Realization of Policies and Stance toward Political Parties
3. Relationship between Political Party Support and Attitudes on Policies
(1) Changes in Voting Behavior among Individuals
The data used here stem from the “Japanese Life Course Panel Surveys”8 con-ducted from 2007 by the Institute of Social Science, Tokyo University. The sources of these data are men and women from all over Japan aged 20–40. As of 2014, the project is ongoing, and a follow-up survey wave is to be conducted once a year.
Included in this panel survey are data on voting behavior for the last four national elections (the 21st House of Councilors election of July 2007, the 45th House of Representatives election of August 2009, the 22nd House of Councilors election of July 2010, and the 46th House of Representatives election of December 2012). In the analysis, the data of the second, fourth, fifth, and seventh waves are used, which includes these voting results.
Table 8-1 follows the change in voting behavior based on the voting behav-ior of 2,587 people over four elections. Among these results, the most numerous sample is that of 216 people (8.3%) who did not vote in any of the four elections.
After that, there were 153 people (5.9%) who voted LDP in all four elections, 148 people (5.7%) who voted DPJ in the first three elections but for another political party in 2012, 121 people (4.7%) who voted DPJ in all four elections, 77 people (3.0%) who voted DPJ in the first three elections but for the LDP in 2012, 67 people (2.6%) who voted for the DPJ in the first three elections but did not vote in 2012, 50 people (1.9%) who voted for New Komeito in all four elections, and 42 people (1.6%) who voted for the JCP in all four elections. Note that of the 434
other parties (e.g., the Japan Restoration Party, Your Party, or the LDP).
There were also many people who did not vote. The number of people who did not vote in both the previous 2010 election and this 2012 election, moreover, exceeded 500, which was close to 15% of this cross table sample. Although it is not possible to verify the voting behavior for the elections prior to this, by looking at this cross table alone, we can tell that there are a certain number of people who were not exercising their vote, either because they were disillusioned with politics or because they had lost interest. Moreover, as we can also see in Figure 7-3, the LDP recovered a large number of seats in this election (winning more than 60%
of all seats), but their vote ratio did not reach 30% of proportional representa-tion. The table shows that the LDP voters were certainly the largest number (878 people). However, this number does not differ much from the other 842 people who voted, and the LDP only gained 37% of the entire vote. Moreover, including the people who chose not to or forgot to vote, the LDP received just over 25% of voter confidence.
Two particular characteristics of the people who voted DPJ in both 2009 and 2012 are that 1) significantly more of them were born between 1966–1975 (rather than between 1976–1986); and 2) significantly less of them worked in sales, were farmers/self-employed, or were blue-collar workers (in contrast to office work). In other words, this means that the non-white-collar strata in the younger group had a propensity not to vote for the DPJ.
(2) Political Party Favorability Rating and Political Attitude
According to Yukio Maeda, who analyzed these panel data, the choice of the sup-ported political party (not the political party voted for in the election) has stability to some extent: there are many people who change from supporting a particular party to not supporting any party at all, but there are not many people who switch allegiances from one party to another or repeatedly change the supported party.
Among the women, there were many who were unwaveringly independent from any party, or who were politically apathetic, while among the people supporting a political party, there were many men. The more the number of LDP supporters, DPJ supporters, and unwaveringly independent people increased, the higher the level of education. However except for the small number of women’s support for the DPJ, the overall relationships between social attributes and political party sup-port are vague. In terms of individual mindset and political policies, for the gap mindset (i.e., that it is necessary for the prosperity of Japan that the income gap is big) and for the marriage mindset (i.e., generally speaking, a married person is happier than a person who is not married), there is a connection with LDP sup-port and unwaveringly independent people; and for both cases, the more the per-son agreed with the statement, the stronger the tendency for that perper-son to be an LDP supporter. In general, opinions on defense policy are clearly divided between LDP supporters and DPJ supporters, and there are no welfare-related grounds causing a division between the two. For defense and welfare, the views of LDP and DPJ supporters clearly differ from those of New Komeito and the JCP. Among the people without a clearly supported political party, there were those who were the DPJ; however, notably at the same time, there were a considerable number of
cases of people not having voted in the previous election deciding to vote for the DPJ in 2009. From this, we can well understand that there were enormous expec-tations for regime change in 2009.
