You are on page 1of 24

Is Political Interest Absolute or Relative?

Andr Blais. University of Montreal andre.blais@umontreal.ca


Carolina Galais. University of Montreal. carolgalais@gmail.com
Shaun Bowler. University of California, Riverside shaunb@ucr.edu
APSA 2014 Annual Meeting Paper

Abstract
An individuals level of interest in politics has both normative and empirical relevance. Normatively,
interest in politics is a component of the definition of a democratic citizen; empirically, interest in
politics is a broadly used predictor of political participation. Despite this importance there has been
little systematic reflection about how it should be measured, and especially about how to minimize
social desirability biases that may prevent some people from admitting a lack of interest. In this
paper we present different ways of measuring political interest and validate these measures by using
them to predict standard measures of political participation. In this paper we determine whether
asking respondents how interested they are in politics compared to other domains produces different
results.
Introduction
An individuals level of interest in politics has both normative and empirical relevance within
political science. Normatively, interest in politics is a component of the definition of a democratic
citizen. Berelson, for example, wrote of the relationship between normative definitions of
democratic citizenry and the level of interest in politics in the following way:
the electorate is required to possess a certain degree of involvement in the process
of political decision, to take an appropriate share of responsibility. Political
democracy requires a fairly strong and fairly continuous level of interest from a
minority, and from a larger body of the citizenry a moderate-to-mild and
discontinuous interest but with a stable readiness to respond in critical political
situations. Political disinterest or apathy is not permitted, or at least not approved.
(Berelson 1952:316)
Under Berlesons definition, being interested in politics is not just part of the definition of being a
democratic citizen it is close to being a requirement for being one.
In empirical studies an individuals level of interest in politics is important, because it is often seen
as a driver of political participation. People participate, according to Verba et als succinct
summary, for three reasons; because they are able to, because they have been asked to or because
they want to (Verba et al. 1995, 271). The first two reasons have generated sizable literatures on the
organizational and institutional features of political participation. The efforts of parties to mobilize
1

voters and the efforts of politicians to remove/construct legal and institutional barriers have seen
substantial and extensive bodies of work documenting the importance of both in driving
participation. The resource model of participation adds a different kind of constraint on
participation, often grounded in socio-economic factors including access to group membership
(social capital) as well as more standard variables such as education or income (Verba et al 1995).
The third reason for participation people participate because they want to has received somewhat
less attention. In part this lack of attention is related to the difficulty in accounting for peoples
preferences: it is hard to explain why people want what they want. Even so, a number of studies have
examined the role of citizen interest in politics. It is clear that an individuals interest in politics is a
marker for that individual wanting to participate in politics, possibly for intrinsic gratification (Brady
et al. 1995, 271; Campbell et al. 1954, Luskin 1990, Van Deth 1990, Verba, Schlozman, and Brady
1995, Delli-Carpini and Keeter 1996, Lupia & McCubbins 1998, Althaus 2003, Prior 2008). Some
scholars express concern over the relatively low levels of interest, especially among youth (e.g.
Mann 1999), while others offer a more optimistic view (see e.g. Dalton, 1998:26). To some extent
this literature does not so much explain why citizens are (or are not) interested so much as document
the degree to which citizens are interested. These concerns clearly tie back to the normative
importance of citizen interest in politics.
Given the importance of an individuals interest in politics form both a normative-theoretical and
empirical standpoints then it is important to develop valid measures of an individuals interest. The
standard measure of political interest is a simple question that asks respondents to indicate their level
of interest for politics. Sometimes there is a battery of questions about interest in politics and in
elections, in national, local, and international politics. Sometimes, there is a battery of questions
about interest as such and related attitudes or behaviors, especially, the level of attention to politics
or the news. In this paper we address the question of whether or not the type of question used to
measure political interest affects the results, proposing two alternative relative measures of
political interest (ratings and rankings). We examine whether the type of question affects the overall
level of reported interest and the relationship with socio-demographic characteristics on the one hand
and political participation on the other hand.
Our findings indicate that presenting valid and worthy domains side by side with politics does not
have much impact on the reported level of political interest and on the relationships between
political interest and socio-demographic variables or political participation. Having information
about other domains of interest does contribute, however, to a broader picture of their mind set and
to a better explanation of their willingness to get involved in politics.
On the (classical) measure of political interest.
From the classical perspective, there are some people who, for whatever reason, like politics or feel
attracted or curious about it and others who do not. These tastes are remarkably stable and they
explain in good part why some devote much more time than others to political activity (Prior 2010).
Yet, there are numerous sources of potential bias and noise that affect reported interest in politics.

For instance, the question leaves to the consideration of the respondent what politics actually means
(Prior 2010).1
Political interest can be defined as a citizen's willingness to pay attention to political phenomena at
the possible expense of other topics (Lupia and Philpot 2005: 1122). Indeed, there are only twentyfour hours in a day, and we cannot afford to invest our energy in an unlimited set of domains. 2 Yet
the most frequent survey question about political interest (How interested are you in politics? ) is
construed in absolute terms. Moreover and perhaps because it is seen in absolute terms- this
classical, absolute question is loaded with social desirability bias.
Social desirability arises when a sincere response entails revealing an embarrassing weakness, not
in line with social norms. One example is the reluctance survey respondents demonstrate in
admitting they did not, in fact, vote (Weiss 1968, Sudman and Bradburn 1974). Reporting an
interest in politics is likely also subject to desirability bias. Citizens are likely to say they are
interested in politics because they are supposed to be interested. As van Deth notes As we all know
from personal experience based on conversations in train compartments and pubs, politics usually is
not considered a popular or interesting topic (van Deth:2000:121). Voters are not likely to be
interested in politics yet are inclined to respond otherwise. Plainly this bias matters given that a
respondents interest in politics is used within the discipline either as a marker for democratic health
or as a representation of voters wanting to participate in politics. Previous work, for example, has
shown that interest in politics is associated with over-reporting of a range of different behaviors
including voting (Presser 1984; Achen and Blais 2014). Similarly, there is consistent evidence that
people who respond to surveys about politics are also more interested in politics (Brehm 1993).
Purging the question of the measurement error due to biases, imprecision or social desirability would
require costly measures (i.e., repeating the survey over time or including several measures of
political interest within a survey) and even impossible counterfactual research methods, such as
interviewing the individuals that did not undertake the survey.
Van Deths work suggests us a way to find a way round some of these problems. Van Deth (2000)
distinguishes between subjective political interest, on the one hand, and relative political saliency (or
importance) on the other hand, especially in relation to competing pressures such as family or work.
Van Deth argues that seeing interest in politics in this relative way should lessen the effects of social
desirability, yielding lower and more realistic levels of political interest.
We agree with van Deths general approach. A relative measure seems to be a plausible and
reasonable way to provide a more accurate measure of interest in politics. But we do not see the
benchmark against which politics is measured in the World Value and the Eurobarometer surveys to
be entirely satisfactory. While we recognize along with van Deth that the standard measure is likely
1

