But the percentage of people who vote in the United States is low by international standards. Furthermore, voting rates are unevenly distributed across socioeconomic groups, with much lower voting rates among the poor. We find that a cash transfer program that increases household incomes has a positive effect on the voting behaviors of children from the poorest households. This suggests that income augmentation programs that help children may have other indirect and long-term benefits to society in the form of increased political participation and civic engagement as adults.
There may be more reasons to champion existing programs and policies aimed at reducing child poverty — not only will they reduce current disparities, they may play a role in ensuring the future electorate will be more representative as well. EF Explains Econofact Chats. See Table 1 for the descriptive statistics of all variables. See Online Appendix 2 for a detailed description of the harmonisation procedure for each variable. Voting In all surveys that were included in the harmonised PolPart dataset, respondents were asked whether they had voted in the last national, parliamentary or presidential election.
Household income was measured differently across surveys. The original deciles or quintiles from these surveys are based on the actual income distribution in the countries. Therefore we based the income quintiles for each country-year in these surveys on the distribution of income in the corresponding sample.
For a detailed explanation of how the income variable was harmonised across surveys, see Online Appendix 2. To be sure that differences in the effect of relative income on voting are not caused by differences in measurement across surveys, we did some additional analyses see Online Appendix 3. The coefficient plots in Online Appendix 3 show the effects of relative income for each country-year for the surveys in which this country-year is included. In general, the effect of income is similar when compared between surveys within country-years.
There are only a few country-years in which we found the effect to be substantially differ between surveys. Altogether, based on these additional analyses we are confident that the differences in the effects of income on voting between country-years are not caused by differences in the operationalization of income.
Respondents were asked about the highest level of education they accomplished in all surveys. The number of answer categories varied between the surveys. The education variables were harmonised into a 5-category variable with categories: No education or only primary education 1 ; Some or lower secondary education 2 ; Completed or higher secondary education 3 ; Some non-university tertiary education 4 ; and University 5.
We also created a quadratic term and included this in our models to account for possible non-linear effects of age. This dataset was constructed based on data from many national and international data sources. The SWIID uses a missing data multiple imputation algorithm to standardize the observations that were collected from the various source datasets in order maximize the comparability of Gini coefficients for the broadest possible sample of countries and years Solt This means that all Gini coefficients are estimated times for each observation i.
We computed the Gini coefficient by taking the mean of the estimates for each observation. We used the Gini index of net income inequality, which measures the distribution of household income after government taxes and transfers. A score of 0 indicates that each household receives an equal share of the income in the country, and 1 indicates that all income is received by a single household.
For easier interpretation of interaction effects, we centred this computed Gini coefficient around the mean at the country-year level in the sample of analysis.
For easier interpretation of interaction effects, we centred this variable around the mean at the country-year level in the sample of analysis. To measure the level of clientelism at the contextual level, we used the clientelism index from the Varieties of Democracy V-Dem dataset Footnote 9 Sigman and Lindberg It measures to what extent politics is based on clientelistic relationships.
The index is formed by combining three components. The first component is vote-buying, measuring the extent to which vote buying was common during the most recent national election in the country.
The second component is the extent to which social expenditures in the national budget are particularistic i. The third component is party linkages, which refers to what sort of goods political parties offer in exchange for political support clientelistic vs.
A high score on this clientelism index reflects that politics is to a large extent based on clientelistic relationships. For easier interpretation of interaction effects, we centred the clientelism index around the mean at the country-year level in the sample of analysis. Before we tested our hypotheses, we did some exploratory analyses at the country-year level, in which we plotted the income gap in voting against economic inequality. To formally test our hypotheses, we performed multilevel logistic regression models with three levels of analysis in order to correct the standard errors for the nested structure of our data, with individuals nested in country-years, nested in countries.
Ignoring the multilevel structure in the data would have led to underestimation of the standard errors Steenbergen and Jones As a first step, we performed a model with the main effects of relative income, the individual-level control variables, and economic inequality and national wealth as independent variables.
