Youth Electoral Study (YES) - Report 5: Youth, schools and learning about politics

Updated: 18 July 2012

12. Bringing it all together: The impact of school variables on voting commitment and political knowledge

The findings regarding the importance of the school have thus far been discussed at a bi-variate level, that is, the simple relationship between aspects of the school and political behaviour. The interrelationship between the many variables of the school (as well as the family), have not been taken into account. Only with respect to the differences between government and private schools have we attempted to show that differences between the schools in student political behaviour are attributable to differences in the student populations of the schools.

In this section, the analysis of the relative effects of the school on student political behaviour will be more complex, and the interrelationships between the variables of both home and the school will be taken into account. We will first examine the relative impact of the school on one aspect of political knowledge, namely being able to give the correct names of the two Houses of Parliament. Then we will use the same analytic model to examine the determinants of voting commitment.

12.1 Political Knowledge: Naming the Houses of Parliament

Virtually all studies of adult political behaviour find that political knowledge is related to voting (Verba et al. 1995; Delli Carpini and Keeter 1996; Nie et al. 1996; Wattenberg 2008). We have found a similar relationship among our students in the YES survey. Using the ability to name the Houses of Parliament as a measure of political knowledge, the Pearson correlation coefficient between this and voting commitment is .262, which is moderately positive. Therefore it is useful to explore the relative importance of the school on political knowledge. In order to obtain a more accurate estimate of the school's influence, we will include family variables as controls, and also to obtain a better picture of the relationship between family and school regarding political learning and behaviour.

In Table 5, the full regression model to explain the ability to name the Houses of Parliament, with school and family variables, is presented.

Table 5: Beta Coefficients for Full Simultaneous Multiple Regression Model, Family and School Variables on Political Knowledge (Name Houses of Parliament)
Demographic/Background Variables Beta School Variables Beta
Father's education .072* Interest in Study Govt .104*
Mother's education .064* Stood for office .084*
Info from parents – voting .060* Get on with teachers .073*
Info from parents – politics .020 Voted in elections .037
Sex of student .003 Like school -.022
  Studied Government .008
Attend private school .008

* p > .05
R2 = .064.

The regression model is not very powerful in predicting who is able to name one or two of the Houses of Parliament. The R2, or variance explained, is only .064 per cent (Compare that with the 21 per cent in the same model to predict voting commitment in Table 6). On the other hand, there are some variables which clearly are important, although the sizes of the Beta coefficients are modest. Also, the results show that the ability to name the Houses of Parliament is determined by both family variables and school variables. However the top three most important variables are school-related.

Interest in Study of Government is the strongest variable in determining the ability to name the Houses of Parliament (Beta = .104), followed closely by Stood for Office (Beta = .084), and Getting on with Teachers (Beta = .073). These three variables represent three dimensions of the school, namely the academic dimension, the informal curriculum (school elections), and the school environment, the teachers. Students who find the study of Australian government interesting, who have been motivated to participate at a higher level in student government, and who have a positive relationship with the teachers, are more likely to be able to name the Houses of Parliament.

The family plays a parallel role to the school. Although not as powerful as the school, the acquisition of political knowledge is affected by a higher educational level of the father and mother (Father's education and Mother's education). Presumably the parental educational levels suggest significant involvement in the learning process of the children. But along with this, the student's own acknowledgement of parental source of information about voting (Info from parents – voting) makes it clear that the family and the school are partners in the political socialization of young people.

Assuming that the ability to name the Houses of Parliament is a surrogate for political knowledge generally, what can we conclude from Table 5 about the role of the school?


  1. Interest in the study of the Australian government is the most important determinant of whether a student can name the Houses of Parliament correctly.
  2. School elections, especially choosing to run for a student representative position, have beneficial effects in the acquisition of political knowledge.
  3. Teachers have an independent influence on the acquisition of political knowledge, probably because students who get along with teachers, will be more receptive to teacher instructions and teacher behaviour. In this respect, teachers serve as role models.
  4. The family and school complement rather than cancel each other. The more educated parents seem to be able to better politically socialize their children. Parents can influence the political knowledge and attitudes of their children, at the same time that the school teaches them in the classroom and in the organizational structure of the school.

12.2 Voting Commitment

We now turn to the question which is similar to the one we posed regarding the effects of private schools on aspects of enrolment and voting. Is there something special about school characteristics which, in and of themselves, develop a sense of civic or political duty, or create a familiarity with voting behaviour in young people, over and above what the family does?