Previous work has shown that this standard measure is sensitive to context and question order (Bishop et al. 1984 ;
Lasora 2009; Sturgis and Smith 2010).
2

There are two responses to this argument. The first is that it is possible to keep interest in politics (or anything else for
that matter) without being able to devote any time to it. After all, we should distinguish what a person likes (interest) and
what she does (participation). The second one is that it is not clear that time constraints are an important determinant of
political engagement (Putnam 2000), and it thus seems possible to maintain interest in many domains. This is, of course,
an empirical issue, and the question thus becomes whether people who are very interested in politics but are also very
interested in many other things are as active in politics as those who are also very interested in politics but are less
interested in other things. We address these questions in the paper.

to over-state the importance of politics the domains he uses as the reference point which include
family and work- are not the most appropriate. A fairer comparison should refer to other worthy
domains that compete for individuals attention in the public sphere. In this way, being interested in
politics is no longer the only way for respondents to pay lip service to the good citizen norm; they
can say that they are interested in lots of other worthy things. We can, then, reduce the social
desirability bias. Furthermore, interest in politics can be set against other leisure interests, such as
arts or sports, that one is free to pursue or not, rather than the more demanding competitors of family
or work. Finally, the concept of saliency does not reflect all affective and cognitive elements of
curiosity (Van Deth 1990), attention (Zaller 1992), willingness to pay attention (Lupia and Philpot
2005) or, broadly speaking, psychological engagement with politics (Teorell 2003, Augemberg
2008, Galais 2008) that are associated with the concept of interest. Hence, we think it better to
ask about interest in competing domains and not importance.
There are at least two ways in which relative interest for politics can be measured: through
comparing ratings of interest in various domains or through having respondents rank order these
various interests. Ratings are less time consuming and more enjoyable for respondents than
rankings, whereas rankings yield higher-quality data than ratings (Pasek and Krosnick 2010).
Whether a ranking produces more valid results than ratings has been the subject of some analyses,
especially when it comes to measuring values. Different studies come to different conclusions (see,
especially, Maio et al. 1996 and Alwin and Krosnick 1985). Ranking forces respondents to prioritize,
to indicate which interests are highest and which ones are lowest, and it eliminates ties (which may
be an easy way to avoid prioritizing) or non-differentiation (Krosnick 1999). Perhaps most
importantly, ranking neutralizes the social desirability bias that leads some people to say that they
are very interested in everything. Rating, however, allows people to say that they are equally
interested in two domains (which people may well be), and it allows them to express the intensity of
their preferences: one may be only slightly more interested in her first choice than in her second
while being a lot more interested in the second than in the third.
There are thus good arguments in support of using both ratings and rankings. We are agnostic about
which is to be preferred but we expect both to be more accurate indicators of interest in politics than
the standard measure.
Our goals are to:
1. Compare the level of interest in politics to the level of interest in other domains.
2. Determine whether socio-demographic characteristics better predict relative interest.
3. Determine whether the relative measures better predict political participation.

Methodology and data


The data we use are from two different countries and two different platforms. In a sense they do
represent 'convenience samples' of a particular kind: the French survey was being conducted
anyway and mturk is inexpensive. But there is an advantage in using such different samples and
obtaining broadly similar results, especially when it comes to a concept as widely used in mass
political behavior as political interest. We can have some confidence that the results we obtain are
4

not the result of quirks of the survey platform or an unusual sample. As we will see below there are
remarkable similarities across the two samples.
The first experiment was embedded in a MTURK online survey on a non-representative sample of
1500 U.S. citizens. The fieldwork was conducted on August 30 and 31, 2013. The second
experiment was embedded in an internet post-election survey conducted in the framework of the
Making Electoral Democracy Work (MEDW) project by Harris International between March 31 and
April 14, 2014 in the days following the second round of the French municipal elections, in the cities
of Paris and Marseilles. The total sample is 1285 (821 in Paris and 464 in Marseilles).3
Table 1 summarizes the experimental design and the question wordings. The baseline condition
consists in asking people to rate their interest in politics without any other element of comparison (an
absolute measure). For this, we simply asked respondents reported level of interest in politics, in
this case on a 0 to 10 scale. For relative interest (treatments 1 and 2), we adopt two approaches. First,
we include measures of interest for other domains and we determine whether including other
interests affects the results. Second, we ask respondents to rank order their interest for various
domains, and we examine whether rankings produce different results than ratings.
When interest in construed in relative terms, the question arises as to what other interests political
interest should be compared with. We must admit that there is an element of arbitrariness in
delimiting these other interests; we want to have a limited number of interests as we were asking
some of our respondents to rank order them. We think of politics as a public domain of activity and
we are looking for other public domains that can compete with politics for peoples attention. After
some intense discussion with colleagues, we came up with the following list of five domains:
politics, sports, religion, arts and culture, and science. 4
Hence, the second group was asked to answer the same question but as part of a battery of interest in
the five aforementioned domains (relative interest). The third group was asked to rank their interest
in the five domains by dragging the domains from top to bottom interest. The assignment of
individuals to these groups was randomized. One-third of the respondents were asked to rate their
interest in politics on a 0 to 10 scale, alone; one-third were asked exactly the same question but were
also invited to rate their level of interest in sports, arts and culture, religion, and science. Finally,
one-third were asked to rank order (from first to fifth) their interest for the same five domains. Note
that the five domains appeared simultaneously on the screen in the two treatments and that the order
was randomized. We have about 500 individuals in each of the three groups in the US and about 470
in France.
In order to test the effect of the alternative measures, we first present the distribution of the three
formulations (absolute, relative rating and relative ranking). We also pay close attention to our
relative measures, which allow us to understand how interest in politics relates to other interests.
Finally, a series of logit and OLS regressions test the effects of age, sex and education on our three
measures, as well as the effects of absolute and relative interest on electoral and non-electoral
participation.