In the second step, we tested whether the effects of relative income on voting varied with the level of economic inequality, by adding a cross-level interactions between relative income and economic inequality to the previous model. Since we expected that this cross-level interaction effect is negative in the less-wealthy countries and positive in the wealthy countries, we did not have a hypothesis for this interaction in the complete sample of countries.
In the third step, we tested the three-way interaction between relative income, economic inequality and national wealth, to test Hypotheses 1 and 2. Based on the results of these analyses we proceeded to the fourth step in order to test Hypothesis 3. Although we expected to find a negative interaction between economic inequality and relative income on voting only in less-wealthy countries, our results showed that this negative interaction holds in the complete sample and does not vary with levels of national wealth.
We examined whether this can be explained by including clientelism. That is, we examined whether including a cross-level interaction between relative income and clientelism can explain the negative cross-level interaction between relative income and economic inequality. First, we ran a series of logistic regression models to test the effect of relative income on voting in each separate country-year subsample in our data.
We controlled for age, gender and education. Subsequently, we plotted the resulting regression coefficients the log-odds , as an indicator of the income gap in voting, against economic inequality in Fig. However, as we expected, the association seems to depend on levels of national wealth. Second, we explored the bivariate relationships between clientelism on the one hand and inequality, national wealth and the income gap in voting on the other hand.
In the next section we examine if the results of these exploratory analyses hold, by estimating multilevel models with the effect of relative income on voting, and its moderators, including control variables at the contextual level.
In Tables 2 and 3 the results of the multilevel analyses are presented. In Model 2 we added the interaction between relative income and economic inequality. Now that we have analysed the moderating effect of economic inequality, we turn to the question whether the moderating effect of economic inequality varies with levels of national wealth.
We included the quadratic GDP p. Hypothesis 2 is therefore only partly supported, since we expected to find a negative interaction between relative income and economic inequality in less-wealthy countries. However, the results suggest that this pattern is present in the complete sample of country-years. Hypothesis 1, which predicted a positive effect of economic inequality on the income gap in voting in wealthy countries, is thus rejected by these results.
The results of our bivariate analyses, where we found differences between more- and less-wealthy countries, do not hold after employing multilevel logistic regression models with contextual-level control variables. Subsequently, because we found a negative interaction between economic inequality and income in our complete sample, we examined the role of clientelism in our complete sample, instead of only in less-wealthy countries.
First, we tested whether the effect of relative income is weaker at higher levels of clientelism. Second, we examined whether this in turn varies with levels of national wealth, since we initially anticipated that clientelism would be especially prevalent in less-wealthy countries.
Third, we analysed whether clientelism partly explains the negative interaction between relative income and economic inequality. The results of Model 7 show that the interaction between relative income and clientelism cannot completely explain the negative interaction between relative income and economic inequality that we found in Model 2. Figure 2 shows the predicted probabilities of voting for all income quintiles, and the marginal effects of relative income, at different levels of inequality left panel and clientelism right panel based on Model 7.
Predicted probabilities of voting by income quintile, and the marginal effects of relative income on voting, as a function of economic inequality left panel and clientelism right panel based on Model 7. Note Model 7 includes the interaction between relative income and economic inequality as well as the interaction between relative income and clientelism. The 10th percentile in the distribution of economic inequality Gini and clientelism at the country-year level was taken as the low level.
The 50th percentile was taken as the medium level, and the 90th percentile was taken as the high level. The left panel of Fig. This effect is slightly stronger for the higher income groups, and therefore the income gap in voting is smaller when economic inequality is higher.
The right panel of Fig. This is in line with the argument that clientelism negatively affects the sense of efficacy and the likelihood to vote for higher income groups. However, the likelihood to vote for the lower income groups is hardly affected by the level of clientelism. This is in not line with our argument that the lower income groups are mobilized when political parties use clientelistic mobilisation strategies. In sum, the prevalence of clientelism may be one of the mechanisms through which economic inequality negatively affects the income gap in voting, but it does not affect lower income groups in the way we anticipated.
In this study we investigated whether the income gap in voting turnout varies with economic inequality at the country-year level. We aimed to bring together previous studies on economic inequality and the income gap in voting, since the results of these studies contradict each other. We did so by formulating new hypotheses. First, we expected the relationship between economic inequality and the income gap in voting to vary by the national level of wealth.
