Since the 1950s, research has shown that people with higher levels of education, especially graduates, have more liberal cultural views than those with less education. These findings have widely been taken to mean that higher education causes liberal views – and, in recent years, some commentators have attacked ‘woke’ universities for ‘indoctrinating’ students.
However, research has begun to show that the effect of education has been overestimated, because people’s pre-existing attitudes, and other important determinants of attitude formation which occur before and after university, haven’t been measured well, or have been missing from the data. That makes it difficult to say if the relationship between education and liberalism is genuinely causal.
Using the data
I used Understanding Society and British Household Panel Study data to explore this question, giving me a maximum sample size of almost 60,000 people – 2,650 graduates and 57,292 non-graduates – over 26 years (1994-2020).
The data tell me people’s attitudes in relation to gender roles, the economy and the environment – and I looked at these attitudes twice. For graduates, I looked at their attitudes pre- and post- university study, and for non-graduates, I observed these no later than when individuals were 20, and again when they were at least 23, because the median age of graduates in my sample was 23.
Theories to explore
Our socio-political attitudes can change throughout life, but tend to be more malleable in childhood and early adulthood – developing as we encounter ‘socialising agents’: our family and friends, the media, school, and the political atmosphere and area we grow up in. So, while it is possible that the experience of studying at university could lead people to develop certain political values, it seems equally plausible that other factors which occur before and after university could contribute, too.
It may be self-selection, meaning that the things that shape our attitudes when we’re growing up – intelligence, parental attitudes, and our family’s socio-economic status – also determine educational attainment. Alternatively, ‘sorting’ is a possibility: having got a degree, people tend to earn more, find more secure work, have higher social status and socialise among different social networks than their less educated counterparts – and these factors influence their attitudes.
It’s a challenge to carry out research which isolates education’s causal effect on attitudes from these other factors, because surveys can’t collect all the data needed to eliminate these confounding effects entirely. I used techniques which control for individuals’ pre-adult experiences and adult status characteristics, such as:
- gender
- cognitive ability (literacy and numeracy)
- attitudes to gender roles, the economy and the environment before adulthood
- parents’ attitudes, income, and education level
- individuals’ occupations in adulthood
These are techniques which have been widely used in existing analyses of the effect of university study on British people’s adult attitudes. However, I also matched all natural, half-, step-, adopted and foster siblings who lived in the same household during childhood, in order to estimate the effects of university study on attitudes only among a sample of people who would likely have experienced the same kinds of environment growing up. Using this within-sibling study design improved my ability to tease out the independent effect of gaining a university degree on British people’s attitudes.
Findings
I found that graduates’ attitudes do, on average, change over the time they’re at university – often more dramatically than those of non-graduates, and often in the opposite direction. Graduates are typically more environmentally friendly and gender egalitarian than non-graduates – but less economically liberal, interestingly.
However, when I took people’s pre- and post-university experiences and characteristics into account, the effects of education shrank considerably. With each attitudinal measure – gender roles, the environment, and the economy – I ran regression models which took into account:
- education on its own
- self-selection
- self-selection and pre-adult attitudes
- self-selection and pre-adult attitudes, again comparing only siblings
In each subsequent model, the effects of university study estimated tended to get smaller. This was particularly true with economic attitudes, where accounting for these confounding factors reduced the effect of education to the point where it became statistically insignificant. So, not only are graduates less economically liberal (or more economically conservative) than their non-graduate counterparts, we can also say that this is entirely down to self-selection and sorting. Differences in British graduates’ and non-graduates’ early life and adult experiences, rather than their differing educational experiences, are the cause of their divergent economic attitudes.
When we look at gender and environmental attitudes, the magnitude of the effect of education also moves towards zero across each subsequent regression model. In this study’s most stringent tests, the within-sibling models, education’s effects on cultural attitudes actually become non-significant. Ultimately, the within-sibling models show that university study has a negative effect on British graduates’ environmental attitudes (with graduates slightly less environmentally friendly than non-graduates) and a small liberalising effect on their reported gender egalitarianism, but neither of these effects are statistically significant.
However, the within-sibling estimates of the effect of university study on British people’s cultural attitudes in my study were considerably less precise than my other estimates, because they were based on smaller sibling-only samples. Had these samples been larger, it is highly likely that even these small effects would have been statistically significant.
Overall, it seemed fair to conclude that although self-selection and sorting are important, they are not the only things driving British individuals’ cultural attitudes. Graduating from higher education does have a direct causal effect on our cultural attitudes, but this effect is very small, and not always in the direction we might expect. There is evidence to suggest that while studying at university does indeed make graduates somewhat more likely to believe in gender equality, it also makes them less environmentally friendly than non-graduates.
What these results tell us
Ultimately, there’s little evidence that people who go to university form distinctively liberal socio-political attitudes as a result. British graduates’ attitudes come about largely because people whose upbringing predisposes them to have certain views are disproportionately likely to go to university. Right-leaning commentators’ claims that universities are hotbeds of left-liberal bias are greatly exaggerated – at least in the British context.
This finding has important implications. It suggests that decades of increasing higher education enrolment rates, which have rapidly boosted the concentration of graduates in British society are unlikely in themselves to have any dramatic, long-term effects on aggregate British attitudes.
The results of this analysis also clearly suggest that within-sibling estimation offers a unique opportunity to provide less biased estimates of higher education’s effect on attitudes. The use of this effective and intuitive technique should be expanded into future research which seeks to isolate the independent effects of gaining educational qualifications on an array of adult outcomes.
One important point that should be made, though, is that the Understanding Society and British Household Panel Survey data does not tell us about people’s social networks. The liberalising effect of education – which is, in any case, very small – could have more to do with people’s peer groups on campus and their conversations and activities together than any ‘indoctrination’ from professors or the curriculum. This is an interesting possibility, which I could not investigate in my study, but should be a priority for future research.
Authors
