IS THE GENDER GAP REAL?

Table 2.20

Left-Right Self-placement by Country and Gender

Own views left or right

Bulgaria Estonia Hungary Poland Slovakia

Male 4.14 4.68 3.79 4.04 4.23

Female 4.03 4.46 3.53 4.03 4.10

F 0.40 4.80* 3.2 (.07) 0.00 0.70

n = 930 790 907 696 858

Source: Own calculations based on data from the LRS of the ILDGP, 2002–03.

Figure 2.18

Within Cohort Gender Gap: Self-placement on Left-Right Scale by Country

0.0

Bulgaria 0.1

0.2 0.3 0.4

Estonia Poland Slovakia

Born 1945–59 Born 1960–

Hungary 0.5

0.6 0.7 0.8 0.9

Left–Right score gap

Source: Own calculations based on data from the LRS of the ILDGP, 2002–03.

Norris. The Inglehart and Norris scale was applied by them to data from national sam-ples of individuals in 70 different countries between 1990 and 2001. The gaps between male and female interviewees show women to the left (towards the “more government involvement” side of the scale) in 65 of these countries. Interestingly, our analysis of local officials closely mirrors the differences found at the mass level for each of the countries, but the point differences are one to three points higher (on the 100-point scale).

A key question is whether the impact of gender is sustained when we control for a variety of other factors. In order to further examine this possibility we have pooled our data sets for the five Central and Eastern European countries included in the survey of local officials. We have already noted that the male-female gender gap is statistically significant for four of the five countries (Bulgaria being the lone exception). In our pooled data set we first look at the “role of government” measure to determine if it re-mains affected by gender among local government officials in all five countries. Using an ANOVA we in fact find that the relationship is highly significant (F=65.1, p<0.001).

Our next strategy is to move beyond the bivariate relationship to a multivariate analysis. We employ an OLS regression to see if the impact of gender remains important after we control for a variety of other factors. The independent variables are handled in four groups: (1) social background factors (age, education, and gender), (2) cultural values (trust in institutions, postmaterialism, and left-right ideology), (3) democratic values (democracy as the best system, political tolerance, minority rights, participation, and media independence), and (4) a series of country dummy variables to sort out country-specific effects.

Education is difficult to compare because of the lack of exact equivalencies across nations. However, we are dealing with elected local government officials, about half of whom have had at least some form of post secondary higher education. Hence we use this break point for its comparative validity. The dummy variable we created is for some higher education (1) or no post-secondary education (0). We also grouped individuals by age to get a post-Cold-War and a pre-Cold-War generation (arbitrarily defined as those born before and after 1960) and created a binary variable for age cohort. We also tested the model with actual age as a variable. The “trust in government” variable is a combined measure of responses to a series of questions on the trust in various government institu-tions (executive, legislature, parties, police, justice system, etc.). The postmaterialism scale is Inglehart’s standard four-point scale used cross-nationally in the World Values and related surveys. We also ran these regressions with dummy variables for each of the countries to further sort out possible country-specific effects.

In table 2.21 we test five models to explain “the role of government” index. In Model One (not shown in the table) we examine the impact of gender alone. It remains quite significant as a factor in our pooled data set. In Model Two we add in social background characteristics. Age, higher education, and gender are all statistically significant in this model. In Model Three we add in the cultural and democratic values dimensions.

As can be seen in this model all three cultural values, trust in institutions, postmaterial-ism, and left-right self-placement are statistically significant contributors. Among the democratic values items, democracy as the best system, support for minority rights, and media independence are statistically significant but neither political tolerance nor the value placed on popular participation figure in the equation. Gender and higher education remain significant but the age variable drops out in Model Three.

In Model Four we add in the organizational membership variables, party member-ship, and membership in an NGO in which they hold a position of responsibility. The other variables that are statistically significant in Model Three remain so, but the two additional variables are significant, and the explained variance increased a bit. It is worth noting that when we ran this same regression for each country sample individually, the organizational membership variables are both significant only in Bulgaria. Party membership but not NGO position is significant in Hungary.

In Model Five we include the same sets of variables but add in dummy variables for country to sort out any country-specific effects. Interestingly, we see significant country effects for Estonia, Hungary, Poland, and Slovakia. However, we can explain the same variance by including only Bulgaria. When any other countries are included with Bulgaria, the combination does not increase the explanatory power of our models.

Bulgaria appears to be somewhat different than the other four countries in the views of local officials on the role of government. In Model Five we see that all of our background factors, including age, all of the cultural values variables, the same three democratic values, and the four country dummy variables are significant. However, the organiza-tional membership variables drop out of the equation (in terms of significance).

These models confirm the ongoing impact of gender on overall perspectives on the role of government in the economy. When we examine the standardized betas we see that the more educated, the more right-wing ideologically, the more postmaterialist, and the greater the level of trust in government institutions, the greater the support for a strong private sector role or less government involvement in the economy and responsibility for the well-being of the country’s citizens. The democratic values items show that greater support for democracy as a system and for a free press are associated with support for more government involvement in the economy, and greater commitment to minority rights is associated with a preference for less government involvement. The impact of gender is consistent with earlier findings. Women are more likely to support a larger government role and collective responsibility for individual welfare than are men, who seem to prefer the private sector and more individual responsibility. Age cohort did not have a significant impact in the pooled data set, but age (as a continuous variable) did.

It has a very limited impact (approaching statistical significance) with the young being only slightly more likely to support a stronger role for government.

When we entered in the country dummy variables, all were statistically significant, and they had some impact on the dependent variable (as measured by the standardized

betas). They did not affect the significance or impact of the other variables. Bulgaria stands out in this respect as being somehow different than the other four. While the signs for the betas for the other four countries are negative (more supportive of greater government involvement), the sign for Bulgaria is positive. That is, Bulgarian local of-ficials differ from their colleagues in the other countries in being much stronger in their support for privatization and a greater role for the private sector. What is consistent across all of these models, however, is the impact of gender. Therefore, our overall conclusion is that the gap associated with gender does indeed have its origins there rather than in other characteristics, and is of some importance (see table 2.21).