These respondents were part of an initial sample of 2024 respondents who participated in a first wave before the first
round of the elections. The sample is representative of the population of both cities in terms of age, gender, and
education.
4

These domains pretty much coincide with Google news domains.

Findings
Table 2a and Table 2b present descriptive information about the responses obtained to the rating
questions, both absolute and relative. We can see that when measured alone, mean rating of interest
in politics is 6.2 in the U.S. and 6.0 in France; about half the respondents give scores between 5 and
8. When measured together with other interests, political interest declines slightly in the U.S., with a
mean of 5.7; the difference is that fewer people rate their interest in politics at 9 or 10. The
differences are minor in France, mean interest increasing slightly when measured in relative terms,
with more people choosing 8 and fewer 5. All in all, our prediction that reported interest in politics
would be lower when measured together with interest in other domains is mildly supported in the US
but disconfirmed in France.
Figure 1 represents an additional effort to compare the resulting average interest in politics for each
of these formulations. For this purpose, we have previously standardized the variables so as to rank
between 0 (minimum interest in politics) to 1 (maximum interest in politics). This required to invert
the values of rankings, hence considering politics the least interesting domain (ranked 5th) has been
coded as 0, whereas considering politics the most interesting thing (ranked 1st) has been coded as 1.
The resulting means with their respective 95% confidence intervals reveal an interesting pattern. The
ranking question produces the lowest levels of self-reported interest in both the US and France.
However, the rating question only behaves according to our expectations in the US, i.e. producing
significantly lower levels of interest than with the absolute formulation. We do not find significant
differences between the absolute and the rating question in France.
(Figure 1 above here)
Table 2 and Table 2b also show that in our samples mean interest is highest for science and arts in
the US, for arts and science in France. Religion and sports come at the bottom, with the highest
variance (due in part to the fact that a substantial minority of respondents are willing to give a 0),
while politics is in the middle. Note also that politics tends to get few extreme scores (0 and 10).
Finally, very low scores (0 to 2) are rare, except for sports and religion. There may be a social
desirability bias with respect to the other domains: the good citizen is supposed to have at least
some interest in politics, arts, and science.
(Table 2a and Table 2b about here)
What about rankings? Table 3a and Table 3b show that most respondents put politics at second, third
or fourth rank. The majority give arts and science the first and second ranks, while religion and
sports usually come fourth and fifth. The picture is similar to that obtained with ratings. Arts and
science are at the top, religion and sports at the bottom, and politics in the middle. Note also that
politics has the lowest proportion of first rank in the U.S. while it is religion in France.
(Table 3a and Table 3b about here)
It is also interesting to see how the other interests correlate with interest in politics (Table 4). The
ratings are all positively correlated with each other, which indicates either that some people are
interested in about everything and others in about nothing or that some people tend to give high
scores and others low (or not so high) scores. The pattern is more pronounced in France than in the
U.S. The weakest correlation is between interest in politics and interest in sports or religion in the
U.S. As for rankings, the correlations are necessarily negative, since giving a high rank to a domain
6

forces one to give a lower rank to other domains. There is one exception, which is the very low
negative correlation between interest in politics and interest in science in the U.S.
(Table 4 about here)
In a second step we can look at the relationship between these different measures of political interest
and socio-demographic characteristics (Table 5). We focus on three characteristics: age, gender, and
education. From here on, all the independent variables and controls (including interest itself) have
been standardized on a 0 to 1 scale to facilitate comparisons. 5 As the literature agrees on the
existence of persisting different political socialization patterns for men and women (Bennett and
Bennett 1989, Verba et al. 1997), we expect political interest to be higher among men. Also, as these
are well-known antecedents for political interest (Denny and Doyle 2008), we expect interest to
increase with age and education.

(Table 5 about here)


It can be seen that whether measured in solo or in combination with other domains, interest in
politics is indeed higher among men, the better educated and increases with age. The relationships
are the same whether interest is measured in absolute or relative terms, with ratings or rankings.
Note, however, that the relationships tend to be weaker with rankings. Whether ratings of interests
are asked in solo or with the other domains does not seem to matter much.
The very last step of our study consists in looking at the relationship between the various measures
of interest and political participation.6 The U.S. survey included a question on respondents electoral
behavior in the 2012 presidential election which was recoded as 1= voted vs. 0= did not vote. The
French survey asked for participation in the first round of the French municipal elections, on March
23, 2014.7 This question was the subject of another experiment, in which the treatment included
many reasons to abstain. Answers to the two versions of the question were coded so as the value 1
identifies those who are sure they voted against the rest (not sure, did not vote for whatever reason).8
Table 6a and Table 6b present the results of a series of logistic regressions relating electoral
participation to our different measures of interest, controlling for age, gender, and education.
Column 1 uses the absolute rating question and confirms that political interest enhances the
propensity to vote; everything else being equal, the probability of voting increases by 58 percentage
points in the U.S. (from 42% to 100%) and by 46 points in France (from 54% to 100%) when
political interest moves from 0 to 10. Column 2 presents the findings with political interest measured
in the multiple domain treatment: the results are similar, though somewhat weaker. In this case,
moving from minimum to maximum political interest caused an increase in the probability of voting
of 48 percentage points in USA and 40 percentage points in France. Column 3 replicates the former

For rankings, lowest rankings pointing to low interest have been coded as 0 and highest rankings as 1.

