How Political Is the Personal? Gender Differences in the Level and the Structure of Political Knowledge

Transcription

1 This article was downloaded by: [Tilburg University] On: 20 July 2011, At: 04:45 Publisher: Routledge Informa Ltd Registered in England and Wales Registered Number: Registered office: Mortimer House, Mortimer Street, London W1T 3JH, UK Journal of Women, Politics & Policy Publication details, including instructions for authors and subscription information: How Political Is the Personal? Gender Differences in the Level and the Structure of Political Knowledge Marc Hooghe a, Ellen Quintelier a & Tim Reeskens a a Catholic University of Leuven Available online: 10 Oct 2008 To cite this article: Marc Hooghe, Ellen Quintelier & Tim Reeskens (2006): How Political Is the Personal? Gender Differences in the Level and the Structure of Political Knowledge, Journal of Women, Politics & Policy, 28:2, To link to this article: PLEASE SCROLL DOWN FOR ARTICLE Full terms and conditions of use: This article may be used for research, teaching and private study purposes. Any substantial or systematic reproduction, re-distribution, re-selling, loan, sub-licensing, systematic supply or distribution in any form to anyone is expressly forbidden. The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae and drug doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material.

2 RESEARCH NOTE How Political Is the Personal? Gender Differences in the Level and the Structure of Political Knowledge Marc Hooghe, Catholic University of Leuven Ellen Quintelier, Catholic University of Leuven Tim Reeskens, Catholic University of Leuven ABSTRACT. In this research note we report on the results of an experimental study among 469 respondents in Belgium. The study shows that the familiar gender differences in political knowledge can be substantially reduced when including more questions on female politicians. Using confirmatory factor analysis and item response theory measurements, we can observe that the structure of political knowledge is identical for women and men implying that women and men do not have different conceptions of what constitutes the political domain. Both for women and men, personal items (in this case: The correct name of the new baby of the female vice-prime minister) were shown to load strongly and one dimensionally with the more traditional institutional Journal of Women, Politics & Policy, Vol. 28(2) 2006 Available online at by The Haworth Press, Inc. All rights reserved. doi: /j501v28n02_05 115

3 116 JOURNAL OF WOMEN, POLITICS & POLICY political knowledge questions. It seems that women and men respond in the same manner to information about the personal lives of politicians. doi: /j501v28n02_05 [Article copies available for a fee from The Haworth Document Delivery Service: HAWORTH. address: <docdelivery@haworthpress.com> Website: < by The Haworth Press, Inc. All rights reserved.] KEYWORDS. Political knowledge, gender differences, Belgium, scale reliability, public/private In survey research on political knowledge, women typically obtain lower scores on political knowledge than men (Delli, Carpini, and Keeter 1993, 1996; Verba, Burns, and Schlozman 1997). Only for topics in which women are more directly involved, such as women s rights, health care, abortion, and local issues do women clearly outperform male respondents (Niemi and Junn 1998). The knowledge gap between men and women can be seen as a stable finding in political science research. Within the field of gender studies, two main objections have been raised against the way mainstream political science investigates political knowledge. A first argument is that most of the political knowledge questions that are routinely used refer to male politicians. A majority of leading political positions are filled by men, but still this might introduce a bias against female respondents. It has been argued that women will find it easier to identify with female politicians (Carroll 2003; Githens, Norris, and Lovenduski 1994). Successful female politicians might act as role models, heightening political interest among women (Campbell and Wolbrecht 2006; Wolbrecht and Campbell 2005). It is also expected that female politicians will devote more attention to topics and policy issues that are highly relevant for women, or that they will better represent women s interests (Lovenduski and Norris 1993). The fact that most political knowledge scales focus on top officials (President, Vice President, Chief Justice, etc.), offices that thus far have not yet been taken by women, would thus lead to an undue advantage to male respondents (Frazer and Macdonald 2003). This hypothesis is confirmed by the finding of Verba, Burns, and Schlozman (1997) that gender differences in political knowledge are smaller in states with at least one female Senator: Regression analyses controlling for respondents education, income, employment status, marital status, and age confirm this result with a winning female senate candidate on the ballot, there is a statistically significant increase in political knowledge and interest for women, but not for men (Verba, Burns, and Schlozman

