway
  • Reporting propensity proxy
  • Media attention proxy
What we checked
  • Reported sexual offence rate per 100k
  • Spain ministerio interior agresion sexual rate
  • Spain ine condenados sexual rate
what this does not prove

A single test is not the whole truth. It narrows the claim under a specific sample, time period, and method. Strong policy conclusions need the pattern to survive nearby tests, alternative data, and serious objections.

verification

4 input datasets, 0 unresolved missing series, provenance status: partial provenance.

Results

engine/runs/spain_reported_sexual_assault_rate_definition_controlled
Loading chart…

Pre-registration

registration ordering unverified
first-spec commit 4c8ce8e · 2026-07-18T22:11:21Z
run generated · 2026-06-29T17:54:01Z
Run timestamp predates this path's first git-add commit (rebase, rename, or pre-git local run). Spec hash is still the path's first-add commit — not repository HEAD — but ordering is not a clean pre-registration proof.

The Spanish police-recorded sexual-offence rate per 100,000 population cannot be interpreted directly across the 7 October 2022 entry into force of Ley Orgánica 10/2022 ("Ley de garantía integral de la libertad sexual", colloquially "solo sí es sí"), because that statute redefined the boundaries of the sexual-offence category: it replaced the prior distinction between "abuso sexual" (without violence/intimidation) and "agresión sexual" (with violence/intimidation) with a single consent-based "agresión sexual" offence category, broadened what conduct is prosecutable, and reorganised statistical reporting categories used by INE and the Sistema Estadístico de Criminalidad of the Ministerio del Interior. The hypothesis operationalises two separable empirical questions. (Q1, descriptive-plus-comparative): on the 2015-2021 sample window — entirely pre-reform — does the Spanish reported sexual-offence rate trajectory deviate from a European comparator pool (FRA, ITA, PRT, DEU, SWE, NLD, BEL) after absorbing common-Europe reporting-culture shocks? (Q2, within-country definitional-shift): when the 7 October 2022 law takes effect, what is the level-shift in the Spanish reported-offence series, how much of it is attributable to the definitional reclassification, and how much residual (if any) is attributable to change in underlying reporting propensity or underlying offending? Neither question permits a raw pre-vs-post comparison of the Spanish series in levels, because the measurement instrument changed.

Falsification criterion — what would disprove this

set before the run · honoured after

This hypothesis is considered falsified if:

The hypothesis has two parts. Q1 is SUPPORTED if, on the pre-reform 2015-2021 window with reporting-propensity and media-attention controls, the Spain indicator is within ±0.10 log points of zero at p > 0.10 (i.e., Spain is NOT distinguishable from the European reporting-trend) — this would corroborate the interpretation that Spain's pre-reform trajectory tracked the European reporting- culture trend rather than representing a Spanish-specific offending surge. Q1 is NOT SUPPORTED if the Spain indicator is significantly positive (Spain deviates upward from Europe pre-reform), in which case the framework reports that Spain's pre-reform trajectory WAS distinct from Europe and the critics' pre-reform-rising-trend framing is data-supported. Q2 is SUPPORTED if the level-shift β at 2022-10-07 is positive AND the definitional-component decomposition attributes the majority (>50%) of β to reclassification — corroborating the interpretation that the post-2022 reported-rate spike is primarily a measurement- regime change rather than an underlying offending surge. Q2 is NOT SUPPORTED if the residual component is positive and large (>50% of β), in which case the framework reports that even after definitional adjustment there is a substantial residual component which it cannot further decompose from aggregate data and which requires dedicated victimisation-survey evidence (Encuesta Nacional de Violencia contra la Mujer waves pre- and post-reform) to interpret. Either outcome is an honest finding. The hypothesis is designed to be falsifiable in both directions and is pre-registered specifically to prevent selective reporting of whichever direction aligns with the commentator's priors.

formal test & threshold
test:      spain_sexual_offence_rate_two_part_Q1_panel_Q2_event_study
threshold: Q1 SUPPORTED if |β_spain_pre_reform| < 0.10 and p > 0.10. Q2 SUPPORTED if level-shift β > 0 and definitional component accounts for > 50% of β.

Method

Template
panel_fe
Fixed effects
country, year
Clustering
country
Sample
8 countries · 2010 – 2023
Evidence type
causal