We should note that questions about interest and participation were separated. Interest questions came at the very
beginning of the questionnaire and those about participation came at the very end.
7

We could have chosen the question on turnout on the second round of the municipal elections but we did not because
some respondents did not have second round in their district (the candidate was elected in the first round).
8

We refer the reader to Appendix III for a detailed description of all the variables employed in our analyses.

analysis but now including interest in the other domains. The coefficient associated with political
interest is unaffected. Interestingly we also see that in the U.S. those who are interested in arts are,
everything else being equal, less prone to vote. In France, we find a similar negative relationship, but
this time with sports, an interest that is detrimental to electoral participation.
(Table 6a and Table 6b about here)
Column 4 is based on ranking of interest in politics. The reference category is those who give the
lowest rank to politics. We can see that those who put politics last are less inclined to vote. But there
seems little evidence of finer grained distinctions. Whether one ranks politics third or fourth does not
seem to matter much. This suggests that only a minimum degree of relative interest is needed to get
someone to vote. This non-linear relationship does not emerge with ratings.
Columns 5 to 7 use rankings for all domains except sports (the reference category). Column 5 is
based on the ranking (from 1 to 5, then reversed, finally standardized to range from 0 to 1) given to
the various domains. With that operationalization none of the interest variables is significant in the
U.S. and only interest in politics is in France. Columns 6 and 7 use dummies for those who put a
particular domain either at the top or at the bottom. The only significant finding is that those who
rank politics last are slightly less inclined to vote. In France, placing politics at the top of ones set of
interests also seems to predict higher chances of voting.
With respect to electoral participation, then, the impact of political interest depends on how it is
measured. When it is measured with ratings, either in solo or in combination with other domains, its
effect is substantial. The multiple domain approach tells us, however, that it is not only interest in
politics that matters, that in the U.S. at least interest in arts is equally important, and that it is
associated with lower turnout. As for the ranking approach, the data suggest that whether politics is
ones priority or not does not matter that much, except perhaps that those who rank politics last are
more tempted to abstain.
As for non-electoral participation, the Mturk survey included questions on whether, in the last 12
months, respondents had expressed their views on an issue by contacting a newspaper or a politician,
participated in a demonstration or march, volunteered for a political party, signed a petition, or
boycotted a product for ethical reasons in the last twelve months. After recoding and adding these
indicators, we obtained a non-electoral participation scale ranging from 0 to 5; then, we rescale it so
as to range from 0 to 1. For its part, the French survey inquired whether the responded had taken part
in demonstrations, bought or boycotted products for political, ethical or environmental reasons,
signed a petition or contacted a public servant during the previous twelve months. After adding the
positive answers to these questions, we obtained a scale with values between 0 (did not perform any
of these activities) to 4 (did perform all these activities). The scale has been normalized so as to
range from 0 to 1 in order to facilitate comparisons.
We estimated those non-electoral participation scales by means of a series of OLS regressions (see
Tables 7a and 7b). In the American case, the dependent variable is a five category scale accounting
for the amount of non-electoral participation acts that the individual had performed in the last 12
months. The scale has later been recoded so as to rank between 0 (no activities undertaken during the
last year) to 1 (the individual has performed all five activities). The first three columns are based on
ratings, and the results are similar whether political interest is measured in solo or not. In all cases,
political interest is a strong predictor of participation. Interestingly, the multiple domain approach
tells us that political participation is also fostered by an interest in arts, which is negatively
8

correlated with turnout. It seems that the artistically inclined are more attracted to new forms of
political participation. Finally, interest in sports has a negative effect. The French data confirm these
findings, as well as the fact that including individuals interest in all domains increases model fit;
this is, helps us to get a better understanding of the phenomenon at stake.
(Table 7a about here)
The findings are relatively similar when we use rankings. In columns 4 to 7, the level of interest in
politics remains a strong predictor of participation. What emerges, however, is that what is required
is that one does not rank politics fourth or fifth in the US, and that it should be first or second in
France. Furthermore, the fact that all the coefficients associated with the five domains (with the
exception of religion, in France) are positive in column 5 indicates that a strong interest for sports
leads to lower political engagement.

Conclusions
As we noted at the outset, individual level interest in politics has long been seen as an important
driver of political participation. But, as we have shown, just how important depends in part on how
we measure political interest. Our findings indicate that presenting interest in politics along with
other valid and worthy domains does not necessarily yield lower levels of reported political interest.
Interest in politics tends to be lower than interest in arts and science and somewhat higher than
interest in religion and sports, the only two domains for which a substantial minority of respondents
in both France and the U.S. report very low interest. With all three measures we find political
interest to increase with age and education and to be slightly higher among men. Ratings of interest
in politics are similarly related to both electoral and non-electoral participation, whether they are
expressed alone or together with other domains. Ranking of interest in politics also tends to produce
similar findings, except in the case of electoral participation in the U.S., where the relationships are
particularly weak. All in all, then, relative measures of political interest do not produce very different
results.
Nevertheless, the use of relative measures does contribute to a better understanding of the causes of
political participation. The relative measures tell us that those who are very interested in sports are
less inclined to get involved in politics. They point to yet another culprit for the relatively low level
of political participation: the popularity of sports. The literature on social capital would seem to
suggest interest in non-political issues should be positively correlated with participation in politics
but our results are not consistent with that suggestion so far as sports are concerned.
Indeed, our findings show that very few respondents are willing to confess a very low level of
interest for arts, science, and politics. These are worthy domains that the good citizen should say
she has at least a modest degree of interest in. The most efficient way to neutralize this bias is to
force respondents to rank their domains of interest. When we do so, most of the relationships that we
observe with the standard solo rating indicator remain but they tend to be weaker. This suggests that
existing research may overstate the true impact of interest on political participation, and also that the
use of relative measures helps us to better understand and correct the social desirability bias
associated with the measurement of peoples interest for politics. Finally, our formulation opens the
door for future research on whether some people (e.g. women, thus allowing to address the gender
9

gap on political interest and political knowledge) wont keep up an interest in politics simply
because they are passionate about other things.

References
Althaus, S. L. (2003). Collective preferences in democratic politics: Opinion surveys and the will of
the people. Cambridge University Press.
Augemberg, K. (2008). Values and politics: Value priorities as predictors of psychological and
instrumental political engagement. ProQuest.
Bennett, L. L., & Bennett, S. E. (1989). Enduring Gender Differences in Political Interest The
Impact of Socialization and Political Dispositions. American Politics Research, 17(1), 105-122.
Bishop, G. F., Oldendick, R. W., & Tuchfarber, A. (1984). What Must My Interest in Politics Be If
I Just Told You I Don't Know?. Public Opinion Quarterly, 48(2), 510-519.
Brady, H. E., Verba, S., & Schlozman, K. L. (1995). Beyond SES: A resource model of political
participation. American Political Science Review, 271-294.
Brehm, J. (1993) The Phantom Respondents: Opinion Surveys and Political Representation.
University of Michigan Press, Ann Arbor, MI.
Campbell, A., Gurin, G., & Miller, W. E. (1954). The voter decides. Oxford; Row, Peterson, and Co.
Carpini, M. X. D. (1996). What Americans know about politics and why it matters. Yale University
Press.
Denny, K., & Doyle, O. (2008). Political interest, cognitive ability and personality: Determinants of
voter turnout in Britain. British Journal of Political Science, 38(02), 291-310.
Galais, C. (2008). Socializacin o contexto? La implicacin poltica subjetiva de los espaoles
(1985-2006) (Doctoral dissertation, Tesis doctoral defendida el 8 de noviembre de 2008,
Departament de Cincies Politiques i Socials, Universitat Pompeu Fabra, Barcelona).
Jones, S. G. (1992). Sport, Politics and the Working Class: Organised Labour and Sport in Interwar Britain. Manchester University Press.
Krosnick, J. A. (1999).Survey research. Annual review of psychology, 50(1), 537-567.
Lasorsa, D. L. (2009). Political Interest, Political Knowledge, and Evaluations of Political News
Sources: Their Interplay in Producing Context Effects. Journalism & Mass Communication
Quarterly, 86(3), 533-544.

10

Lasorsa, D. L. (2009). Political Interest, Political Knowledge, and Evaluations of Political News
Sources: Their Interplay in Producing Context Effects. Journalism & Mass Communication
Quarterly, 86(3), 533-544.
Lupia, A. and Philpot, T. S. (2005), Views from Inside the Net: How Websites Affect Young Adults
Political Interest. Journal of Politics, 67: 11221142. doi: 10.1111/j.1468-2508.2005.00353.x
Lupia, A., & McCubbins, M. D. (1998). Political Credibility and Economic Reform. A Report for
the World Bank.
Luskin, R. C. (1990). Explaining political sophistication. Political Behavior,12(4), 331-361.
Pasek, J., & Krosnick, J. A. (2010). Optimizing survey questionnaire design in political science:
insights from psychology. Oxford handbook of American elections and political behavior, 2750.
Presser, S. (1984).Is inaccuracy on factual survey items item-specific or respondentspecific?. Public Opinion Quarterly, 48(1B), 344-355.
Presser, S. (1990) Can changes in context reduce vote overreporting in surveys? Public Opinion
Quarterly, 54 (4) (1990), pp. 586593
Prior, M. (2008). The stability of political interest over the life cycle. Presented in the Annual
Meeting of the Midwest Political Science Association, Chicago (IL)(en lnea). http://www.
princeton. edu/csdp/events/Prior052208/Prior052208. pdf, ltimo acceso (Vol. 1).
Prior, M. (2010). Youve either got it or you dont? The stability of political interest over the life
cycle. The Journal of Politics, 72(03), 747-766.
Sturgis, P., & Smith, P. (2010).Fictitious issues revisited: Political interest, knowledge and the
generation of nonattitudes. Political Studies, 58(1), 66-84.
Sudman, S. and Bradburn, N. M. (1974) Response effects in surveys: a review and synthesis.
Number 16 in National Opinion Research Center Monographs in Social Science. Aldine, Chicago,
IL.
Teitel, R. (1992). Critique of Religion as Politics in the Public Sphere. Cornell L. Rev., 78, 747.
Teorell, J. (2003). Linking Social Capital to Political Participation: Voluntary Associations and
Networks of Recruitment in Sweden1. Scandinavian Political Studies, 26(1), 49-66.
Van Deth, J. W. (2000). Interesting but irrelevant: Social capital and the saliency of politics in
Western Europe. European Journal of Political Research,37(2), 115-147.
Van Deth, Jan W. (1990): Interest in Politics, in Continuities in Political Action, ed. M. Kent
Jennings, Jan W. van Deth et ai, Berln: Walter de Gruyter.
Verba, S., Burns, N., & Schlozman, K. L. (1997). Knowing and caring about politics: Gender and
political engagement. The Journal of Politics, 59(04), 1051-1072.

11

Verba, S.; Schlozman, K. L. & Brady, H. E. (1995). Voice and equality: Civic voluntarism in
American politics. Harvard University Press.
Weiss, C.H. (1968) Validity of welfare mothers' interview responses. Public Opinion Quarterly, 32 ,
pp. 622633

12

Figure 1 : Adjusted predictions of interest in politics (standardized) with 95% CIs


France

.4

.4

.45

.45

Linear Prediction
.5
.55

Linear Prediction
.5
.55

.6

.6

.65

.65

USA

absolute

rating

ranking

absolute

rating

ranking

13

Table 1: Experimental design.


Group

Control

Treatment 1

Treatment 2

N
USA

482

504

509

N
Variable
France

449

475

468

Absolute political
interest

Relative political
interest. Rating.
Battery of 5 items

Relative political
interest .Ranking.