Based on the relative power theory Solt , we expected a positive association between economic inequality and the income gap in voting in relatively wealthy countries H1. Based on recent studies that included a wider sample of countries Amat and Beramendi ; Matsubayashi and Sakaiya , we expected a negative association between economic inequality and the income gap in voting in less-wealthy countries H2.
Second we hypothesized that clientelism would explain the negative interaction between income and inequality at lower levels of national wealth. Namely, we anticipated that the prevalence of clientelism would increase the likelihood to vote for lower income groups, decrease the likelihood to vote for higher income groups, and therewith weaken the positive effect of income on voting. Our final sample of analysis consists of 68 countries from different continents, and country-years, including , individuals, and thus exceed the most elaborate sample used in previous studies that explicitly tested the relationship between income inequality and voting.
We found that, in general, higher income groups are more likely to vote. The positive effect of income was found to be weaker at higher levels of economic inequality. We found that this pattern does not vary with levels of national wealth, which means that Hypothesis 1 is not supported.
This finding does not support our predictions based on relative power theory. They controlled for a wider variety of country -year characteristics, their data covered a longer time period but a smaller sample of countries, and the individual-level control variables differ from our models.
For example, they included union membership and marital status, but not education. When combining our results with those of earlier studies, it seems that altogether there is inconclusive evidence about whether higher net income inequality is related to a higher income gap in turnout in wealthy countries. Future studies should therefore still try to uncover the presumed mechanism through which economic inequality would widen the income gap in voting more precisely, and the conditions under which this would occur, in order to shed more light on the inconclusive findings in the literature so far.
Low rates of participation may be regarded as a sign of disaffection and a threat to the functioning of democracy Lijphart ; Norris ; but see Rosema ; Lutz and Marsh Moreover, that the income gap in voting does not vary with economic inequality in wealthier countries does not necessarily mean that political inequality remains unaffected. Higher socioeconomic groups participate to a higher rate in other forms of politics as well e.
Similarly, Filetti and Janmaat showed that in Europe, income inequality widens the gap between the rich and the poor for boycotting products and signing petitions, but not for voting. Moreover, political inequality is more than inequality in political participation. It also comprises the extent to which various groups are involved in civil society organizations, are represented in government, are able to set the political agenda, and influence political decisions and the implementations of those decisions Houle ; Schakel When taking such a broad conceptualization of political equality, it was found to be negatively related to economic inequality at the macro level in a global sample of countries Houle Moreover, policy in advanced democracies is more responsive to higher income groups than it is to middle- and low-income groups Schakel Altogether, our results combined with other studies do not imply that economic and political inequality cannot reinforce each other, but it is important to study their interrelationships for various forms of political participation and other forms of power, to get a better insight in how economic inequality affects political inequality, and vice versa, in wealthy democracies.
The negative association between economic inequality and the income gap in voting is in line with Hypothesis 2 for the less-wealthy countries. However, since we found this pattern in the complete sample of countries, we investigated the moderating role of clientelism in the complete sample. These analyses showed that the effect of relative income is weaker when levels of clientelism are higher, which was constant across levels of national wealth.
When clientelism and economic inequality were both included as moderators, both interactions were significant, but the interaction between economic inequality and income was reduced. It suggests that although clientelism might partially explain why economic inequality reduces the income gap in voter turnout, there must also be other mechanisms that explain a negative relationship between economic inequality and political inequality.
Hypothesis 3 was therefore only partly supported by our results. Moreover, the prevalence of clientelism does not exactly work as we argued. Since higher income citizens are not the target of clientelistic strategies, they might perceive political parties that employ such strategies as corrupt and unfair, negatively affecting their sense of political trust and efficacy, and their likelihood to vote.
Following Kasara and Suryanarayan , , higher income groups are more likely to vote when the potential tax exposure of the rich is high. It is likely that the poorer, more unequal countries with high levels of clientelism in our sample are also countries with low bureaucratic capacity Kitschelt and Wilkinson ; Kasara and Suryanarayan This study was limited in certain aspects and therefore future research could try to improve upon our study in several other ways.