Looking at a comparison of 2010 with 2009, while there were not many people who swung from the LDP to the DPJ, a good number of people swung from the DPJ to the LDP. As a whole, however, there were many people who contin-ued to vote for the the DPJ. In this election, the DPJ’s vote ratio and seat ratio fell compared not only with the previous House of Representatives election but also with the House of Councilors election of 2007. Despite the air of defeat sur-rounding the DPJ, which was suffering ongoing party turmoil characterized by the Hatoyama administration’s collapse over the relocation of the Futemma Base, in terms of the actual vote ratio, this was still higher for the DPJ than the LDP.
Ultimately, however, due to the nature of the electoral system, the number of LDP seats increased (Figure 8-1). In the panel survey’s sample, which is limited to a pro-portion of the younger generations, and not therefore representative of all of the voters, there were 554 proportional representation LDP voters and 924 propor-tional representation DPJ voters, showing that there were overwhelmingly more DPJ voters. It should be pointed out that among the people who voted DPJ in the previous House of Representatives election but swung on this election, compared with those who swung to voting LDP, there were quite a number of them that swung to vote for another party (e.g., Your Party) or chose not to vote at all.
We see a completely changed situation, however, when we look at the House of Representatives election of 2010 and the House of Councilors election of 2012.
Among the people who voted for the DPJ in 2010, those who voted for DPJ in 2012 were not the majority; the overwhelming majority had swung to one of the Figure 8-1 Trends in Results of House of Councilors Elections
Source: MIC Notes: YP=Your Party
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
21st 22nd
LDP DPJ NK JCP
SDP YP Other
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
21st 22nd
LDP DPJ NK JCP
SDP YP Other
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
21st 22nd
LDP DPJ NK JCP
SDP YP Other
This shows that the preferences (or scores) among the individuals are more dis-persed for these parties.
Next, Table 8-3 shows the trend of scores for attitudes on policies and political opinions. The values range between 1 and -1, with positive scores indicating ment. The movement that stands out over this period is a sharp increase in agree-ment for the strengthening of Japan’s military defense and support for the Security Treaty, perhaps reflecting, in particular, the deteriorating relationship with
Table 8-2 Changes in Political Party Favorablity Rating (max score = 100)
2008 2010 2011 2013
LDP Ave. 42.65 40.48 44.67 52.05
Std. Dev. 20.62 20.29 20.45 20.81
DPJ Ave. 47.37 44.45 37.60 34.38
Std.Dev. 18.73 20.87 21.40 20.72
NK Ave. 33.28 30.95 31.88 32.62
Std.Dev. 22.61 22.75 22.89 22.82
JCP Ave. 33.50 33.78 32.97 32.32
Std.Dev. 21.43 21.39 21.76 21.62
SPJ Ave. 33.98 33.55 32.99 30.71
Std.Dev. 20.24 20.48 20.60 20.92
N 3785 3070 4071 3675
Source: JLPS
Table 8-3 Changes in Score of Political Attitudes
2008 2010 2011 2013
A) Strengthen defence Ave. .132 .096 .301 .389
Std. Dev. .738 .714 .682 .674 B) Maintain Security Treaty Ave. −.017 .034 .193 .290 Std. Dev. .661 .651 .628 .621
C) Narrow income gap Ave. .350 .293 .275 .276
Std. Dev. .677 .698 .675 .664 D) Secure employment through public Ave. .370 .439 .454 .448
works Std. Dev. .688 .686 .670 .667
E) Enhance welfare Ave. .663 .625 .561 .467
Std. Dev. .571 .590 .639 .674 F) Not entitled to welfare Ave. −.003 .066 .146 .279 Std. Dev. .774 .773 .753 .732
N 3895 3127 4226 3736
Source: JLPS
repulsed by political parties and others with no feeling of opposition against politi-cal parties; and there was a strong general tendency among the former to vote LDP in 2005 and for the latter to vote DPJ in 2009 (Maeda, 2013).