To address this question, we will return to our regression model which we used in Figure 12 and Table 5. This will indicate whether there is a unique link between school and voting behaviour, over and above the importance of family variables. The results of this analysis are given in Table 6.

Table 6: Beta Coefficients for the Full Simultaneous Multiple Regression Model, Family and School Variables on Commitment to Vote
Demographic/Background Variables Beta School Variables Beta
Info from parents – politics .139* Interest Study of Gov .215*
Info from parents – voting .116* Voted in elections .116*
Father's education .101* Get along with teachers .087*
Sex of student -.024 Like school .060*
Mother's education .011 Stood for office .046*
  Attend private school .024
Studied Government .023

* p > .05
R2 = .21.

As in Table 5, the figures in Table 6 are Beta regression coefficients. They are in standard deviation (sd) units, and they all come from the same regression model. They also are grouped separately into family and school, and they are ranked by size in order to facilitate interpretation. The asterisks indicate the variables which are statistically significant, that is, they are of sufficient size as to be significantly different from zero. The non-asterisked variables could be zero, and therefore are not statistically significant and cannot be regarded as having any direct independent influence.

Because these figures came from the one regression model, each figure indicates the impact of that variable on voting commitment, controlling for all the other variables in the model. Thus, in the full model, the variable that exercises the largest impact on voting commitment is Interest in Study of Government, which has a Beta weight of .215, which means one increment in Interest results in a .215 sd increment in voting commitment. Furthermore, this impact occurs irrespective of the value of the other variables, that is for both boys and girls, for any parental educational or occupational level, whether students like school or get along with their teachers, and whether the students are in a government or private school.

What do the figures tell us about the importance of the school on voting commitment? Firstly, within the context of the variables in the model, the type of school the students attend – government, Catholic or independent – does not have any unique impact on commitment to voting. Nor does taking a subject about the Australian government have an impact. Whatever relationship might exist at the bivariate level with these variables, does disappear when the other variables are taken into account.

On the other hand, by far the variable with the strongest unique impact, either among the school or family variables, is Interest in the Study of Government, as noted above. The next most important school variables are Voted in a School Election, with a Beta of .116, followed by Get Along With Teachers, Like School, and Stood for Office. Clearly school elections, and the general satisfaction of the student with the school, have significant impacts on the students' commitment to voting, irrespective of whether the student is male or female, or whatever socioeconomic background the student comes from.

What is interesting in this model is the coexisting impact of the family. In other words, the student's commitment to voting is not a function of only the family, or only the school, but a combination of both. In the family, the extent to which the student learns about politics from their parents, and the extent to which the student claims to be influenced about voting by parents, both appear very important in affecting the student's commitment to voting. Again, these effects occur independently of the student's relationship with the school.

Using five family variables and seven school variables, it is possible to make some strong statements about the effect of the school on commitment to voting. The regression model explained 21 per cent of the variance, which by sociological standards, indicates a fairly strong model. What are some conclusions?


  1. Of the twelve variables in the model, Interest in the Study of Government is by far the most important school variable which determines whether a student says they would vote, even if they did not have to. (Beta = .215)
  2. School elections, especially the experience of having voted (Voted in Elections) is the second most important school determinant of voting commitment, but is only about half as powerful as Interest in the Study of government (Beta = .215 compared to .116).
  3. Having a positive attitude toward schooling (Like School), and being integrated in the school (Get Along with Teachers) also contribute to voting commitment. These variables relate to the school environment.
  4. Standing for a political position in school elections (Stood for Office) has a small but significant positive impact on voting commitment, but is only half as important as voting in school elections (Beta = .046 compared to .116).
  5. In the context of the other variables, the two school variables of taking subjects about the Australian Government (Study about the Australian Government), and the type of school a student attends (Attend Private School), have no effect on voting commitment. This may seem surprising for type of school (See the baseline regression in Table 4). In other words, any bivariate relationships between these variables and voting commitment are a function of other family or school variables in the model.
  6. Family variables are important determinants of voting commitment, concomitantly with school variables. The family as a source of political knowledge (Beta = .139), and the family as a source of influence about voting (Beta = .116) are both independently important, and this could have policy implications.
  7. The father appears more important than the mother with respect to voting commitment, and education has a positive effect.