THE STUDY OF POLITICAL PARTY AFFILIATION

Abstract

This study, based on the panel surveys made in September, 1956 and February, 1957, is an attempt at an analysis of: the reasons given by the respondents for supporting certain political parties, the characteristics of the groups who, showed throughout the first and the second survey no change in thier support of particular political parties, and the multidimensional positioning of the factors for supporting a political party, and finally at an attitude analysis through the modified latent structure analysis.
The questionnaire contains such items as: biased questions to measure the respondent's political party affiliation, reasons therefore, reasons for a turnover from the party affiliation, opinions about social insurence, government controls, scandals among government employees, international relationships and rearmament.
1. Analysis of the Reasons for Supporting a political Party
The reasons for the support of political parties given by the panel respondents were coded and a matrix was made of the reasons given in the first and second surveys. The e_ij represents the number of the respondents who gave the reason i in the first survey and the reason j in the second survey. The reason i is given the numerical value x_i and the cases with approximate values of xi were grouped together. This means that the reasons in the matrix are moved in such a way that those with greater frequencies are as much concentrated near the diagonals as possible.
Q=-^RΣ_i=1^RΣ_j=1e_ij(x_i-x_j)²
The greater the value e_ij is, the smaller will the value between x_i and x_j become and the smaller the value e_ij is, the greater will the separation between the two become. It is, therefore, possible to make Q in the above obtain the maximal value.
When the formula ∂Q/∂x_l=0 is written under the condition where 1/nΣx_i²=a², the maximal value obtained by the following equation is the greatest value in this case.
{-ⁿΣ_j=1j≠la_lj}x_l+ⁿΣ_j=1j≠la_ljx_j=λx_l(l=1, 2, ……, R).(a_lj=e_lj+e_jl).
Then we can find out x_i (i=1, 2, ……, R), which is the maximal value, under the condition where x=0.
The results enabled the grouping of reasons. It has become clear that the reasons for supporting the Liberal-Democratic Party and the Socialist Party can respectively be divided into 3 groups. Since the reasons have been coded into 12 items, this means that the reasons, though expressed in different wording, do not differ much in contents.
2. Characteristics of the Groups Consecutively Supporting the Liberal or the Social Party
Here an analysis was made regarding which questions the two groups were most likely to be characterized by. When the quantity given to each item is α_i and the average of α_i for n people for each item category is α,
α=1/_nⁿΣi=1αi.
Its variance is
σ²=1/n_nΣ_i=1α_i-α².
The average of each grouped stratum, αi, and the variance between each stratum, σ_b², are as follows: