No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Estimating subnational preferences across the European Union
Abstract
Subnational analyses of political preferences are substantively relevant and offer advantages for causal inference. Yet, our knowledge on regional political preferences across Europe is limited, not least because there is a lack of adequate data. The rich Eurobarometer (EB) data is a promising source for European-wide regional information. Yet, it is only representative for the national level. This paper compares state-of-the-art methods for estimating regional preferences from nationally representative EB data, validating predictions with regionally representative surveys. Our analysis highlights a number of challenges for estimating regional preferences across Europe, such as data availability, variable selection, and over-fitting. We find that predictions are best using a Bayesian additive regression tree with synthetic post-stratification.
... As the ESS survey is nationally but not regionally representative, ESS responses cannot be aggregated to the regional level to obtain accurate measures of regional religion. Instead, we rely on the machine learning algorithm and post-stratification strategy used in Lipps and Schraff (2021) to predict the Catholic and Protestant populations in each NUTS II region. 8 While this strategy only yields religion and culture measures for 104 regions across eight countries, 9 this more fine-grained measure allows us to test whether sub-national culture affects individual preferences when national culture may not. ...
... Next, we replace the national religious culture measures with the regional measures predicted by the machine-learning method from Lipps and Schraff (2021). 10 Catholic culture at the regional level has a more significant effect on debt preferences, but again indicates that respondents in regions with a larger Catholic community believe that reducing the public deficit and debt is a priority. ...
Popular media and politicians have often blamed the high public debt of some EU countries on cultural differences. These claims are most apparent in the discourse contrasting ostensibly prudent Northern Europeans with spendthrift Southern Europeans. Despite the prominence of these and similar narratives and evidence that culture plays a nontrivial role in other economic outcomes, there is no systematic evidence that culture influences attitudes towards sovereign debt in the EU. We provide the first empirical test of this claim using over 233,000 responses to a Eurobarometer question about the salience of national debt. Our analysis reveals that national and sub-national differences explain very little of the variance in debt preferences. Further, the differences that do emerge do not fit existing cultural narratives. Additional analysis reveals that established measures of national culture or religious observance, at the national and regional levels, do not correlate with debt attitudes as cultural arguments would predict.
... We improve the generalizability of the survey responses using MRP to reweight the responses so that they better represent the full population of previously successful noncompleters in our partner colleges. While MRP has been used extensively in political science to measure public opinion in the United States (Gao et al., 2019;Gelman et al., 2010;Gelman & Little, 1997;Howe et al., 2015;Kastellec et al., 2019;Lax & Phillips, 2009;Lei et al., 2017;Lipps & Schraff, 2019;Little, 1993;Pacheco, 2011;Park et al., 2004;Wang et al., 2015;Warshaw & Rodden, 2012), and is increasingly used by political scientists outside of the United States (Lipps & Schraff, 2019;Toshkov, 2015), sociologists (Fairbrother & Martin, 2013), and epidemiologists (Downes et al., 2018;Eke et al., 2016;Zhang et al., 2014) to generate representative estimates from nonrepresentative data, it has been seldom used in surveybased educational research. 3 Following our general description above, MRP is implemented using this two-step process: ...
... We improve the generalizability of the survey responses using MRP to reweight the responses so that they better represent the full population of previously successful noncompleters in our partner colleges. While MRP has been used extensively in political science to measure public opinion in the United States (Gao et al., 2019;Gelman et al., 2010;Gelman & Little, 1997;Howe et al., 2015;Kastellec et al., 2019;Lax & Phillips, 2009;Lei et al., 2017;Lipps & Schraff, 2019;Little, 1993;Pacheco, 2011;Park et al., 2004;Wang et al., 2015;Warshaw & Rodden, 2012), and is increasingly used by political scientists outside of the United States (Lipps & Schraff, 2019;Toshkov, 2015), sociologists (Fairbrother & Martin, 2013), and epidemiologists (Downes et al., 2018;Eke et al., 2016;Zhang et al., 2014) to generate representative estimates from nonrepresentative data, it has been seldom used in surveybased educational research. 3 Following our general description above, MRP is implemented using this two-step process: ...
Even though a postsecondary degree can offer economic, social, and civic benefits, many community college students leave without earning a degree—including some who have performed well academically and made substantial progress toward graduation. To better understand the factors contributing to early exit, we surveyed a number of former students in a large community college system. We improve the generalizability of the survey responses through multilevel regression with poststratification, which we use to reweight the responses to better represent the population in our original survey frame. We find that tuition and fees, living expenses, and no longer being eligible for financial aid are the factors contributing to early exit for the largest share of students. We also find variation in both financial and nonfinancial factors across subgroups, suggesting that targeted supports may be useful in helping students persist or return to college and complete their degree.
... Our four-wave sample has a median number of 511 observations per NUTS2 region. This allows us to avoid many of the issues regarding representativeness and precision on the regional level that usually plague European comparative surveys (Lipps & Schraff 2019). The four-wave design makes our analysis a longitudinal study that can investigate change on the regional and national level. ...
Inequality is a central explanation of political distrust in democracies, but has so far rarely been considered a cause of (dis‐)trust towards supranational governance. Moreover, while political scientists have extensively engaged with income inequality, other salient forms of inequality, such as the regional wealth distribution, have been sidelined. These issues point to a more general shortcoming in the literature. Determinants of trust in national and European institutions are often theorized independently, even though empirical studies have demonstrated large interdependence in citizens’ evaluations of national and supranational governance levels. In this paper, we argue that inequality has two salient dimensions: 1) income inequality and 2) regional inequality. Both dimensions are important antecedent causes of EU trust, the effects of which are mediated by evaluations of national institutions. On the micro‐level, we suggest that inequality decreases a person's trust in national institutions and thereby diminishes the positive effect of national trust on EU trust. On the macro‐level, inequality decreases country averages of trust in national institutions. This, however, informs an individual's trust in the EU positively, compensating for the seemingly untrustworthiness of national institutions. Finally, we propose that residing in an economically declining region can depress institutional trust. We find empirical support for our arguments by analysing regional temporal change over four waves of the European Social Survey 2010–2016 with a sample of 209 regions nested in 24 EU member states. We show that changes in a member state's regional inequality have similarly strong effects on trust as changes in the Gini coefficient of income inequality. Applying causal mediation techniques, we can show that the effects of inequality on EU trust are largely mediated through citizens’ evaluations of national institutions. In contrast, residing in an economically declining region directly depresses EU trust, with economically lagging areas turning their back on European governance and resorting to the national level instead. Our findings highlight the relevance of regional inequality for refining our understanding of citizens’ support for Europe's multi‐level governance system and the advantages of causal modelling for the analysis of political preferences in a multi‐level governance system.
Existing research mainly analyzes mass attitudes towards the European Union (EU) from the national and individual‐level perspective. This paper adds to this literature by focusing on the relationship between EU support and subnational economic conditions, using harmonized survey data covering 40 years and 1.1 million respondents in 197 European regions. We first describe Europe's changing subnational conditions in terms of catch‐up, wealthy, declining, and glass‐ceiling regions. The paper then develops and tests a set of hypotheses regarding the temporally dynamic relationship between EU attitudes and regions’ long‐ and short‐term economic conditions. Our analyses reveal important longitudinal variations in this relationship with low levels of geographic differentiation in public opinion giving way to clear spatial differences in recent years. Our findings are consistent with the idea that the Great Recession and Brexit have generated a new geography of both euroscepticism in Europe's declining regions and EU support in its wealthy and catch‐up regions. This article is protected by copyright. All rights reserved
Using a new regional database of national and European parliament elections on NUTS 2 level in 28 countries, we test the main theories explaining the electoral support for the European far right. Accounting for differences between the extremist (ER) and populist radical right (PRR), we find evidence in support of both economic insecurity and cultural backlash theses. The ER vote is associated mostly with economic insecurity and the PRR vote mostly with cultural backlash. Whereas micro and macro-level analyses have often produced conflicting results, unemployment, immigration and income inequalities have significant and robust effects at the meso level, indicating that the factors determining the far right vote might at large be operating at a sub-national level. In line with the " contact " and " salience-of-change " hypotheses, the effects of economic insecurity are more pronounced in regions that undergo sudden changes compared to those with high levels of immigration.
Europeans’ confidence in political institutions has dropped precipitously since the onset of the Euro-crisis in 2009. The decline in trust in government varies across countries and occupational and educational groups. Economic factors explain much of the cross-national and over-time variation. The baseline level of trust is influenced by a person’s position in the labor market: across European countries, citizens with more education and higher levels of skills trust government more than those educational and occupational groups that have benefited less from European integration. Residents of debtor countries with high unemployment rates are also much less likely to trust national government than those in creditor countries that have fared better during the economic crisis, while the unemployed have lost faith in government to a greater degree than other parts of the population. Cultural, ideational, and political factors remain important for baseline levels of trust, but cannot explain the acute, asymmetrical decline in citizen trust observed over the last decade.
Political scientists interested in estimating how public opinion varies by constituency have developed several strategies for supplementing limited constituency survey data with additional sources of information. We present two evaluation studies in the previously unexamined context of British constituency-level opinion: an external validation study of party vote share in the 2010 general election and a cross-validation of opinion toward the European Union. We find that most of the gains over direct estimation come from the inclusion of constituency-level predictors, which are also the easiest source of additional information to incorporate. Individual-level predictors combined with post-stratification particularly improve estimates from unrepresentative samples, and geographic local smoothing can compensate for weak constituency-level predictors. We argue that these findings are likely to be representative of applications of these methods where the number of constituencies is large.
We develop a Bayesian “sum-of-trees” model where each tree is
constrained by a regularization prior to be a weak learner, and
fitting and inference are accomplished via an iterative Bayesian
backfitting MCMC algorithm that generates samples from a
posterior. Effectively, BART is a nonparametric Bayesian
regression approach which uses dimensionally adaptive random
basis elements. Motivated by ensemble methods in general, and
boosting algorithms in particular, BART is defined by a
statistical model: a prior and a likelihood. This approach
enables full posterior inference including point and interval
estimates of the unknown regression function as well as the
marginal effects of potential predictors. By keeping track of
predictor inclusion frequencies, BART can also be used for
model-free variable selection. BART’s many features are
illustrated with a bake-off against competing methods on 42
different data sets, with a simulation experiment and on a drug
discovery classification problem.
Does trust in national institutions foster or hinder trust in the institutions of the European Union (EU)? There is no agreement in the literature on popular support for the EU about the direction of the relationship between trust in national and European institutions. Some scholars argue that both will be positively related, others have proposed the opposite hypothesis: low levels of trust in national institutions will lead citizens to higher levels of support for the EU. We argue that both hypotheses are true but operate at different levels: whereas more trusting citizens tend to be so in both the national and the European arenas, we also find that at the country level the relationship is negative: living in a country with highly trusted and well-performing institutions hinders trust in the European Parliament. We test our hypotheses using data from the European Social Survey and Hierarchical Linear Modeling.
This paper investigates the role of economic variables in predicting regional disparities in reported life satisfaction of
European Union (EU) citizens. European subnational units (regions) are defined according to the first-level EU nomenclature
of territorial units. We use multilevel modeling to explicitly account for the hierarchical nature of our data, respondents
within regions and countries, and for understanding patterns of variation within and between regions. Main findings are that
personal income matters more in poor regions than in rich regions, a pattern that still holds for regions within the same
country. Being unemployed is negatively associated with life satisfaction even after controlled for income variation. Living
in high unemployment regions does not alleviate the unhappiness of being out of work. After controlling for individual characteristics
and modeling interactions, regional differences in life satisfaction still remain, confirming that regional dimension is relevant
for life satisfaction.
KeywordsLife satisfaction-Regional disparities-Multilevel models
There has been considerable recent debate about the importance of local context as an influence on political attitudes and voting behaviour in Great Britain. Resolution of that debate has been difficult, because analytical methods have not been available with which to evaluate the relative importance of both individual voter characteristics and the characteristics of their milieux as independent correlates of attitudes and behaviour. The technique of multi-level modelling has been developed by educational researchers to do just that. It is introduced here and illustrated using data for the 1987 British general election. The preliminary results suggest that place clearly does matter as a component of the processes that influence voters' choices.
Political scientists often find themselves analyzing data sets with a large number of observations, a large number of variables, or both. Yet, traditional statistical techniques fail to take full advantage of the opportunities inherent in “big data,” as they are too rigid to recover nonlinearities and do not facilitate the easy exploration of interactions in high‐dimensional data sets. In this article, we introduce a family of tree‐based nonparametric techniques that may, in some circumstances, be more appropriate than traditional methods for confronting these data challenges. In particular, tree models are very effective for detecting nonlinearities and interactions, even in data sets with many (potentially irrelevant) covariates. We introduce the basic logic of tree‐based models, provide an overview of the most prominent methods in the literature, and conduct three analyses that illustrate how the methods can be implemented while highlighting both their advantages and limitations.
Anticipating the competitive disadvantage of economically weak regions in an integrated European single market, the European Union (EU) redistributes money to alleviate economic inequalities and increase cohesion. However, the amount of European redistribution is very moderate and the recent years have shown that Eurosceptic parties gain ground, especially in economically weak areas. So is Eurosceptic voting related to an insufficient compensation of the losers of EU integration? Combining European Social Survey data with information on regional funding for 123 EU regions, I demonstrate that the probability of a Eurosceptic vote is highest under insufficient compensation. Insufficient compensation occurs among middle income regions that are cut-off from the bulk of funding due to the regional policies’ targeted approach. Moreover, some of the poorest regions miss out as well, as the more developed areas among the poor are favored in funds allocation. A taming effect of funding on Eurosceptic voting is therefore restricted to the more prosperous regions in Europe’s lagging areas.
The comparative study of subnational units is on the rise. Multilevel regression and poststratification (MrP) has become the standard method for estimating subnational public opinion. Unfortunately, MrP comes with stringent data demands. As a consequence, scholars cannot apply MrP in countries without detailed census data, and when such data are available, the modeling is restricted to a few variables. This article introduces multilevel regression with synthetic poststratification (MrsP), which relaxes the data requirement of MrP to marginal distributions, substantially increases the prediction precision of the method, and extends its use to countries without census data. The findings of Monte Carlo, U.S., and Swiss analyses show that, using the same predictors, MrsP usually performs in standard applications as well as the currently used standard approach, and it is superior when additional predictors are modeled. The better performance and the more straightforward implementation promise that MrsP will further stimulate subnational research.
This book addresses two questions - why some political systems have more centralized systems of interpersonal redistribution than others, and why some political unions make larger efforts to equalize resources among their constituent units than others. This book presents a new theory of the origin of fiscal structures in systems with several levels of government. The argument points to two major factors to account for the variation in redistribution: the interplay between economic geography and political representation on the one hand, and the scope of interregional economic externalities on the other. To test the empirical implications derived from the argument, the book relies on in-depth studies of the choice of fiscal structures in unions as diverse as the European Union, Canada and the United States in the aftermath of the Great Depression; Germany before and after Reunification; and Spain after the transition to democracy.
This text reports the results of an evaluation of the performance of multilevel regression modeling and poststratication (MRP)
in reconstructing state-level estimates from federal-level data. The evaluation makes use of Eurobarometer data and relies
on the fact that Eurobarometer provides representative survey data for each European Union state to further explore the performance
of MRP. I repeatedly draw subsets of the entire Eurobarometer sample, then I compute adjusted country means using MRP with
census data, and I compare the resulting estimates to the true country means from the full sample. I do that for ten survey
items from various Eurobarometer waves. The results show that MRP is generally successful in producing estimates that are
highly correlated with the true values (mean of 0.90). But the approach is less capable of reconstructing the relative rankings
of the country means and hitting the range of plausible values of the individual state means. I also show that the great part
of the adjustment comes from the modeling of the state means and not from poststratification, and that population-weighted
samples perform no worse than samples in which countries have equal shares of the pool of respondents.
Multilevel regression and poststratification (MRP) is a method to estimate public opinion across geographic units from individual-level
survey data. If it works with samples the size of typical national surveys, then MRP offers the possibility of analyzing many
political phenomena previously believed to be outside the bounds of systematic empirical inquiry. Initial investigations of
its performance with conventional national samples produce generally optimistic assessments. This article examines a larger
number of cases and a greater range of opinions than in previous studies and finds substantial variation in MRP performance.
Through empirical and Monte Carlo analyses, we develop an explanation for this variation. The findings suggest that the conditions
necessary for MRP to perform well will not always be met. Thus, we draw a less optimistic conclusion than previous studies
do regarding the use of MRP with samples of the size found in typical national surveys.
Using multilevel regression and poststratification (MRP), we estimate voter turnout and vote choice within deeply interacted subgroups: subsets of the population that are defined by multiple demographic and geographic characteristics. This article lays out the models and statistical procedures we use, along with the steps required to fit the model for the 2004 and 2008 presidential elections. Though MRP is an increasingly popular method, we improve upon it in numerous ways: deeper levels of covariate interaction, allowing for nonlinearity and nonmonotonicity, accounting for unequal inclusion probabilities that are conveyed in survey weights, postestimation adjustments to turnout and voting levels, and informative multidimensional graphical displays as a form of model checking. We use a series of examples to demonstrate the flexibility of our method, including an illustration of turnout and vote choice as subgroups become increasingly detailed, and an analysis of both vote choice changes and turnout changes from 2004 to 2008.
Due to insufficient sample sizes in national surveys, strikingly little is known about public opinion at the level of Congressional and state legislative districts in the United States. As a result, there has been virtually no study of whether legislators accurately represent the will of their constituents on individual issues. This article solves this problem by developing a multilevel regression and poststratification (MRP) model that combines survey and census data to estimate public opinion at the district level. We show that MRP estimates are excellent predictors of public opinion and referenda results for both congressional and state senate districts. Moreover, they have less error, higher correlations, and lower variance than either disaggregated survey estimates or presidential vote shares. The MRP approach provides American and Comparative Politics scholars with a valuable new tool to measure issue-specific public opinion at low levels of geographic aggregation.
The paper discusses quality monitoring and assessment in quantitative survey research from a cross-national perspective. It takes standards of best practice advocated in national survey research as a starting point from which to discuss cross-national research quality and comparability. It illustrates how the lack of adequate documentation at each stage of crossnational research seriously hampers monitoring and evaluation of project quality and comparability across countries. It outlines different kinds of information needed to begin to monitor and evaluate quality properly in cross-national survey research and points towards developments in cross-cultural survey methods which can follow on from such information becoming publicly available.
In order to address classic questions about democratic representation in countries with winner-take-all electoral districts, it is necessary to understand the distribution of political preferences across districts. Recent formal theory literature has contributed new insights into how parties choose platforms in countries with a continuum of heterogeneous districts. Meanwhile, increases in survey sample sizes and advances in empirical techniques have made it possible to characterize the distribution of preferences within and across electoral districts. This review addresses an emerging literature that builds on these new tools to explore the ways in which the geography of political preferences can help explain the parties that compete, the platforms and policies they choose, and even the rules under which they compete. Building on insights from economic and political geography, it pays special attention to electoral and policy biases that can emerge when there is an asymmetric distribution of preferences across districts.
We compare two approaches for estimating state-level public opinion: disaggregation by state of national surveys and a simulation approach using multilevel modeling of individual opinion and poststratification by population share. We present the first systematic assessment of the predictive accuracy of each and give practical advice about when and how each method should be used. To do so, we use an original data set of over 100 surveys on gay rights issues as well as 1988 presidential election data. Under optimal conditions, both methods work well, but multilevel modeling performs better generally. Compared to baseline opinion measures, it yields smaller errors, higher correlations, and more reliable estimates. Multilevel modeling is clearly superior when samples are smaller—indeed, one can accurately estimate state opinion using only a single large national survey. This greatly expands the scope of issues for which researchers can study subnational opinion directly or as an influence on policymaking.
BART: Bayesian additive regression trees
- H A Chipman
- George Mcculloch
- Re