Q1 (pre-reform 2015-2021, panel): TWFE with country and year fixed effects. Outcome: log reported sexual-offence rate per 100k. β on a Spain indicator identifies the Spanish deviation from the comparator pool's time-average. Reporting-propensity and media-attention proxies enter as covariates. The test: is Spain's pre-reform 2015-2021 trajectory distinguishable from the European reporting- culture trend, or does it track the pool? Q2 (within-Spain event study, 2018-2023): interrupted time series around the 7 October 2022 effective date of LO 10/2022. Monthly Ministerio del Interior series preferred; annual Eurostat series used where monthly is unavailable. Specification: log(rate) = α + β·post_2022_10_07 + γ·t (trend) + δ·t·post (trend break) + month/ year FE + ε. β and δ estimate the level shift and trend-break at the reform date. The level-shift β in Q2 has TWO empirical interpretations that must be reported separately rather than conflated: (i) DEFINITIONAL component: the portion of the level-shift attributable to offences previously categorised as "abuso sexual" (or other categories) being reclassified into the post-reform "agresión sexual" category. This is estimated by summing the pre-reform abuso + agresión cells and comparing to the post-reform consolidated cell — a same-definition comparison that removes the reclassification mechanically. (ii) RESIDUAL component: the portion remaining after the definitional reclassification is removed. This residual is itself a composite of (a) change in underlying offending rate, (b) change in victim reporting propensity induced by the law's public-communication campaign and surrounding media environment, and (c) change in police classification/recording practices under the new statute. The framework CANNOT separate (a)/(b)/(c) from aggregate police-recorded data alone; this limitation is pre- registered and the result card must state it. Robustness: (R1) drop DEU from Q1 pool due to 2017 StGB §177 reform creating a similar definitional discontinuity in the German series; (R2) use convictions-based INE series for Spain as triangulation; (R3) report Eurostat's own footnote flags for the Spanish cell and honour Eurostat's data-comparability notes explicitly.

Data

VariableSourceTransform
reported_sexual_offence_rate_per_100k
outcome
eurostat:crim_off_cattier 1
rate_per_100k
spain_ministerio_interior_agresion_sexual_rate
outcome
manual: Ministerio del Interior — Sistema Estadístico de Criminalidad, Balances Trimestrales / Anales. Not currently in an API fetcher; manual-drop under data/manual/spain_mi/.rate_per_100k
spain_ine_condenados_sexual_rate
outcome
manual: INE Estadística de Condenadosrate_per_100k
spain_post_2022_10_07
treatment
constructed:indicator = 1 if country=ESP and year >= 2023 (or month >= 2022-10 for monthly specification); 0 otherwise. Effective datier 5
indicator
reporting_propensity_proxy
channel
constructed:European victimisation-survey series on reporting rate to police, where available (FRA-CVS, ITA-ISTAT delitti, ESP-ENVS tier 5
level
media_attention_proxy
channel
constructed:Google Trends index for country-language terms 'agresión sexual' / 'violación' / equivalents; annual average. Proxy for tier 5
level
migration_composition_share
channel
world_bank_wdi:SM.POP.TOTL.ZStier 2
level
log_population
control
world_bank_wdi:SP.POP.TOTLtier 2
log
urbanisation
control
world_bank_wdi:SP.URB.TOTL.IN.ZStier 2
level
share_population_age_15_34
control
world_bank_wdi:SP.POP.1564.TO.ZStier 2
level
unemployment_rate
control
world_bank_wdi:SL.UEM.TOTL.ZStier 2
level

● ready  · ● pending  · ● reconstruct-needed

Detailed result card

Result card — spain_reported_sexual_assault_rate_definition_controlled

Verdict: SUPPORTED — coef=+1.2e+04 (sign matches claim +), p=0.0569

Pre-registration

  • Claim: The Spanish police-recorded sexual-offence rate per 100,000 population cannot be interpreted directly across the 7 October 2022 entry into force of Ley Orgánica 10/2022 ("Ley de garantía integral de la libertad sexual", colloquially "solo sí es sí"), because that statute redefined the boundaries of the sexual-offence category: it replaced the prior distinction between "abuso sexual" (without violence/intimidation) and "agresión sexual" (with violence/intimidation) with a single consent-based "agresión sexual" offence category, broadened what conduct is prosecutable, and reorganised statistical reporting categories used by INE and the Sistema Estadístico de Criminalidad of the Ministerio del Interior. The hypothesis operationalises two separable empirical questions. (Q1, descriptive-plus-comparative): on the 2015-2021 sample window — entirely pre-reform — does the Spanish reported sexual-offence rate trajectory deviate from a European comparator pool (FRA, ITA, PRT, DEU, SWE, NLD, BEL) after absorbing common-Europe reporting-culture shocks? (Q2, within-country definitional-shift): when the 7 October 2022 law takes effect, what is the level-shift in the Spanish reported-offence series, how much of it is attributable to the definitional reclassification, and how much residual (if any) is attributable to change in underlying reporting propensity or underlying offending? Neither question permits a raw pre-vs-post comparison of the Spanish series in levels, because the measurement instrument changed.
  • Falsification rule: The hypothesis has two parts. Q1 is SUPPORTED if, on the pre-reform 2015-2021 window with reporting-propensity and media-attention controls, the Spain indicator is within ±0.10 log points of zero at p > 0.10 (i.e., Spain is NOT distinguishable from the European reporting-trend) — this would corroborate the interpretation that Spain's pre-reform trajectory tracked the European reporting- culture trend rather than representing a Spanish-specific offending surge. Q1 is NOT SUPPORTED if the Spain indicator is significantly positive (Spain deviates upward from Europe pre-reform), in which case the framework reports that Spain's pre-reform trajectory WAS distinct from Europe and the critics' pre-reform-rising-trend framing is data-supported. Q2 is SUPPORTED if the level-shift β at 2022-10-07 is positive AND the definitional-component decomposition attributes the majority (>50%) of β to reclassification — corroborating the interpretation that the post-2022 reported-rate spike is primarily a measurement- regime change rather than an underlying offending surge. Q2 is NOT SUPPORTED if the residual component is positive and large (>50% of β), in which case the framework reports that even after definitional adjustment there is a substantial residual component which it cannot further decompose from aggregate data and which requires dedicated victimisation-survey evidence (Encuesta Nacional de Violencia contra la Mujer waves pre- and post-reform) to interpret. Either outcome is an honest finding. The hypothesis is designed to be falsifiable in both directions and is pre-registered specifically to prevent selective reporting of whichever direction aligns with the commentator's priors.
  • Falsification test: spain_sexual_offence_rate_two_part_Q1_panel_Q2_event_study

