Value Change in Global Perspective

Notes

Notes Chapter 1 1. Muller and Seligson's analyses are also distorted by the data they employ. They use the Freedom House codings of levels of democracy. This coding schema uses the stable democracies to define the top level of their scales. From the start, stable democracies have been assigned the maximum possible score, and cannot become more democratic. As Muller and Seligson's analysis is based upon change, this means they cannot attain high scores on their dependent variable. They are testing hypotheses about short-term change, but their methodology virtually guarantees that they cannot detect change among stable democracies. 2. As of November 1, 1993, the European Community was officially renamed the European Union. As all of the surveys we employ in this book were conducted before this change, we refer to these countries as members of the European Community. Chapter 2 1. Surveys of Luxembourg began in 1973 and of Northern Ireland in 1976, but these samples have been relatively small. Greece has been surveyed from 1980 onward and surveys began in Spain and Portugal in 1986. The time series provided by Greece, Spain, and Portugal are not long enough to effectively analyze the impact of generational replacement. 2. Beginning in the fall of 1990, the EuroBarometer surveys include the newly formed states from the territory that used to comprise East Germany. To maintain overtime comparability, we exclude respondents from these states from our analyses. 3. The results for 1993 are based upon EuroBarometer #39, conducted in the spring of 1993. These results were provided directly by EuroBarometer staff, which sent us the overall distribution of scores for each country on the four-item values index. As of this writing, we have not analyzed the relationship between age and values for the 1993 data. With a single exception, we discuss the 1993 results only in this chapter. 4. We combine results for 1970 and 1971 to be consistent with results presented by Inglehart 1977. For all subsequent reports we present results according to single years. For 1977, 1978, and for 1980 through 1992, we combine the results of two survey years. Clarke and Dutt (1991) and Duch and Taylor (1993) also employ a single annual result for their analyses, although Thomassen and van Deth (1989) provide results for each semiannual survey. 161 162 Notes to Pages 13-37 5. More specifically, the percentage difference index is equivalent to a mean score that assigns each Postmaterialist a score of 100, each respondent with mixed values a score of 0, and each Materialist a score of -100. In the vast majority of analyses, there is a monotonic relationship among the three categories of this measure. For example, in the 1992 U.S. presidential election Postmaterialism was positively correlated with voting for Bill Clinton. According to our analysis of the 1992 NES survey, among major-party voters who were Postmaterialists (N = 247),75 percent voted for Clinton, among those with "mixed" values (N = 860), 57 percent voted for Clinton, and among major-party voters who were Materialists (N = 212), only 44 percent voted for Clinton (based upon weighted Ns using the timeseries weight). 6. As noted above (note 3), as of this writing we have not analyzed the 1993 results by age. 7. Factor analyses demonstrate that the beautiful cities item does not clearly tap Postmaterialist values, and in our scoring we classify it as measuring neither Materialist nor Postmaterialist values. 8. Somewhat different scoring procedures are used when the twelve-choice measure is employed across the entire range of 40 societies sampled in the 1990-1991 World Values Survey. See chapter 7 for a discussion. Chapter 3 1. As there are only six data points for the United States, we cannot conduct the type of multivariate time-series analysis of value change that we will present for the Western European countries. 2. Economic data in this chapter are based upon statistics published by the Statistical Office of the European Communities. German economic statistics for 1990, 1991, and 1992 exclude the states that used to comprise East Germany. 3. Chatfield (1989, 62) points out that conventional tests of autocorrelation lack power when there is an N of less than 100. He suggests that the most powerful test is to examine the first-order autocorrelation function coefficient. As there are 18 observations for Denmark and Ireland, a finding of +I - .471 indicates the presence of autocolTelation; as there are 19 observations for the other countries and for the sixnation sample, a...