First, we only looked at the effect of net income inequality measured with the Gini coefficient. Future studies may compare different measures of economic inequality e. Second, our measure of relative income is not perfectly comparable between surveys, although we created income quintiles for each country-year for each survey.
We explored the extent to which the differences in measurement between surveys affected the estimate for the effect of relative income on voting in Online Appendix 3. These comparisons lead us to cautiously conclude that, overall, differences in the effect of income are not caused by different operationalizations of the income variable between surveys. To control for survey differences in our models, we included dummy variables for the surveys in each model. Notwithstanding these limitations, this study adds some important insights to the literature on economic and political inequality.
Importantly, our results do not imply that economic inequality is positive for democratic representation, since economic inequality was found to depress the likelihood of voting for all income groups, and other studies indicated that economic inequality is associated with political inequality in several other ways than through voting.
Principal Investigator is Prof. Bert Klandermans. Unfortunately, we could not include data from the African Barometer in our analyses, since no comparable measure of relative income is available from this survey. Data file edition 1. Edition 2. Source : Asian Barometer. Asian Barometer Survey Data Release: —, —, —, — Source : Inglehart, R.
Haerpfer, A. Moreno, C. Welzel, K. Kizilova, J. Diez-Medrano, M. Lagos, P. Norris, E. Puranen et al. Version: www. Madrid: JD Systems Institute. ISSP Political scientists have shown conclusively that voting participation, at least in the US, is strongly correlated with income e. Verba et al. Richer people vote more. This form of social inequality is troubling on a number of levels. In practical terms, this pattern appears to have distortionary downstream effects on representative government — reinforcing patterns that bias public policy towards the wealthy Schlozman et al.
Indeed, the most compelling empirical research on this topic tends to show that who participates affects who gets elected and the policies they implement Fowler , Griffin and Newman , Leighley and Nagler , Madestam et al.
Is the income gradient in voter turnout due to a causal link between financial resources and civic participation? Or does income correlate with another characteristic, such as household interest in politics, that is the true driving factor behind the observed correlation?
In the absence of a random shock to income that does not affect other household preferences, these two alternatives are hard to disentangle. In our recent paper, we show that that children who grow up in households that experienced a positive external shock to unearned income due to a quasi-experimental government cash transfer are more likely to vote as adults, even if their parents did not vote Akee et al. Consistent with theory that predicts a diminishing effect of extra income on civic participation as incomes grow, this finding holds only for children from households that were in the bottom half of the income distribution before the unconditional transfers began.
The income transfers were generated from newly opened gaming operations on an American Indian reservation and were distributed to all tribally enrolled members regardless of initial income. We used panel data on tribally enrolled members, their children, and non-American Indians residing in the surrounding counties who were not affected by the transfers over two decades, linked to the voter registration rolls.
Examining the voting patterns of parents and children from the same household, we find evidence of strong correlations between the two. Poorer parents are less likely to vote, and so are their children. We find that the children from the bottom half of the initial income distribution increase their voting by 10 to 20 percentage points as adults compared to their counterparts who did not receive the increased incomes during adolescence.
In Figure 1, we show how the income transfers affect voting behaviour among children from households initially below and above the median income. One very likely interpretation is that the increase in household incomes disrupts the intergenerational transmission of voting behaviour, at least among those who are initially least likely to vote.
We also test whether the parents change their own voting behaviour in response to the increased household incomes. We find no such effects. In Figure 2, we plot the differences in voting probability for the parents over the period before the transfers started prior to and the time after that, separating the observations depending on whether they experienced the increased income or not.
As the figure demonstrates, there is no systematic impact on parental voting across these two groups. This suggests that propensity to vote may be set relatively early in life. Therefore, interventions during adulthood, such as information campaigns and additional income, may have little immediate impact on voter turnout. There is relatively little modern empirical research on these issues, and the mechanisms that bring the effects we describe about are not well understood.
We put forth several hypotheses that may explain the findings.
0コメント