We saw in the previous section (Section 6) that political party identity is an important factor in explaining why some young people are committed to voting, and others are not. We also saw in Section 3.3 that there is considerable family inheritance in party identity when we examine each party individually. Of course family inheritance in party identity can be explained in many ways, for example by means of political party "cues" which young people pick up from their parents, or through the perceived link between the social status of the party and that of the family [Edwards, 2006 #220; Ventura, 2001 #231].
Although the body of research linking party identity and perceived parental identity has been documented, researchers agree that these identities can change, and are subject to other non-family influences [Beck, 1991 #244].
But the important question to be addressed is whether or not there are other factors which also explain the existence of party identity among young adults. In other words, what are some of the characteristics which differentiate between those students who identify with a political party (who named a party) and those who do not?
In order to answer this question, we will repeat the regression analysis in Figure 7, but instead of explaining propensity to vote, we will explain the propensity to name a political party. All other variables in the model are the same. But there is one major difference which has to be taken into account, namely the nature of the variable that we are trying to explain. In Figure 7, we were explaining commitment to vote, a variable which consisted of five categories. The type of regression analysis we used was appropriate for this variable. However with party identity, we are using a variable with only two categories, namely does the student identify with a party or not?
When the variable we want to explain (the dependent variable) only has two categories, the usual multiple regression approach is not appropriate because some key statistical assumptions are violated, one of which is that the dependent variable cannot be normally distributed (Hair Jr, Anderson, Tatham, and Black 1995). In order to get around this condition, another form of regression procedure is used, namely logistic regression.
The selected results from the logistic regression are presented in Figure 8. This figure presents the "odds ratios" for only those variables that significantly predict whether a student named a political party that he or she "feels closer to".
The odds ratios can be interpreted as follows. Let us first consider the first variable, "prepared to vote". When the student's score on the "preparedness to vote" scale increases by a unit value, the chance that the student will name a political party increases by a value of 1.995, or roughly 100%, controlling for the other variables in the model. At the lower end of the figure, when "information from parents" increases by one unit value, the odds ratio of that person naming a political party increases by a value of 1.096, or roughly 9.6%. In other words, the increase is significant, but not by very much. An odds ratio of 1 would indicate that the variable in question is not related to an increase in the likelihood of naming a political party. All of the variables reported in Figure 8 have an odds ratio significantly greater than one.
Thus from Figure 8, we see the characteristics which are related to an increase the likelihood that a student named a political party, controlling for the other variables in the model. Thus, from the model we can see that "feeling prepared to vote" is clearly the most related to an increase, followed by "interest in politics". Similarly, even as a student progresses from Year 10 to Year 12, the likelihood of naming a party increases, though not by much.
Perhaps of equal interest is the collection of variables that were not significantly related to naming a political party, when controlling for all variables in the model. These variables are as follows: 1) "father's occupation", "finding the study of civics interesting", "liking school", "national pride", and "having stood for election as a student representative".
In conclusion, given the fact that the ability or willingness to name a political party is significantly related to commitment to voting (see Figure 7), it is useful to know some of the characteristics which differentiate between those who name a party and those who do not. Furthermore, if some of the characteristics can be incorporated into a program aimed at the improvement of youth enrolment and voting, either through the media or the school, in the awareness of, and identification with a political party among young people, especially those who are nearing the voting age, then the level of youth participation in enrolment and voting can be increased.
Do the findings in Figure 8 give us something to work with in this regard? The answer is "yes". With the exception of gender and the family, all the remaining five determinants can be incorporated into an educational program. The feeling of "being prepared to vote", the "interest in politics", the "conviction of the importance to vote", and more "positive engagement with the school" (interaction with teachers), are all factors which can be targeted and incorporated into the school curriculum.