Replicating Incels

The Institute for Family Studies is running an article about Incels, Male Sexlessness is Rising But Not for the Reasons Incels Claim, which uses data from the General Social Survey. The GSS is easy to process and use, so I decided to try to replicate the chart in the article

That’s a very steep increase from 2010, and I’m skeptical, especially because, as the article notes, the rise is not supported by NHANES nor NSFG.

The only solid information I could find about how the GSS data is processed is at the botton of the chart:

NHANES and GSSS pooling on samples on 2 years of each side of the label year. … GSS “Never Married”,

To do this analysis, I used the GSS Data Explorer to extract these variables:

year      Gss year for this respondent                       
sexfreq   Frequency of sex during last year
partners  How many sex partners r had in last year
marital   Marital status
id_       Respondent id number

You can download the data package from my GSS project for this analysis .

Based on the comment on the chart, the analysis process:

  • Extract the records for males who are never married and between 22 and 35 years old, inclusive.
  • Group by year and count the number of respondents who report no sex in last year for sexual frequency
  • Group by year and count the number of respondents who report no sex partners in last year
  • Average the two previous counts, per year.
  • Compute the 5 year rolling average of the counts
  • Divide the rolling average by the number of respondents

The query to extract the sub-group of respondents is:

"year >= 1988 and marital == 'Never married' and sex == 'Male' and age >=22 and age<= 35 "

The result is a completely different chart.

I’m not sure what the difference is, but either I or the IFS author, or both, made some errors in the analysis.

Leave a Reply