4 Research Note , 1,069). One obvious manner to reduce the gender difference in political knowledge scores therefore would be to include more questions on female politicians. Second, it has been argued that women and men differ with regard to the way they conceptualize politics. Verba, Burns, and Schlozman (1997) claim that, women and men have different political interests, and as a result, also develop different forms of political knowledge. This argument holds that women are more interested in, and they know more about, topics and policy areas that traditionally have been defined as female (education, family...). In the seminal study by Niemi and Junn (1988), women were clearly more knowledgeable than men about topics such as civil liberties, freedom of religion, and some historical figures. These findings seem to support the point of view of Lovenduski (1986, 119), when she claims: In short, men and women do differ politically, but in neither the manner, nor for the reasons suggested by traditional studies. Since traditional instruments for measuring political knowledge put a strong emphasis on institutional aspects of political knowledge, it could be argued that they measure a male-defined form of politics, introducing a bias against women when answering these questions. Therefore, including more traditionally female topics might reduce the observed gender gap in political knowledge scores. In this research note, we report on an experimental study assessing whether both arguments can be supported empirically. First, we test whether women indeed score better when female politicians are systematically included in the tests of political knowledge. Second, we assess whether the structure of the responses indicates that women and men tend to use different concepts of what politics actually is. The study was conducted among 469 Belgian first-year university students, in their first weeks at the university. We assume that these students have not yet been exposed to any substantial influence from the university environment. Participation in the test was a course requirement, so that in practice more than 90 percent of all enrolled students from the Faculty of Social Sciences at the Catholic University of Leuven actually participated in the experiment. It should be noted that all Belgian universities have open admission policies, so there has not been any prior selection of the respondents in this study. The sample includes 266 women and 203 men. Self-evidently, results from this experiment cannot be used for generalization toward the general population. But since our main research question is to ascertain whether women and men delineate the political sphere in a different manner, this kind of experiment offers a valid research design.

5 118 JOURNAL OF WOMEN, POLITICS & POLICY The students received ten questions as part of a web-based survey. We balanced questions on male and female politicians and we included a number of female topics, while we also tested participants knowledge about the family life of a leading female politician (Table 1). The results from this test confirm our expectations to some extent. Gender differences were substantially reduced compared with the results of traditional research, although male respondents still outperformed female respondents (5.53 versus 5.11 on a 0-10 scale, significant at the.02 level). When looking at item level differences (Table 2) we notice that gender differences tend to be strongest in the traditional questions referring to male politicians. Both the question on Mr. Barroso and on Mr. Vandeurzen showed significant gender differences. It is striking to observe that in the questions on the female ministers Arena and Onkelinkx there are no significant gender differences at all. Only the question on Ms. Lieten does not behave according to expectations with men outperforming women, probably because she is associated with a nontraditional field for women, public transport. In 4 out of 10 questions, women outperform men, although the differences are not statistically significant. TABLE 1. Summary of the Questions Included in the Experiment Questions on Male Politicians Who is the president of the Christen-Democratic party? (correct answer: Mr. Jo Vandeurzen) Who is the president of the European Commission? (correct answer: Mr. José Manuel Barroso) Questions on Female Politicians Who is the Minister of Justice? (correct answer: Ms. Laurette Onkelinkx) Who is the minister of formation in the Walloon region? (correct answer: Ms. Marie Arena) What is the current function of Ms. Caroline Gennez? (correct answer: Vice-president of the Socialist Party) Who is the chairperson of the Public Transport Company? (correct answer: Ms. Ingrid Lieten) Traditional female topics The quality of child care is controlled by... (correct answer: The Government Agency Child and Family ) Child allowance is a... (correct answer: Federal competence) In which country does the new government have as many female as male ministers? (correct answer: Norway) Personal question This summer, vice prime minister Ms. Freya van den Bossche gave birth to a daughter. What is the name of that daughter? (correct answer: Billie)