Although people can only choose one “supported political party,” as previously mentioned, an individual may give a high favorability rating to a political party even if it is not their first choice. Under such circumstances, it is conceivable that the individual could decide to change which political party to vote for. In this survey, there are questions that use the sentiment thermometer. Because the sub-jects were asked for their favorability ratings of the five parties (LDP, DPJ, New Komeito, JCP, and SDP) over the course of four surveys, we will here look at the changes in these. Moreover, it is pertinent to investigate whether this change was influenced by individual environmental (or attributable) factors and whether it relates to a change in mindset toward the policies.
However, the mindset on education policy, the policy directly relevant to this book, is not included in the survey. Instead, we will therefore use these data to ana-lyze items of mindset related to various attitudes on politics and policy. We exam-ine here whether there is some kind of relationship between changes in mindset and changes in political party favorability ratings. A question concerning mind-set asked, “Do you agree or disagree with the following opinions? For each opin-ion, please mark a circle to indicate your degree of agreement/disagreement.” The opinions stated on the survey were: A) Japan’s military defense capabilities should be strengthened; B) The Treaty of Mutual Cooperation and Security between the United States and Japan should be strengthened; C) It is the responsibility of the government to narrow the income gap between people with a high income and people with a low income; D) Public works projects are important for securing local employment; E) Even when the budget must be tough, social welfare such as pensions and healthcare for the elderly should be enriched as much as possible;
and F) Apart from the elderly and people with mental or physical disabilities, all people must live without counting on social welfare. Each opinion had five levels from agree to disagree.
For this analysis, however, “Agree” and “Somewhat agree” were collated and given the score of 1, while “Disagree” and “Somewhat disagree” were collated and given the score of −1. “Neither agree nor disagree” was given the score of 0. The answer “Don’t know” was also an option, and this was also given the score of 0 for the sake of convenience.
First, let us look at the trend of the favorability ratings of the political parties.
Table 8-2 lists the results. As already stated, using only the data from 2007, 2009, and 2012 ensures that the targeted data belong to surveys with data of national-election voting behavior. Even these can only be treated as a rough trend. The favorability rating for the LDP was at its lowest in 2010 directly after the change of government. Thereafter, it began to improve, reaching over 50 in 2013. The favor-ability rating of the DPJ, on the other hand, follows a downward course. There are not any significant changes in the favorability rating of the New Komeito, the JCP, and the SDP, with each lower than the LDP and the DPJ. New Komeito and the JCP show an increase in the standard deviation compared with the other parties.
The explanation of exact details of the analysis is provided in Allison (2009) and Nakazawa (2012). One thing that should be mentioned, however, is that the coefficients that are constant variables show a relationship between the individ-ual characteristics that were originally possessed by the respondent and his or her favorability ratings towards political parties, which is the dependent variable (in other words, it changes between individuals). Allison does not proactively provide interpretation for a coefficient of the average among the individuals for the vari-ables that may change. One thing that should be mentioned, however, is that the coefficient shows the difference between the individuals with average scores for political attitude as the dependent variable when the average among the individual has risen by 1. On the other hand, the coefficient of the variable that took the dif-ferences between the average among the individuals and the score of each time point among the individuals shows the amount of change of the political attitude as the dependent variable when there is a change of one explanatory variable for an individual. This coefficient matches the coefficient of the fixed effect model in the econometric model in the case that data have no loss (balanced panel data). In other words, when you want to see “the favorability rating of the political party of a person who has had a certain thought from the start” you should look at the coef-ficient of the average for the individual, and when you want to see “how favorabil-ity rating for a political party changes among a person who comes to hold certain thoughts” then you should look at the average coefficient for an individual.
Table 8-4 lists the analysis results. The occupation and household income were removed from the list as they had no significant effect on the favorability ratings of political parties, and they were unnecessarily complicated. The dummy variable coefficients at the time of the survey were also removed from the results because they only indicated the changes of the favorability ratings of the political parties from 2008, and the general trends can be evaluated from Table 8-2.
Looking at the differences among individuals, New Komeito and the SDP had low scores among people with high academic records. The reason for this was not clear, but at least on distribution, compared with junior/senior-high school gradu-ates, New Komeito received low scores of 2.897 points among two-year/technical college graduates and 5.635 points among university/graduate school graduates.
The SDP received a low score of 1.546 points for university/graduate school grad-uates. In contrast to this, women tended to give significantly higher points to New Komeito, the JCP, and the SDP.
