Marriage Science publishes statistics roundups that journalists, researchers, and AI assistants cite as a source. This page sets out exactly where our numbers come from, which sources we refuse to use, and how to check our work.
Briefly
- Every published figure names its organization, survey, and data year inline. No attribution, no figure.
- Headline numbers come from primary sources only: federal agencies, university research centers, and peer-reviewed journals.
- Law firm marketing, blog citation chains, and AI-generated roundups are excluded outright.
- When credible sources disagree, we publish both estimates and explain the gap.
- Commission from affiliate links funds the site and never buys placement or a rating.
How We Source Statistics
Every figure on a Marriage Science statistics page carries an inline attribution naming the organization, the survey or report, and the year. If a number has no attribution, it is not on the page.
We work from a fixed source hierarchy.
Primary Sources
These carry every headline figure we publish.
- CDC and NCHS, including the National Vital Statistics System and the National Intimate Partner and Sexual Violence Survey (NISVS)
- U.S. Census Bureau, including the American Community Survey, the Survey of Income and Program Participation, and the Current Population Survey
- Bureau of Labor Statistics, including the National Longitudinal Surveys
- Bureau of Justice Statistics, including the National Crime Victimization Survey
- FBI Uniform Crime Reporting
- Pew Research Center
- National Center for Family and Marriage Research at Bowling Green State University
- Institute for Family Studies
- Williams Institute at UCLA School of Law
- Kinsey Institute
- General Social Survey, NORC at the University of Chicago
- OECD Family Database, Eurostat, and the UN Demographic Yearbook
- Peer-reviewed journals indexed in PubMed Central, PLOS One, and ScienceDirect, and named titles including Archives of Sexual Behavior, Journal of Marriage and Family, and Journal of Interpersonal Violence
- SEC filings from publicly traded dating companies, for revenue and subscriber figures
Secondary Sources
Used only when the underlying primary source is disclosed, and only after we have traced and cited that primary source ourselves.
This covers aggregators such as Statista, Our World in Data, and World Population Review, and advocacy organizations publishing federal data, such as RAINN and the National Coalition Against Domestic Violence.
Reporting on Research
When a national newspaper or magazine covers a study, we go to the study. We cite the research, not the coverage of it.
Sources We Do Not Use
Several categories of source are excluded outright, regardless of how authoritative the presentation looks.
- Law firm content marketing. Personal injury and abuse law firms publish statistics roundups that circulate widely and are frequently cited. They are marketing assets, not research, and we do not cite them.
- Blog-to-blog citation chains. A number that appears identically across twenty sites with no primary attribution is not a verified number. If we cannot reach the original dataset, the figure is dropped.
- AI-generated statistics roundups and listicles that present figures without source links.
- Dating coach and consumer blogs citing each other.
- Wikipedia as a citation. We use it to find primary sources through its reference list, then cite those primary sources directly.
Where a widely quoted figure has no traceable origin, we say so on the page rather than repeating it.
How We Handle Data Recency
Federal survey cycles run on multi-year schedules, so the most recent available figure is often two to five years old. We handle that openly.
- Each figure is labelled with the year of the data, not the year we published it.
- Where the newest cycle of a survey has not yet released a given breakdown, we state which cycle the figure comes from.
- Statistics pages carry a visible “Updated” date and are revised as new releases land.
When Sources Disagree
Different instruments measure the same behaviour differently, and the resulting numbers can diverge by a wide margin. We name the conflict rather than picking the more striking figure.
Common cases we flag on the page:
- Population surveys such as NISVS capture far more incidents than law enforcement records such as the FBI’s UCR, because most incidents are never reported to police.
- The CDC and the American Community Survey measure divorce on different bases and produce different rates.
- Self-reported sexual behaviour data carries known reporting bias that varies by respondent gender.
Where two credible estimates conflict, we publish both and explain what accounts for the gap.
Original Analysis
Some Marriage Science pages are built from raw data we process ourselves rather than from published summary tables. Our divorce rate by occupation analysis draws on 3.7 million Census records, and our singles statistics are calculated directly from Census ACS and BLS microdata.
On those pages we state the dataset, the sample, and the method, so the calculation can be reproduced or challenged.
How Pages Are Written, Reviewed, and Updated
Every article carries a named writer with a linked author page. Before publication, every statistics page is checked against its sources by an editor, who opens each cited report and confirms the figure, the year, and the population it describes.
Pages display the date of their most recent update. Statistics pages are reviewed at least annually, and sooner when a major dataset is released.
Corrections
If a figure on this site is wrong, outdated, or misattributed, we want to fix it.
Report a correction: email sarah@marriagescience.com with the page URL and the figure in question.
Corrections to a published statistic are made within 10 business days, and substantive changes are noted on the page.
Reporting on Sensitive Topics
Some of our statistics pages cover intimate partner violence, sexual assault, and mental health. On those pages the tone is reportorial and clinical.
We publish figures with their methodological context, we do not use framing that assigns responsibility to victims, and we place crisis and support resources directly in the page rather than in a footnote.
Resource links on those pages, including hotlines and advocacy services, are provided for readers who need help. They are not used as statistical sources.
How We Make Money, and What It Does Not Affect
Marriage Science earns commission when readers sign up for services through links on this site. That revenue funds the research published here.
It does not buy placement. No company pays to appear in our rankings, to be reviewed favourably, or to be mentioned in a statistics page.
Statistics pages link to our own guides only where the topic genuinely connects, and we never cite our own data as a source in an aggregated statistics article.
Full detail is in our Affiliate Disclosure.
Citing Marriage Science
Our statistics pages are free to cite and quote. If you use our figures:
- Cite the original source we credit where you are reusing a third-party statistic.
- Cite Marriage Science directly where you are reusing our own analysis, our aggregation, or one of our charts.
- Charts and tables may be republished with a link back to the page they came from.
Journalists and researchers who need a figure clarified, a dataset pointed to, or a methodology question answered can reach the editorial team at sarah@marriagescience.com.