6 Research Note 119 TABLE 2. Correct Answers on the Items % correct Women Men T-value P-value Ms. Onkelinkx, Minister Mr. Barroso, President Child Care Ms. Gennez, Vice-president Mr. Vandeurzen, President Child allowance Daughter, Billie Ms. Arena, Minister Norwegian Government Ms. Lieten, Transport Average Score N Notes: Entries for the first three columns are the percentage of correct answers on the ten knowledge question, according to gender. The last two columns summarize the results of the t-tests, which compare the means for the two groups (df = 467). Therefore, this test confirms our first hypothesis. Questions on male politicians resulted in large and significant differences, while this is not the case for most questions on female politicians where differences are minimal and not significant. It should be noted that including questions on female politicians does not lead to the introduction of a new form of bias (this time in favor of women), but rather to a neutral measurement, where differences between female and male scores are not significant. The second hypothesis assumes that there are also differences in the structure of political knowledge. If women and men indeed delineate the political realm in a different manner, this should imply that the structure of their political knowledge answers should be different. In that case we could expect that for women female topics like child care or child allowance will be an integral part of the latent variable political knowledge, while this would not be the case for male respondents, who are more likely to perceive this as a private problem. If we argue that women have a more personal and family oriented outlook toward politics and society (Okin 1989), tests should show that for them the name of the baby of a female politician (the Billie question ) is linked to political knowledge, while for men this is not the case. To ascertain whether these predictions hold, we will conduct validity tests for women and men separately. The correlation between the single item and the overall scale determines whether an item empirically belongs to the latent variable we want

7 120 JOURNAL OF WOMEN, POLITICS & POLICY to measure (Wilson 2005). If the structure of political knowledge is indeed different for women and for men, this should result in different validity scores for the two groups. Because this is a crucial test to determine the structure of political knowledge, we will use two different techniques. Item Response Theory (IRT) tests whether all the items can be placed on this scale; multi group confirmatory factor analysis (CFA) tests whether the political knowledge scale can be considered as one-dimensional for both groups. Item Response Theory (IRT) is routinely used in education research to test the validity of a measurement instrument. The test takes into account both the internal cohesion of the scale and the difficulty of every single item (van der Linden and Hambleton 1997). There are two different ways to conduct an IRT test. The first, known as the Rasch model, is rather straightforward: It allows us to predict the likelihood of a correct answer on an item, taking into account the overall score of the student and the difficulty of that specific item. In the Rasch model, the alpha correlation of the scale is higher for men (.52) than it is for women (.41) indicating that this scale works slightly better for men than for women. For all the different items, however, the gender differences are minimal (Table 3). A second, statistically more sophisticated approach to IRT TABLE 3. IRT Tests for the Validity of the Scale Women Men Rasch OPLM Rasch OPLM R it (w) R it (w) R it (w) R it (w) Ms. Onkelinkx, Minister Mr. Barroso, President Child Care Ms. Gennez, Vice-president Mr. Vandeurzen, President Child allowance Daughter, Billie Ms. Arena, Minister Norwegian Government Ms. Lieten, Transport Alpha Correlation Notes: Rasch and OPLM models for the validity of the scale. Entries are correlations between item and overall test (R it (w)), and for the OPLM models also the slope of the answering curve of the item ( ).