Question
On a scale from 0 to 10,
where 0 means that you are
not interested at all and 10
means that you are very
interested, how interested
are you personally in
politics?
On a scale from 0 to 10,
where 0 means that you are
not interested at all and 10
means that you are very
interested, how interested
are you personally in the
following:
Politics
Sports
Arts and Culture
Religion
Science
Please order the five
domains below according
to how interested you are,
such that the one in which
you are the most interested
is on top. Rank the domains
by dragging them to the list
on the right.
Politics
Sports
Arts and Culture
Religion
Science

Response options
( ) "Not interested at all"
0, ( ) 1, ( ) 2, ( ) 3, ( ) 4, (
) 5, ( ) 6, ( ) 7, ( ) 8, ( )
9, ( ) "Very interested"
10
France included a DK
option. N=3 (.7%)

( ) "Not interested at all"


0, ( ) 1, ( ) 2, ( ) 3, ( ) 4, (
) 5, ( ) 6, ( ) 7, ( ) 8, ( )
9, ( ) "Very interested"
10
France included a DK
option. N=7 (1.5%)

1___________
2___________
3___________
4___________
5___________
France included a DK
option. N=7 (1.5%)

14

Table 2a. Distribution of Interest in politics: ratings. United States.

0
1
2
3
4
5
6
7
8
9
10
Total N
Mean
Std. Dev.

%
Absolute
interest in
politics:
rating
2.7
3.1
5.2
5.8
5.8
12.5
13.1
17.6
14.9
11
8.3
482
6.2
2.6

% Relative interest, rating:


Politics

Arts &
culture

Religion

Science

Sports

4.2
3.6
6.5
8.9
8.1
11.1
12.5
19.8
12.1
7.5
5.6
504
5.7
2.6

.4
1.8
3.4
3.2
6.7
12.7
10.1
19.4
18.6
10.1
13.5
504
6.8
2.3

18.6
8.7
10.1
9.3
4.8
8.7
6.3
8.7
9.3
4.8
10.5
504
4.4
3.4

1.2
1
2.4
2.8
4.2
9.9
11.9
15.1
18.8
13.9
18.8
504
7.2
2.3

11.5
10.5
10.7
8.1
6.8
7.7
9.9
8.1
7.9
6.2
12.5
504
4.8
3.3

Table 2b. Distribution of Interest in politics: ratings. France.

0
1
2
3
4
5
6
7
8
9
10
Total N
Mean*
Std.
Dev.

%
Absolute
interest in
politics:
rating
4.0
3.1
4.5
7.4
6.5
15.3
11.0
17.5
12.1
7.2
11.4
446
6.0
2.7

% Relative interest, rating:


Politics

Arts &
culture

Religion

Science

Sports

5.8
3.0
3.6
5.6
5.6
10.9
10.3
17.3
18.42
8.35
11.13

1.7
.4
3.0
3.0
2.6
8.4
12.0
19.5
19.9
12.9
16.7

22.1
8.8
9.6
5.6
5.8
13.1
12.4
6.6
5.6
4.3
6.2

1.7
1.5
2.6
4.1
4.3
12.9
12.9
16.7
16.3
12.5
14.6

14.1
7.9
8.3
4.7
6.2
12.0
9.0
11.1
13.7
5.34
7.69

467
6.2

467
7.2

467
3.9

466
6.8

468
4.9

2.8

2.3

3.2

2.4

3.2

15

Table 3a. Distribution of Interest in politics: ranking. United States

1st
2nd
3rd
4th
5th
Total N
Mean
Std. Dev.

Politics

Arts &
culture

Religion

Science

Sports

9.8
18.3
28.1
29.3
14.5
502
3.2
1.2

31.7
28.9
21.7
12.2
5.6
502
2.3
1.2

12.4
8.8
11.9
25
42
502
3.8
1.4

30.7
32.7
20.3
11.3
5
502
2.3
1.2

15.5
11.3
17.9
22.3
32.9
502
3.5
1.4

Table 3b. Distribution of Interest in politics: ranking. France.

Politics
1st
2nd
3rd
4th
5th
Total N
Mean*
Std. Dev.

20
22.4
21.7
21.7
14.1
460
2.88
1.34

Arts &
culture
39.6
26.5
17.4
12
4.6
460
2.15
1.19

Religion

Science

Sports

7.6
6.1
12.6
28.8
44.9
459
3.97
1.22

19.4
28.9
30.5
15.5
15.7
459
2.6
1.13

13.7
15.9
17.7
22
30.7
159
3.4
1.4

Table 4. Correlations between other interests and interest in politics

Politics,
rating
Arts and culture
Religion
Science
Sports
*

USA
politics,
ranking

France
Politics,
politics,
rating
ranking

.19*

-.29*

.3*

-.21*

.11*

-.26*

.32*

-.29*

.22*

-.09*

.32*

-.3*

.1*

-.26*

.24*

-.3*

p < .05

All measures have been standardized so to range between 0 (minimum interest in politics) to 1
(maximum interest in politics).

16

Table 5. Interest as a dependent variable. Ratings OLS regressions


USA

Female
Age
Education
Constant
R2
N
+

France

Absolute
interest in
politics:
rating

Politics,
ratings

Politics,
ranking

Absolute
interest in
politics:
rating

Politics,
ratings

Politics,
ranking

-.08**
(.02)
.17**
(.05)
.14*
(.06)
.54**
(.04)
.06
482

-.07*
(.02)
.13*
(.05)
.22**
(.06)
.44**
(.03)
.06
504

-.06*
(.03)
.03
(.05)
.2*
(.07)
.36**
(.04)
.03
502

-.09**
(.02)
.24**
(.07)
.24**
(.06)
.38**
(.06)
.08
446

-.11**
(.02)
.17*
(.07)
.42**
(.06)
.32**
(.06)
.14
467

-.07*
(.03)
.31**
(.08)
.25**
(.08)
.28**
(.07)
.06
460

p < .10, * p < .05, ** p < .01

All variables have been recoded so that values range between 0 and 1. The ranking variable is the result of
reversing the original rankings; hence, the maximum interest in politics (when it has been ranked as the 1st
interest) takes here the value1. When politics is ranked last, it takes the value 0 .

