Gender and political orientation in childhood

Abstract

This study explores whether boys and girls differ in their levels of political orientation and the extent to which race/ethnic heritage mediates such an association. We analyze survey data for 14,855 children across 20 states using a fixed-effects analytical technique that confines the children to their immediate environments. We find that girls surpass boys in political interest and activity, and this persists without a significant drop in teen years as might be expected. This pattern is evident whether political orientation is measured as a composite indicator or as discrete items tapping specific activities and opinions. Within subgroups, White and Native American girls consistently displayed higher levels of orientation than comparable boys. Among Black, Hispanic, and Asian Americans, observed advantages for girls attenuate with full specification of control variables; at worst, Black, Hispanic, and Asian girls are equally as political as comparable boys. These findings have implications for theorizing about political orientation in childhood.

Acknowledgements

A previous version of this manuscript was delivered at the 1998 Annual Meeting of the Western Political Science Association, Los Angeles, California, March. We thank Susan J. Carroll, Warren E. Miller (deceased), Heather Campbell, Robert Benedetti, and Syd Goldstein of the Kids Voting USA program for their helpful comments on earlier drafts of this manuscript.

Notes

1 Socialization also has been advanced as an explanation for women’s occupational patterns, see Kanter (1977). Theorizing on the biological foundations of these differences is persistent, although the initial force of this line of inquiry has dissipated, see Kelly and Boutilier (1978); Bourque and Grossholtz (1974).

2 The states and samples are: Alaska, 527; Arizona, 827; Arkansas, 820; California, 814; Colorado, 460; Florida, 722; Georgia, 992; Kansas, 1055; Kentucky, 513; Maryland, 1401; Michigan, 735; Minnesota, 943; Mississippi, 708; North Carolina, 806; Ohio, 533; South Carolina, 724; South Dakota, 881; Tennessee, 323; Texas, 438; Washington State, 633. Students are from 199 schools.

3 Of course, asked in a vacuum, this question would be inappropriate for these children. However, as part of a project they were very knowledgeable about, they understood the question.

4 Variable Coding. Did you ask questions about voting at home? Frequently or Occasionally=1; Never=0. How often did you watch things about voting on TV? Frequently or Occasionally=1; Never=0. How often did you read things in the newspaper about voting? Frequently or Occasionally=1; Never=0. How often did you hear things about voting on the radio? Frequently or Occasionally=1; Never=0. How often did you read things about voting in magazines? Frequently or Occasionally=1; Never=0. When political and social issues are debated at your school, do you participate in those discussions? Frequently or Occasionally=1; Never=0. Would you like to have Kids Voting as part of what you learn in school the next time there is an election? Yes=1; No=0. Did you go to the polls and vote on election day? Yes=1; No=0. How important is it for people to vote on election day? Very or Somewhat Important=1; Not Very Important=0. How often does your family have political discussion at your home? Frequently or Occasionally=1; Never=0. How often did your teacher talk to you in class about voting? Frequently or Occasionally=1; Never=0. Does your family subscribe to or read a newspaper on a daily basis? Yes=1; No=0. Gender: Female=1; Male=0. Grade level: two levels of coding. Level one: actual grade level. Level two (grouped dummies): grades 4–6; grades 7–9; and grades 10–12.

5 McLeod et al. (1999) reported that higher education and income levels are positively related to newspaper readership. Dominick (1999) reported that education and age help explain the decline in newspaper readership. A 1998 national report commissioned by two newspaper trade associations reported that newspaper readership increases with education, household income, job responsibility, value of home and higher paying occupations (American Society of Newspaper Editors and The Newspaper Association of America, 1998).

6 Models are of the form: I=f(G, X, Zj), where I is the index of political orientation, G a dummy comparing girls to boys, X a vector of other factors, and Z the set of school dummies.

7 We compared the indices for the sexes across grade and for grade groupings 4–6, 7–9, and 10–12. Girls’ indices were higher than boys’ throughout. Except in the 11th grade, these differences were significant, with a slight attenuation of gender differences in the middle and higher grades.

8 Analysis of group composition in the population (children ages 10–19 as of July 1995, Statistical Abstract 1996) and the sample (data for Fall 1994) did not necessitate weighting: non-White population=29.6%, sample=30.5%; Blacks=13.7%, sample=15.4%; Asians=3.4%, sample=3.9%; Hispanics=11.3%, sample=6.6%; and Native Americans=1.0%, sample=4.7%.

9 The fixed effects feature introduced into the analysis did not allow for the specification of regional dummies. We reestimated Model 2 (Table 3) with regional dummies, the South serving as the reference category: POI=3.33^∗∗+.56^∗∗(Female)+.06^∗∗(Grade)−.34^∗∗(Black)−.21^∗(Asian)−.34^∗∗(Hispanic)−.18^∗(N.Am.)+1.24^∗∗(Pol.Dis.)+.60^∗∗(Newspaper)+.99^∗∗(Teacher)−.01(Gender×Grade)−.21(Gender×Teacher)+.26^∗∗(Midwest)+.14^∗∗(West). (^∗p<.01; ^∗∗p<.001).

10 We explored the relationship between the interaction terms and gender (Model 2) for Blacks, Hispanics, and Asians. Both terms are relevant, although none is significant. Thus, not allowing the gender effect to vary by grade/teacher exaggerated the additive results for Black, Hispanic, and Asian girls. Any gender differences across grade are not a significant explanation for gender differences in orientation within these groups.