general social survey – Eddiethoughts

The General Social Survey asks respondents a few questions about the use of police force. The baseline question is worded, “Are there any situations you can imagine in which you would approve of a policeman striking an adult male citizen?” (Since this is rather wordy and not easily condensed, I’ll be referring to it generally as “approve of police use of force” or something similar henceforth.)

Personally, this feels like a low bar. Off the top of my head, I would approve an officer striking a man who was assaulting someone and not stopping, or who otherwise might be a danger to themselves or others.

It turns out this view isn’t necessarily common, though. Race, education, and sex are important factors, while political ideology seems—surprisingly, in my opinion—not to have a significant relationship to views on the use of force.

Large racial differences in approval of force

There is a strong relationship between race and approval of police use of force. The black-white gap is particularly striking: black people are between 35% and 49% less likely than whites to say they can imagine approving of a policeman striking a man. That’s a massive opinion gap on a pretty fundamental aspect of policing.

Being afraid to walk one’s neighborhood at night doesn’t correspond to increased approval of officers’ use of force. For blacks, being afraid to walk their neighborhood at night actually may be associated with decreased approval of police use of force (p=.06).

There weren’t enough responses for inference on “other” races.

Having a personal relationship to an officer does seem to make black people more likely to be able to imagine an officer justifiably striking a citizen. Notably, black and white people are about equally as likely to have a relationship to a police officer.

Education positively associated with theoretical approval of use of force

Okay, this was a big surprise. Personally, I’d assumed more education would mean less approval of force by police. This felt like a pretty easy extrapolation from existing social science research that negatively correlates educational attainment with authoritarian tendencies, which I figured would make someone more likely to be okay with the police using force. I also thought that since people with advanced degrees tend to live in safer neighborhoods, they’d feel less need for forceful policing.

Admittedly it’s hard to tell if we’re just observing collinearity between race and education, so I created facets by race. Unfortunately, some of the sample sizes are too small for reliable inference (e.g. there are only 11 black graduate degree-holders in the 2018 sample, hence massive confidence intervals). There were only consistently adequate sample sizes among whites, for whom the pattern holds broadly.

No significant relationship between political ideology and views on force

Another foray, another surprise! I’d expected large ideological gaps in approval, which I certainly didn’t find.

Men vs. women

If the above suggest that the people least enthusiastic about police use of force are those most likely to experience it (minorities, people with less education), women’s reliably lower rates of approval of police force complicate that narrative.

First, a bit of housekeeping: I’ve heard from a few people that the previous site theme made it hard to figure out how to follow this blog, so I changed it. If you want to follow, click on the button to the right. Right over here >>>

Also, I’m “proud” to say eddiethoughts.com has already met its implicit 2020 goal of surpassing 2019’s “record-breaking” readership statistics.

Moving on.

Everyone’s familiar with the cat lady archetype: She owns many cats, to which she devotes inordinate attention. She’s eccentric, romantically challenged, getting on in the years, childless, and quite possibly a lesbian. A spinster, if you will. She’s in her 30s, in her 40s, and the Simpsons’ platonic in her golden years.

But how does this stereotype match reality? Thanks to advances in social science (ie the General Social Survey adding questions about pet ownership in 2018) we may finally have the answer.

Defining terms and building a sample

Unfortunately, the GSS doesn’t ask respondents, “Would you call yourself a ‘cat lady?'” so I had to cobble together the parameters from other variables, a task I undertook with all the gravity it deserved. Following are the variables we’ll be looking at. Their GSS synonyms are in parentheses next to the layman’s terms I’m using. If you’re curious about the wording of the questions, you can refer to the codebook.

Independent variables

The following are the criteria to be deemed a certified cat lady:

Cat ownership (CAT)
Owning multiple pets (NUMPETS)
Almost always considering pets members of the family (PETFAM)
Not owning a dog (DOG)

Women who don’t meet these rigorous standards are relegated to the status of “not cat ladies.” I also took a resample of the total sample to create a control group, which I misnamed “total population.”

Note: I restricted the sample populations to women between the ages of 35 and 60. This was a somewhat arbitrary choice I made to reflect the perception that cat ladies are of a certain age while trying to prevent generational differences and age from skewing results. When all was said and done, I was left with a sample size of 573—11 cat ladies and 562 non-cat ladies. I bootstrapped 5,000 samples from this sample to increase statistical significance.

Dependent variables

These are the variables we want to test for group differences, broken into three groups:

Mental health
- Mental health problem diagnosis (DIAGNOSD)
- Number of poor mental health days in the past 30 days (MNTLHLTH)
Social life
- Satisfaction with social activities and relationships (SATSOC).
- How often they dine with three or more friends who aren’t family (DINEFRDS)
Romantic and family life
- Sex frequency in the last year (SEXFREQ)
- Marital status (MARITAL)
- Number of children (CHILDS)
- Sexual orientation (SEXORNT)

Results

The stereotype holds to varying degrees in different areas. In some, it appears flat-out wrong—in others, shockingly accurate. I’ll resist the urge to speculate about the findings and let you make judgements for yourself. Remember to refer back to the “defining terms” section to remind yourself what constitutes a cat lady. Also remember that my definition is…pretty loose and arbitrary. Don’t extrapolate beyond the information presented, and don’t take it too seriously.

Mental health

When it comes to mental health, the stereotype appears thoroughly vindicated. Cat ladies are significantly more likely than the general population of 35-60 year old women and non-cat ladies to have been diagnosed with a mental health problem—nearly 34% of cat ladies have been diagnosed with a mental health problem versus an average rate of about 22%.

Perhaps it’s not surprising, then, that they also report a higher average number of poor mental health days (around 5.23 out of the past 30) than non-cat ladies (4.33) and the total population (4.38).

Crossbars represent 95% confidence intervals.

Social life

By their own estimation, cat ladies are about twice as likely to have unsatisfying social lives as their peers. About 8.8 percent of cat ladies report poor satisfaction with their social activities and relationships, versus 4.2 percent of non-cat ladies and 4.5 percent of the total population of women aged 35-60.

But when it comes to meeting up with friends, cat ladies seem to do as well or better than other women. Sixty-three percent of cat ladies dined with three or more friends at least several times a year, compared with about 56 percent of their peers.

Romantic and family life

In accordance with stereotype, cat ladies seem to have fewer children than average. On average, cat ladies have 2.09 children, compared with about 2.25 for their peers. They are more likely not to have any children, but not by too much.

But in contradiction to the archetype, cat ladies are less likely to be unmarried or have never married than their peers. They are, however, more likely to be divorced.

The sample appears to show that zero percent of cat ladies are widows. This is likely not actually true but a quirk caused by the fact that the relatively small sample (n=11) from which I bootstrapped contained no widowed cat ladies, which isn’t very surprising, if you think about it. It’s a good reminder of the limitations of the techniques I’m using and my own ability. Perhaps someone with greater knowledge would know how to work around this situation.

When it comes to cat ladies’ sex lives, our cultural preconception appears very wrong. Cat ladies are only slightly more likely not to have had sex in the past year than other women, but they’re also over-represented at the higher end of the sex frequency distribution. Forty-eight percent of cat ladies had sex at least once a week, compared to around 40% of other women.

Finally, the question that’s been keeping you all up at night: are cat ladies more likely to be lesbians than the general population? The answer: not really. I’m holding back a plot or summary statistic because I’m planning a related blog post.

Okay, that’s all. If you have questions about the results or methodology, or you’d like another variable investigated, leave it in the comments or contact me directly.