17

Table 6a. Interest as a predictor of electoral participation. USA (Logistic regression)


1
2
3
4
5
6
7
Absolute Relative Relative Political Ranking
Top
Bottom
political political interests, interest,
of
Ranking Ranking
interest, interest,
ratings
ranking. interests
rating
rating
Int. in politics

2.61**
(.46)

1.82**
(.4)

Int. in religion
Int. arts & cult.
Int. in science
Int. in sports

2.19**
(.45)
.20
(.33)
-2.00**
(.55)
.16
(.52)
-.01
(.34)

Politics: 2nd
Politics: 3rd
Politics: 4th
Politics: 5th
(ref.)

Age
Education
Constant
Pseudo R2
N
+

.72
(.44)
.19
(.39)
.31
(.33)
.29
(.30)

-.54+
(.31)
.05
(.24)
-.30
(.45)
.03
(.50)

(ref.)

(ref.)

(ref.)

.28
(.23)
1.24**
(.45)
2.37**
(.55)
-1.38*
(.67)
.06
487

.19
(.23)
1.26**
(.45)
2.40**
(.55)
-1.15*
(.38)
.06
487

.26
(.21)
1.33**
(.45)
2.36**
(.55)
-.82*
(.36)
.06
487

.92*
(.45)
.40
(.34)
.53+
(.32)
.53+
(.31)

Politics: 1st

Female

.62
(.40)
.13
(.37)
.05
(.42)
.21
(.40)

(ref.)
.15
(.23)
2.4**
(.55)
1.36*
(.64)
-1.81**
(.44)
.14
471

.27
(.22)
.44
(.45)
2.12**
(.58)
-1.32**
(.38)
.08
491

.41+
(.25)
.37
(.45)
2.33**
(.60)
-.50
(.55)
.11
491

.29
(.21)
1.32**
(.45)
2.33**
(.55)
-1.36**
(.39)
.07
487

p < .10, * p < .05, ** p < .01

All the variables in the model have been recoded so that values range between 0 and 1
Interest is measured through a series of dichotomous variables in model 4 that takes the value 1 when politics
has been ranked as 1st, 2nd, etc. interest.
Interest in model 5 is measured with the ranking question, reversed and standardized; which produces a series
of variables that range from 0 (least interesting) to 1 (most interesting). Model 6 includes a series of
dichotomous variables that capture which domain was selected as the most interesting. Model 7 measures the
effect of selecting each of these five domains as the least interesting.

18

Table 6b. Interest as a predictor of electoral participation (2014 municipal elections. Paris, Marseille).
France (logistic regression)

Interest in politics

1
Absolute
political
interest,
rating
2.39**
(.48)

2
Relative
political
interest,
rating
1.79**
(.45)

Religion
Arts and culture
Science
Sports

3
Relative
interests,
rating
2.17**
(.52)
-.57
(.45)
-.26
(.60)
.66
(.58)
-.77+
(.45)

Politics: 1st

Politics: 3rd
Politics: 4th

Age
Education
Constant
Pseudo R2
Obs.

5
Ranking
of
interests

6
Top
ranking

7
Bottom
ranking

.94*
(.45)
-.22
(.46)
.66
(.46)
-.02
(.49)
(ref.)

.71+
(.41)
-.55
(.45)
.57
(.35)
.02
(.38)
(ref.)

-.91**
(.34)
.20
(.28)
.25
(.61)
.45
(.59)
(ref.)

-.29
(.26)
1.76*
(0.70)
0.60
(0.61)
-.43
(.79)
.05
459

-.31
(.25)
1.89**
(0.66)
0.77
(0.59)
-.02
(.57)
.054
468

-.15
(.24)
2.01**
(0.67)
0.57
(0.60)
.32
(.63)
.058
468

1.30**
(.39)
.68*
(.34)
1.60**
(.41)
.92**
(.35)
(ref.)

Politics: 2nd

Politics: 5th
(ref.)
Female

4
Political
interest,
ranking

.34
(.26)
2.22**
(.73)
.96
(.64)
-1.55*
(.63)
.096
446

-.20
(.25)
2.11**
(0.72)
0.85
(0.65)
-.84
(.62)
.075
467

-.27
(.28)
1.71*
(0.76)
0.64
(0.70)
-.36
(.73)
.087
464

-.17
(.24)
1.73**
(0.67)
0.54
(0.61)
-.46
(.57)
.07
468

Standard errors in parentheses


+
p < .10, * p < .05, ** p < .01

19

Table 7a. Interest as a predictor of non-electoral participation. US. (OLS regression)

Int. in politics

1
Absolute
political
interest,
rating
(T1)
.31**
(.04)

2
Relative
political
interest,
rating
(T2)
.34**
(.04)

Int. in religion
Int. arts &
cult.
Int.t in science
Int. in sports

3
Relative
interests,
ratings

4
Political
interest,
ranking.

5
Ranking
of
interests

6
Top
interest

7
Bottom
interest

(T2)
.3**
(.04)
-.03
(.03)
.20**
(.04)
.07
(.04)
-.08**
(.03)

(T3)

(T3)
.25**
(.04)
.09*
(.04)
.09+
(.04)
.11*
(.04)

(T3)
.12*
(.04)
.02
(.04)
.04
(.04)
.07
(.03)

(T3)
-.14**
(.03)
-.07**
(.03)
-.08
(.05)
-.15**
(.05)

(ref.)

(ref.)

(ref.)