First, let us take a look at the coefficient of the average for the individual. The LDP average for the individual is positively significant by defense capacity, security treaty, and public works project, while the income gap cancellation is significantly negative. In other words, originally the person with an affirmative opinion gives the Liberal Democratic Party a high score with respect to the three former issues and has an affirmative opinion of the latter one, giving the Liberal Democratic Party a low score. The coefficient of the average for the individual reflects the dif-ference in ideology between political parties to some extent. People with an affir-mative stance toward military defense and the Security Treaty tend to give high points to the LDP, and they also rate the DPJ highly (although not as much as neighboring countries. In addition, the opinion that welfare should be enhanced
remained dominant, but the score considerably decreased; in contrast, the oppo-site opinion toward welfare increased. It is not possible to evaluate the vote ratio of the House of Representatives election of 2012 as entirely representing a return to the LDP. A new conservative, third force in the political arena emerged in the shape of the Japan Restoration Party and Your Party, and both these parties advo-cate small government. We can expect the change in orientation to these political parties to be reflective of the trend of political opinions.
(3) Analysis of Panel Data
In Chapter 7, JGSS data were used to analyze the relationship between social attributes and political attitudes. As JGSS represent cross-sectional data (as data obtained from one survey), even if some relationship between mindset and social attributes were to be detected, the most rigorous assertion that can be made is that there is some kind of relationship in the distribution. For example, it is not possible to determine whether it is because people originally had that social status that they tend to hold these attitudes, or whether people tend to assume these attitudes by acquiring such social status. In other words, when there is a relation-ship between independent variables and dependent variables, it either means that dependent-variable change is brought about by independent-variable change, or it just means that the person with independent-variable characteristics also tends to have specific dependent-variable ones, and that independent-variable changes do not necessarily lead to dependent-variables change. Unless panel data are viewed, then it is not possible to distinguish which is the case. In order to perform this rig-orous classification, it is essential to perform panel data analysis.9
Although there are several methods that can be used for panel data analysis, we will examine changes in the favorability ratings of political parties using the hybrid model that is advocated by Paul D. Allison (Allison, 2009).
The explanatory variables used here are sex (the dummy variable of woman = 1), birth cohort (the dummy variable assumes 1 if the birth year is within the range of 1966–75, with the standard category assigned to birth year within the range of 1976–86), educational background (three categories of junior/senior high school graduate, two-year/technical college graduate, university/graduate school graduate), occupation (as in Chapter 7, occupational classification is based on the Erikson-Goldthorpe-Portocarero typology), household income, marital sta-tus, and dummy variable showing the time of survey (2008 is made the standard variable, and 2010, 2011, and 2013 are included as dummy variables). Among these variables, sex, birth cohort, and educational background are those that do not change in the individual within the observation period. The other variables may change within the observation period. Concerning the other variables that may change, except the survey time dummies, models were cast for both the aver-age score among the individuals and the score showing the difference from the average among the individuals and the score for each time period. This represents the hybrid model. For the calculation, random effect estimation was conducted for the analysis of the panel data.
the LDP). In the case of the Security Treaty, however, the DPJ is significant at the 10% level, and the coefficient is also small. This may reflect the party’s stance on the U.S. base problem in Okinawa. Among the people supporting New Komeito, which is in coalition with the LDP, they are negative about strengthening military defense, although they are affirmative about the Security Treaty. The JCP and SDP supporters are also negative with regard to strengthening military defense, but in the SDP, the negative degree is weak and not significant concerning the Security Treaty.
The LDP/New Komeito stance and the JCP/SDP stance are in opposition to each other on the issue of the income gap, but the positive coefficient of the DPJ is not significant. On the issue of public works projects, opinion was significantly negative among the DPJ supporters but significantly positive among the LDP and New Komeito supporters. The support base for the SDP, which was in coalition with the DPJ when they formed the government, tended to be affirmative about securing employment through public works projects. On the issue of the enhance-ment of welfare, the divide in opinion takes the shape of the LDP on one side and the other parties on the other.