8 Research Note 121 testing also takes into account the fact that different items have different difficulty and discrimination curves ( ). This OPLM analysis provides us with a better test for the validity of the scale, with alpha correlations of.59 (men) and.57 (women). Again, we observe that the scores for women and men do not differ in a systematic manner. For both groups, the questions on child care and child allowances do not belong to the political knowledge scale, while the Billie question performs extremely well, both for men and women. Therefore, we do not observe any gender differences with regard to the delineation of the political realm. The second test in this analysis is a confirmatory factor analysis. First women and men are analyzed separately. In a second phase, the answers of the two groups are used in a multiple group confirmatory factor analysis (Table 4). The restriction on this model is a metric invariance one (Steenkamp and Baumgartner 1998, 80), which means that the factor TABLE 4. (Multiple Group) Confirmatory Factor Analysis for the Validity of the Scale Women Men Multiple group Mr. Vandeurzen, President a Ms. Onkelinkx, Minister Mr. Barroso, President Child Care Ms. Gennez, Vice-president Child Allowance Daughter, Billie Ms. Arena, Minister Norwegian Government Ms. Lieten, Transport Chi-square test b ; df = 24; p = ; df = 27; p = ; df = 53; p = RMSEA NNFI Notes: Results from two confirmatory factor analysis models for the two groups and one multiple group confirmatory factor analysis model with metric invariance restrictions. Entries are factor loadings from the factor political knowledge on the items. a The factor loading on the first item is always restricted to 1. b The WLSMV estimation method corrects for the chi-square and the associated degrees of freedom.

9 122 JOURNAL OF WOMEN, POLITICS & POLICY loadings are kept equal across the two groups. Because the data are categorical (dichotomous: correct or not correct), the weighted least squares means and variances adjusted estimation method is used (Muthén and Muthén 2006). The test reveals that the political knowledge scale performs well, both for women and men. To put it differently: the concept of political knowledge does not have a different meaning for women than for men. The multiple group model does not reveal any problems: the structure of political knowledge is not different for women than for men. The analysis also confirms that the Billie question is a valid element of a political knowledge scale. The conclusion from this experiment, therefore, is that measured gender differences in the level of political knowledge can be reduced substantially compared with the results obtained by conventional measurement instruments, by including more questions on female politicians. For items on female politicians, most often there are no significant gender differences revealed in the level of political knowledge. Put another way, including women-specific items increases the measured level of political knowledge for women without reducing it for men. Both the IRT test and the confirmatory factor analysis, however, also support the argument that there are no gender differences with regard to the structure of political knowledge. Both for women and men, the same items load well (or not) on the political knowledge scale. In this scale, we also included a question on the name of the recently-born daughter of a female politician. We included this item to test the assumption that women are more inclined to add this personal dimension to their interest in politics. The results of the experiment did not confirm this hypothesis: there were no significant gender differences in the test scores for the Billie item. Both for women and men this very personal item can be seen as part of a valid political knowledge scale: those who perform well on the traditional political knowledge questions (institutions, functions, etc.) also score well on this personal question. For all practical purposes, therefore, the tests show that the Billie question is indeed a valid measurement for political knowledge. Whether this also can be seen as a relevant form of political knowledge is of course a question that cannot be answered by using only statistical analysis. The one dimensionality we observe here could indicate that men and women are generally interested in the personal lives of public figures. Or such information might be especially relevant for young respondents, who are faced with the challenge of building their career and starting a family simultaneously. For them it might be relevant to

10 Research Note 123 know that a professionally successful woman can get pregnant, take a few months off, and go back to work. Or the finding might imply that the traditional separation between public and private to some extent has become obsolete (or was never useful). Politicians themselves now use elements from their private life as part of their political and public profiling effort, so it should come as no surprise that among the public, too, this border is blurred. In this case, the minister was quite public about her pregnancy, although she has effectively shielded her baby and her partner from any press attention. Do our findings imply that political knowledge scales should be altered? To those who would argue that political knowledge should not include the personal, we respond that political knowledge scales are not the appropriate platform to defend certain normative notions about the separation between the public and the private spheres. What knowledge scales should do, rather, is to measure as adequately as possible what people themselves consider to be politically relevant. Our test results suggest that women and men respond in the same manner to information about politicians personal lives. We believe it would be appropriate for political knowledge scales to recognize that for both genders the boundary between the private and public sphere may be more permeable than previously assumed. Textbooks in political science often take it for granted that women score lower on political interest and political knowledge than men do, and indeed thus far this was seen as a constant finding in political science research. Our data show, however, that a number of caveats are in order. First, the observed difference might be due at least partially to the measurement method being used. In recent years a number of arguments have been raised about the bias in current measurement methods, and we believe that by now these arguments should be tested in a largescale experiment on political knowledge. Only if we conduct such experiments can we ascertain whether gender differences remain present, even when using a gender-sensitive measurement instrument. Second, our findings suggest that institutional political items and more personal political items empirically belong to the same political knowledge scale. Our tests, however, cannot solve the question whether this is also a form of relevant political knowledge. To ascertain whether knowledge about institutional items has the same effects as knowledge on more personal items about the private lives of leading politicians, a thorough analysis is called for on the effects of different kinds of political knowledge. We know already that political knowledge leads to mobilization and participation. If this effect turns out to be just as strong for the personal items

