The Effect of Education on Political Knowledge: Evidence From Monozygotic Twins
| Author | Aaron C. Weinschenk,Christopher T. Dawes |
| DOI | 10.1177/1532673X18788048 |
| Date | 01 May 2019 |
| Published date | 01 May 2019 |
| Subject Matter | Articles |
research-article2018
Article
American Politics Research
2019, Vol. 47(3) 530 –548
The Effect of Education
© The Author(s) 2018
Article reuse guidelines:
on Political Knowledge:
sagepub.com/journals-permissions
https://doi.org/10.1177/1532673X18788048
DOI: 10.1177/1532673X18788048
journals.sagepub.com/home/apr
Evidence From
Monozygotic Twins
Aaron C. Weinschenk1
and Christopher T. Dawes2
Abstract
Political scientists have long been interested in the determinants of
political knowledge. In many studies, education is the strongest predictor
of political knowledge. However, some studies have found that education
has no effect on knowledge once confounding variables are taken into
account. In addition, some recent work suggests that education remains the
strongest predictor of knowledge even after accounting for confounders like
personality traits and intelligence. We provide new evidence on the effect
of education on political knowledge by utilizing the co-twin control design.
By looking at the relationship between education and knowledge within
monozygotic twin pairs, we are able to circumvent sources of confounding
of the relationship due to genetic factors and early-life family environment
because monozygotic twins share both. We find that the relationship
between education and political knowledge is highly confounded by genes
and/or familial environment. The results from a naive model that does not
take into account unobserved family factors indicate that education has a
positive and statistically significant effect on political knowledge. However,
in a twin fixed-effects model that accounts for confounding due to genetic
factors and familial socialization, we find that the effect of education on
1University of Wisconsin–Green Bay, WI, USA
2New York University, New York City, NY, USA
Corresponding Author:
Aaron C. Weinschenk, Department of Political Science, University of Wisconsin–Green Bay,
2420 Nicolet Drive, Green Bay, WI 54311, USA.
Email: weinscha@uwgb.edu
Weinschenk and Dawes
531
political knowledge drops substantially and is not statistically significant at
conventional levels.
Keywords
education, political knowledge, discordant twin design, co-twin control
design, monozygotic twins
Introduction
Political knowledge is a central concept in the political behavior literature.
Indeed, scholars have spent a great deal of time trying to identify its causes
and consequences (Arceneaux, Johnson, & Maes, 2012; Delli Carpini &
Keeter, 1996; Galston, 2001; Highton, 2009; Luskin, 1990).1 Theoretical
accounts of democratic citizenship and empirical studies in political behavior
suggest that political knowledge is important because it underlies many desir-
able behaviors and attitudes, including political participation and political
tolerance. In summarizing the consequences of political knowledge, Delli
Carpini and Keeter (1996) note that “all things being equal, the more informed
people are, the better able they are to perform as citizens” (p. 219).