Estimate

  • Method: linearmodels.PanelOLS
  • Coefficient (treatment): +1.2e+04
  • Std error: 6220
  • p-value: 0.0569
  • Observations: 112, countries: 8
  • Within R²: -0.282
  • Fixed effects: entity=True, time=True
  • Clustering: country

Variables resolved

  • eurostat:crim_off_cat → reported_sexual_offence_rate_per_100k (outcome, publisher=eurostat, n=671)
  • constructed: indicator = 1 if country=ESP and year >= 2023 (or month >= 2022-10 for monthly specification); 0 otherwise. Effective date of LO 10/2022 is 7 October 2022; year-resolution treatment is coded from 2023 with 2022 treated as a transition year in robustness. → spain_post_2022_10_07 (treatment, publisher=constructed, n=112)
  • constructed: European victimisation-survey series on reporting rate to police, where available (FRA-CVS, ITA-ISTAT delitti, ESP-ENVS 2019). Sparse; enters as a level covariate where country-year coverage permits. → reporting_propensity_proxy (decomposition_channels, publisher=constructed, n=112)
  • world_bank_wdi:SM.POP.TOTL.ZS → migration_composition_share (decomposition_channels, publisher=world_bank_wdi, n=2080)
  • world_bank_wdi:SP.POP.TOTL → log_population (controls, publisher=world_bank_wdi, n=14447)
  • world_bank_wdi:SP.URB.TOTL.IN.ZS → urbanisation (controls, publisher=world_bank_wdi, n=16965)
  • world_bank_wdi:SP.POP.1564.TO.ZS → share_population_age_15_34 (controls, publisher=world_bank_wdi, n=16965)
  • world_bank_wdi:SL.UEM.TOTL.ZS → unemployment_rate (controls, publisher=world_bank_wdi, n=6874)

Variables missing data

  • manual: Ministerio del Interior — Sistema Estadístico de Criminalidad, Balances Trimestrales / Anales. Not currently in an API fetcher; manual-drop under data/manual/spain_mi/. (outcome, name=spain_ministerio_interior_agresion_sexual_rate) — vintage not on disk
  • manual: INE Estadística de Condenados (outcome, name=spain_ine_condenados_sexual_rate) — vintage not on disk
  • constructed: Google Trends index for country-language terms 'agresión sexual' / 'violación' / equivalents; annual average. Proxy for reporting-climate effects (MeToo waves 2017-2018, La Manada verdict 2018-2019, public reaction to LO 10/2022). (decomposition_channels, name=media_attention_proxy) — vintage not on disk

Generated by scripts/run_panel_fe.py at 2026-06-29T17:54:01+00:00

Strongest opposing argument

Every hypothesis ships with its charitable opposing argument. The framework earns credibility by handling objections at their strongest, not weakest.

Notes

This hypothesis is the framework's response to a politically-charged topic where raw descriptive statistics are routinely deployed without the instrumental-discontinuity controls they require. The framework's value-add is not to take sides on the LO 10/2022 debate — which is a legal and political debate the framework has no special standing on — but to enforce the measurement discipline that the reform created a definitional discontinuity in the reported-rate series and that any claim about trajectory across 2022-10-07 must decompose definitional from residual components before it can be interpreted. Data access limits: Eurostat crim_off_cat has harmonisation caveats Spain-specific; Ministerio del Interior series requires manual-drop fetching as of v1; INE Estadística de Condenados requires separate access. Victimisation-survey triangulation (ENVS 2019, pre-reform) is available but only one wave within the sample; a post-reform ENVS wave is essential for the strongest version of Q2 and is called out as a v2 requirement. This hypothesis MUST NOT be read as advancing a normative claim about LO 10/2022. It advances only the measurement claim that the reported-rate series cannot be interpreted raw across the reform date.

Authored framework. Read the transparency note.