11 124 JOURNAL OF WOMEN, POLITICS & POLICY as for the institutional items, we would indeed have every reason to consider the personal items as a form of relevant political knowledge. Third, we can observe that women tend to pay more attention to the activities of female politicians. At this moment we cannot decide why this is the case. Some authors, like Wolbrecht and Campbell (2005), might argue that leading female politicians act as role models. Others might argue that female politicians pay more attention to the political priorities and preferences of female voters. No matter what kind of causal mechanism is involved, however, it is clear that we can expect that the rise in the number of leading female politicians will have an effect on political interest and political knowledge among women. If countries like Germany and Chile are now being led by female politicians, if in the United States, the Speaker of the House and the Secretary of State are women, we can expect that female voters will become increasingly mobilized into politics. Our hypothesis for future research, therefore, would be that the presence of visible and successful female politicians will lead to an increase in the cognitive mobilization and the civic engagement of women. Self-evidently, this is quite a complex research question since it is just as likely that an already mobilized female electorate increases the odds that more women will get elected. Solving this question therefore requires longitudinal and comparative research. The 2007 presidential elections in France, and the 2008 presidential elections in the United States, however, might provide excellent opportunities to conduct exactly this kind of research. REFERENCES Campbell, David and Christina Wolbrecht See Jane Run. Women Politicians as Role Models for Adolescents. Journal of Politics, 68(2): Carroll, Susan, ed Women and American Politics. New Questions, New Directions. Oxford: Oxford University Press. Delli Carpini, Michael and Scott Keeter Measuring Political Knowledge. Putting First Things First. American Journal of Political Science, 37(4): Delli Carpini, Michael X. and Scott Keeter What Americans Know about Politics and Why it Matters. New Haven, CT: Yale University Press. Frazer, Elisabeth and Kenneth Macdonald Sex Differences in Political Knowledge in Britain. Political Studies, 51(1): Githens, Marianne, Pippa Norris, and Joni Lovenduski, eds Different Roles, Different Voices. Women and Politics in the United States and Europe. New York, NY: Harper Collins.

12 Research Note 125 Lovenduski, Joni and Pippa Norris, eds Gender and Party Politics. Thousand Oaks, CA: Sage. Lovenduski, Joni Women and European Politics. Contemporary Feminism and Public Policy. Sussex, UK: Wheatsheaf Books. Muthén, Linda and Bengt Muthén Mplus: User s Guide (Version 4). < Niemi, Richard and Jane Junn Civic Education: What Makes Students Learn? New Haven, CT: Yale University Press. Okin, Susan Moller Justice, Gender, and the Family. New York, NY: Basic Books. Steenkamp, Jan-Benedict and Hans Baumgartner Assessing Measurement Invariance in Cross-National Consumer Research. Journal of Consumer Research, 25(1): van der Linden, Wim and Ronald Hambleton, eds Handbook of Modern Item Response Theory. New York, NY: Springer. Verba, Sidney, Nancy Burns, and Kay Lehman Schlozman Knowledge and Caring about Politics. Gender and Political Engagement. Journal of Politics, 59(4): Wilson, Mark Constructing Measures. An Item Response Modelling Approach. Mahwah, NJ: Lawrence Erlbaum. Wolbrecht, Christina and David Campbell Do Women Politicians lead Adolescent Girls to be more Politically Engaged? Paper presented at the Annual Meeting of the American Political Science Association, 1-4 September 2005, Washington, DC. doi: /j501v28n02_05