-.02
(.02)
-.04
(.04)
.01
(..05)
-.01
(.07)
.08
502

-.02
(.02)
-.04
(.04)
.03
(.05)
-.20
(.04)
.02
502

-.02
(.02)
-.02
(.04)
.04
(.05)
.31**
(.04)
.05
502

.16**
(.04)
.16**
(.04)
.10**
(.03)
.03
(.03)

Politics: 1st
Politics: 2nd
Politics: 3rd
Politics: 4th
Politics: 5th
(ref.)
Female
Age
Education
Constant
R2
N
+

(ref.)
.05*
(.02)
-.03
(.04)
.08
(.05)
-.02
(.04)
.12
481

.02
(.02)
-.09*
(.04)
.05
(.05)
.03
(.03)
.15
504

-.00
(.02)
-.09*
(.04)
.04
(.05)
-.07
(.05)
.21
504

-.00
(.02)
-.05
(.04)
.01
(.05)
.18**
(.04)
.06
502

p < .10, * p < .05, ** p < .01

All the variables in the model have been recoded so that values range between 0 and 1
T1, T2 and T3 refers to the original wording and treatment group
Interest is measured through a series of dichotomous variables in model 4 that take value 1 when politics has
been ranked as 1st, 2nd, etc. interests.
Interests in model 5 is measured with the ranking question, reversed and standardized; which produces a
series of variables that range from 0 (least interesting) to 1 (most interesting). Model 6 includes a series of
dichotomous variables that capture which domain was selected as the most interesting one. Model 7 measures
the effect of selecting each of these five domains as the least interesting.

20

Table 7b. Interest as a predictor of non-electoral participation. France. OLS regression

Interest in politics

1
Absolute
political
interest,
rating
.29**
(.06)

2
Relative
political
interest,
rating
.19**
(.05)

Religion
Arts and culture
Science
Sports

3
Relative
interest,
rating
.20**
(.06)
-.05
(.04)
.17**
(.06)
-.02
(.06)
-.08+
(.04)

Politics: 1st

Politics: 3rd
Politics: 4th
Politics: 5th

Age
Education
Constant
R-Squared
Obs.

5
Ranking
of
interests

6
top
ranking

7
bottom
ranking

.18**
(.05)
-.00
(.06)
.15**
(.06)
.09
(.06)
.18**

.06
(.05)
.03
(.06)
.09*
(.04)
.04
(.05)
(ref.)

-.09+
(.05)
.03
(.03)
-.06
(.07)
-.15*
(.06)
(ref.)

-.05
(.03)
.05
(.08)
.17*
(.07)
-.08
(.09)
.075
440

-.05+
(.03)
.11
(.08)
.23**
(.07)
.05
(.07)
.047
449

-.04
(.03)
.10
(.08)
.22**
(.07)
.12
(.08)
.065
449

.09*
(.05)
.20**
(.05)
.06
(.05)
.05
(.05)
(ref.)

Politics: 2nd

(ref.)
Female

4
Political
interest,
ranking

-.04
(.03)
-.03
(.08)
.14+
(.08)
.06
(.08)
.079
427

-.04
(.03)
.07
(.07)
.23**
(.07)
.01
(.07)
.089
452

-.06*
(.03)
.00
(.08)
.17*
(.07)
.03
(.08)
.115
450

-.03
(.03)
.11
(.08)
.19**
(.07)
.04
(.07)
.083
449

Standard errors in parentheses


+
p < .10, * p < .05, ** p < .01

21

Appendix I: Dependent variables and covariates. USA.


1.

In the past 12 months, have you:


Yes

No

Expressed your views on an issue by contacting a newspaper or a politician

()

()

Participated in a demonstration or march

()

()

Volunteered for a political party

()

()

Signed a petition

()

()

Boycotted a product for ethical reasons

()

()

2.

In talking to people about the last presidential election in November 2012, we find a lot of people were not able to
vote because they weren't registred, they were sick, or they just didn't have time.
How about you, did you vote in the last presidential election in November 2012?
( ) I did not vote
( ) I usually vote, but didn't this time
( ) I definitely voted
( ) I did not have the right to vote

3.

What is the last grade that you completed in school?


( ) Elementary school
( ) Some high school
( ) High school graduate
( ) Some college, no degree
( ) Vocational training/2-year college
( ) 4-year college/bachelor's degree
( ) Some postgraduate work, no degree
( ) 2 or 3 years' postgraduate work/master's degree
( ) Doctoral/law degree

4.

Are you a male or a female?


( ) Male
( ) Female

5.

Which category corresponds with your age?


( ) 18-25
( ) 26-35
( ) 36-45
( ) 46-55
( ) 56-65
( ) 65 and older

22

Appendix II: Dependent variables and covariates. France.


1.

In the past 12 months, have you:

Contacted a public servant?


Participated in a demonstration?
Signed a petition?
Bought or boycotted a product for political, ethical or environmental reasons?
2.

Yes
()
()
()
()

No
()
()
()
()

The question on the vote at past municipal experiment was the subject of an experiment. The two treatments have
been combined to produce a single question. Only I am sure I voted in the election and yes have been coded as
the value 1 (voted) versus the rest.

2A) In each election, some choose not to vote, others are unable to do so. Which of the following best describes your
situation during the first round of the past election?
() I have not voted at this election
() I wanted to vote but did not go vote
() I usually vote but did not go this time
() I'm sure I voted in the election
() Do not know / prefer not to answer
2B) In each election, many people are not able to vote because they were not registered to vote, they were sick or they do
not have time. Have you been able to vote in the first round of the past election?
() yes
() no
() Do not know/ prefer not to answer.
Covariates
3.

What is the highest level of studies that you have completed?


() None
() Primary school
() Ancien Brevet, BEPC
() CAP, BEP
() Vocational training
() High School leading to university
() 2-year college
() 4-year college/bachelor's degree / Master/ doctoral degree

4.

Gender
( ) Male
( ) Female

5.

Age
( ) 18-96

23

Appendix III: Summary of descriptive.


Observations

Mean

Std.Dev.

Min

Max

Variable

USA

France

USA

France

USA

France

USA

France

USA

France

Turnout

1456

1933

.71

.57

.45

.49

1494

1336

.24

.3

.24

.3

1495

1933

.5

.54

.5

.50

Education

1495

1933

.54

.72

.2

.20

Age

1495

1933

.26

.35

.25

.20

Non-electoral
participation
Female

24

You might also like