Data & construction

The scandals dataset

A hand-coded record of U.S. Catholic clergy abuse scandals — each one geolocated to an exact institution and dated to the day the accusation became public. Below: how it was built, what every field means, and downloads in several formats and levels of aggregation.

At a glance

Scandal events by year

Type-A (red) and Type-B (blue) events per year. The 2002 spike follows the January 2002 Boston Globe investigation that triggered a wave of allegations nationwide.

Type-A Type-B
Methodology

How each scandal was coded

The dataset combines several public sources to pin down, for each accusation, where the accused clergy member was and when the accusation became public.

1

Start from Bishop Accountability

The base list of accused clergy comes from BishopAccountability.org, a non-governmental organization that compiles a public list of U.S. Catholic clergy who have faced a substantiated sexual-abuse allegation (newspaper articles from a reputable source or legal documents filed in court).

2

Identify appointments & the first news article

Complementary sources — newspaper archives, the Official Catholic Directory, the official websites of Catholic institutions, and Google Maps — are used to identify each clergy member's appointment at the time of the accusation and at the time of the alleged abuse, and to find the date of the first news article mentioning each accusation.

3

Geolocate to an exact institution

Every relevant parish, school, or institution is matched to an exact street address and geocoded to latitude/longitude. The analysis does not depend on whether an allegation is true — only on its media repercussions.

4

Classify each event as Type-A or Type-B

Because decades often passed between abuse and revelation, each scandal is one of two types, defined by where the accused clergy member was at the time of the accusation (see below).

Type-A · red dot

A clergy member currently working in a Catholic institution is accused for the first time. Location: the institution where the clergy member works at the time of the public accusation. Date: the first article mentioning the abuse.

Type-B · blue dot

A clergy member is accused of abuse committed while working at an institution in the past. Location: the institution where the abuse allegedly occurred. Date: the first article mentioning the abuse.

Note. If an accusation involves a clergy member working at the same institution where the abuse allegedly took place, it generates one Type-A and one Type-B event. The paper's baseline specification does not discriminate between the two types.
This release

What's in the dataset

This release contains 2,941 geolocated, dated scandal events — 1,117 Type-A and 1,824 Type-B — covering 1977–2010 across the United States, involving 2,076 distinct clergy members.

Built with the published paper's sample-selection criteria (below), this release closely reproduces the counts reported in the 2015 article — 1,125 Type-A and 1,899 Type-B scandals over 1980–2010 — confirming the data construction. Small differences reflect ongoing refinement of geocoding and dates.

Downloads

Data in multiple formats

Event-level data plus aggregations by year and state, and as zip×year and county×year panels — ready for Stata, R, Python, or GIS. Every file carries the Type-A / Type-B split.

Posted for research purposes. This data is made freely available for research and teaching. Preferred citation:
Bottan, N. L., & Perez-Truglia, R. (2015). Losing my religion: The effects of religious scandals on religious participation and charitable giving. Journal of Public Economics, 129, 106–119. doi.org/10.1016/j.jpubeco.2015.07.008

Event level — one row per scandal

FileFormatRowsSize
scandals_events.csvCSV2,9411.0 MBDownload
scandals_events.jsonJSON2,9411.9 MBDownload
scandals_events.geojsonGeoJSON2,9412.1 MBDownload

Aggregations

FileLevelRowsSize
scandals_by_year.csvYear311 KBDownload
scandals_by_state.csvState551 KBDownload
scandals_by_year_state.csvYear × State5409 KBDownload
scandals_by_county_year.csvCounty × Year 1,55639 KBDownload
scandals_by_zip_year.csvZip × Year 2,49748 KBDownload

Unbalanced panel — location-years with zero scandals are not included.

The county×year and zip×year files are sparse panels — one row per location-year with at least one scandal event; location-years with no events are omitted, so cross-sectional totals are recovered by summing over year. The build script that generates every file above is included: build_data.py.

Reference

Data dictionary — event level

Fields in scandals_events.csv / .json. The compact map file uses short keys (e.g. t, p, inst) documented in build_data.py.

FieldTypeDescription
idintUnique event id (sorted by date)
typeA / BScandal type — A (red) or B (blue), as defined above
priesttextAccused clergy member (first + last name)
rolecodeP=Priest, B=Brother, D=Deacon, N=Religious, S=Seminarian
diocesetextDiocese associated with the clergy member
minor_or_adultcodeWhether the alleged victim was a minor (M) or adult
gendercodeReported gender of the alleged victim, where known
n_accusationsintNumber of accusations recorded for the clergy member
dateYYYY-MM-DDDate the accusation became public (the "scandal" date)
year / month / dayintComponents of date
institutiontextParish / school / institution at the event location
addresstextStreet address, city, state, zip of the institution
postaltextZip code of the institution
county / statetextCounty and state of the institution
lat / lonfloatLatitude / longitude (WGS-84)
news_linkurlSource news article for the event (often BishopAccountability)
ba_linkurlBishopAccountability record for the clergy member