Scholarship on the determinants of political knowledge has shown that a
variety of factors influence political knowledge, including numerous demo-
graphic attributes and some environmental factors. One finding that has
emerged across many studies is a positive relationship between educational
attainment and political knowledge. In their seminal work on political knowl-
edge, Delli Carpini and Keeter (1996) find that even in the presence of 19
demographic and attitudinal control variables, education is “the strongest sin-
gle predictor of political knowledge” (p. 188). The finding that education posi-
tively impacts political knowledge has not gone unscrutinized, however. Some
scholars have noted that the relationship between education and political
knowledge is likely confounded by other variables, such as personality traits
and cognitive ability, both of which are partially heritable, family background,
and/or socialization. In fact, Highton (2009) goes so far as to argue that
due to interpretative difficulties relating to cognitive skill and the exclusion of
factors that likely cause sophistication and are also related to educational
attainment, virtually all previous studies of the causes of political sophistication
are severely limited in their ability to ascertain whether the variable most
commonly associated with sophistication is actually a cause of it. (p. 1566)
In this research note, we provide new evidence on the effect of education
on political knowledge. As we describe below, there is disagreement across
532
American Politics Research 47(3)
empirical studies about the extent to which education exerts a causal effect on
political knowledge. In light of this debate, it is critical to develop an under-
standing of whether the relationship between education and political knowl-
edge is causal or confounded. If, after several decades of research on the
underpinnings of political knowledge, we remain unsure whether or not the
relationship between education and knowledge is causal, then our under-
standing of political knowledge would appear to lack depth. Importantly, if it
turns out that education and knowledge are correlated but not causally related,
then many scholars have been misinterpreting the link between these two
variables for some time now.2 Our goal is to clarify the nature of the relation-
ship between educational attainment and political knowledge.3 We follow the
recent line of quasi-experimental research by using a design not previously
employed to study the effect of education on political knowledge—the so-
called “co-twin control design” (McGue, Osler, & Christensen, 2010), in
which the association between education and knowledge is analyzed within
monozygotic (MZ) twin pairs. The strength of this design lies in the fact that
MZ twins are genetically identical and have been exposed to the same family
environment (given that they grew up together), which allows us to bypass
some of the most likely unobserved variables threatening to bias the esti-
mated impact of education on political knowledge, such as early-life social-
ization in the family and heritable psychological traits.4
Existing Evidence
Many studies have examined the relationship between education and politi-
cal knowledge, and scholars often make the assumption that education
plays a causal role in fostering political knowledge.5 It is worth noting that
many studies on the relationship between education and knowledge about
politics, including foundational work by Delli Carpini and Keeter (1996),
rely on cross-sectional survey data. Of course, one key concern when using
cross-sectional surveys is that it can be difficult to sort out questions about
causality—Unmeasured factors may affect the relationship between cause
and effect.
There are a number of possible factors that might confound the relation-
ship between education and political knowledge. Cognitive ability is an obvi-
ous potential confounder. As Highton (2009) points out, cognitive ability
likely fosters political knowledge and influences whether one attends and
graduates from college. Personality traits are another possible source of con-
founding. The Big Five personality traits, for example, are correlated with
political knowledge (Gerber, Huber, Doherty, & Dowling, 2011a, 2011b;
Mondak, 2010) and educational attainment (Anger, 2013; van Eijck & de
Weinschenk and Dawes
533
Graaf, 2004). Previous studies have demonstrated that both cognitive ability
and personality traits are partially heritable (Deary, Spinath, & Bates, 2006;
Haworth et al., 2009; Jang, Livesley, & Vernon, 1996; Loehlin, McCrae,
Costa, & John, 1998). Political socialization is yet another potential con-
founding variable. Family experiences (or attributes like socioeconomic sta-
tus) may influence one’s educational attainment and political knowledge
(Highton, 2009).
A few studies have attempted to account for some of the confounders men-
tioned above. For example, Luskin (1990) is able to include a measure of
intelligence in his model of political knowledge; he makes use of interviewer
ratings of respondent intelligence. He finds that intelligence has an important
effect on knowledge but that education does not have an independent effect,
suggesting that the effect of education on knowledge is confounded. In a
more recent study, Rasmussen (2016) uses data from two cross-sectional sur-
veys conducted in Denmark and finds that education remains an important
predictor of political knowledge even after accounting for intelligence and
the Big Five personality traits.6 In a departure from Luskin (1990) and
Rasmussen (2016), both of whom attempted to include some possible con-
founders but were only able to use cross-sectional data, Highton (2009) uses
data from four waves of the Youth-Parent Socialization Study, a well-known
panel study in political science. He finds that
Differences in political knowledge associated with attending and graduating
from college that are apparent when people are in their 20s, 30s, and 50 are not
caused by attending college because the differences are evident when people
are 18 before any college education has taken place. Factual political knowledge
is caused by factors that are also associated with attending and graduating from
college. (Highton, 2009, p. 1570)
Given the different conclusions that have emerged across the studies out-
lined above, we...
To continue reading
Request your trial