When looking at change within the individual, in other words, the coefficient becomes significant for the LDP, as it did for the average for the individual. This means that there is a relationship whereby the favorability rating for the LDP tends to be higher among people with affirmative opinions on defense, the Security Treaty, and public works, and lower among people with affirmative opinions on eliminating the income gap and enhancing welfare. Those among the DPJ sup-porters originally with affirmative opinions on the Security Treaty (at the 10%
level), had the tendency to give a high favorability rating, although rising only by 0.534 points, but when there was a change in the individual toward having an affirmative opinion on the Security Treaty, we can see the favorability rating for the DPJ lowering by 0.745 points. Looking at New Komeito, the influence on the favorability rating from changes in opinion was practically nonexistent. The favor-ability rating can mostly be explained by the differences in ideology and orienta-tion held from the start. The same can be said for the JCP, in that the propororienta-tion that is explained by difference in orientation from the start is large. Concerning the opinion that “people in society should subsist without counting on welfare,” it is held from the start, although there was not a significant effect for the favorability rating itself. If we connect the favorability rating to an affirmative opinion, we can understand the favorability rating toward the political party to be low. As for the SDP, the average for the individual was positively significant for public works projects, but for the change in the individual, it was significantly negative (at the 10% level); and considering the people who held an affirmative view on securing employment through public works projects, we can see that their favorability rat-ing toward the political party lowered.
The table does not show the effect of change on the favorability rating toward a political party for changes in occupation or income because there was very little.
However, with respect to occupation, it is worth mentioning that those people among New Komeito, JCP, and SDP that were originally blue collar tended to Table 8-4 Result of Panel Data Hybrid Model Analysis
LDP DPJ NK JCP SDP
Coef. S.E. Coef. S.E. Coef. S.E. Coef. S.E. Coef. S.E.
Female .094 .579 .439 .578 3.665 .703 *** 3.887 .654 *** 5.474 .598 ***
Born 1966–1975 .607 .547 .814 .545 .001 .664 −.977 .616 −1.765 .564 **
2-year/tech college .389 .655 .534 .653 −2.897 .795 *** .234 .739 −.631 .676 University/graduate school .974 .690 .995 .689 −5.624 .839 *** −1.083 .779 −1.546 .713 * Ave. in individual
Married .037 .643 −.439 .578 *** 2.104 .779 ** .038 .724 .799 .663
Strengthen Defence 1.710 .550 ** .814 .545 *** −5.989 .663 *** −4.271 .618 *** −6.955 .566 ***
Maintain Security Treaty 8.114 .644 *** .534 .653 + 5.606 .777 *** −2.277 .724 ** −.271 .664 Government should reduce
income gap −5.160 .501 *** .995 .689 −1.334 .606 * 2.583 .564 *** 1.272 .516 **
Local employment by public
works 6.372 .510 *** −1.113 .510 * 5.873 .616 *** .407 .573 1.573 .525 **
Improve elderly healthcare
regardless of fiscal state −.844 .574 2.739 .573 *** 1.256 .693 + 1.522 .645 * 3.004 .591 ***
No entitlement to welfare .509 .447 −.753 .447 + −1.319 .539 * −.630 .502 −.519 .461 Change in individual
Married −1.280 .798 1.511 .874 + −1.039 .763 −.633 .794 −.990 .783
Strengthen Defence 1.612 .325 *** −.189 .356 .237 .311 −1.026 .324 ** −1.011 .319 **
Maintain Security Treaty 1.989 .330 *** −.745 .362 * .548 .316 + −.209 .329 −.605 .324 + Government should reduce
income gap −.709 .306 * .243 .335 .163 .292 .220 .304 .335 .300
Local employment by public
works 1.290 .307 *** −1.297 .337 *** .082 .294 −.104 .306 −.498 .301 +
Improve elderly healthcare
regardless of fiscal state −.654 .326 * 1.165 .357 ** .233 .313 .375 .325 1.112 .320 **
No entitlement to welfare .463 .250 + −.292 .274 −.134 .240 −.545 .249 * −.224 .245
N of observations 12617 12629 12587 12583 12560
N of persons 4055 4057 4049 4048 4047
R2 (within) .122 .135 .007 .006 .017
R2 (between) .145 .056 .099 .065 .113
R2 (overall) .150 .088 .081 .056 .096
+p < .10 *p < .05 **p < .01 ***p < .001
Note: Other explanatory variables considered were employment and household income (both ave. in individual and change in individual).