"What Politicians Believe About Their Constituents: Asymmetric Misperceptions and Prospects for Constituency Control"
A nifty paper, but it seems badly marred by a failure to ask or test the most obvious explanation for persistent incongruency: that their assessments do not reflect their 'constituents' but does reflect the voters. A half-hearted gesture in that direction at the very last paragraph does not make up for this sustained silence & failure, especially when their existing dataset seems like it could check at least one possibility: whether greater levels of incongruency correlate with greater risk of not being elected (if it doesn't, that's evidence that the views are congruent with voters).
"We reexamine prospects for constituency control in American politics with original data describing nearly 2,000 state legislative candidates’ perceptions of mass opinion in their districts and recent advances in public opinion estimation that allow us to determine actual district-level opinion with precision. Actual district opinion explains only a modest share of the variation in politicians’ perceptions of their districts’ views. Moreover, there is a striking conservative bias in politicians’ perceptions, particularly among conservatives: conservative politicians systematically believe their constituents are more conservative than they actually are by over 20%, while liberal politicians also typically overestimate their constituents’ conservatism by several percentage points. A follow-up survey demonstrates that politicians appear to learn nothing from democratic campaigns or elections that leads them to correct these shortcomings. Electoral selection has a limited impact on whether the chosen representative is congruent with the majority of her constituents...For perspective, 20 percentage points is roughly the difference in partisanship between California and Alabama. Most politicians appear to believe they are representing constituents who are considerably different than their actual constituents.
...We first investigate the general relationship between elite perception of public opinion and actual public opinion and show that it is remarkably linear on average – an increase in support for same-sex marriage of 10 percentage points in a district is associated with an equally-sized 10 percentage point increase in politicians’ perceptions of their districts’ opinions on average. The same holds true for districts’ support for universal healthcare and politicians’ average perceptions of this support. Yet although politicians’ perceptions do vary directly with their constituencies’ actual views on average, the correlations between public opinion and politicians’ perceptions of it are at best modest – 0.43 for universal health care and 0.51 for same-sex marriage.
...We exploit the timing of our survey, August 2012 – prior to the most intense period of the fall election campaign, and prior to when the first election itself had occurred in newly drawn legislative districts – to examine whether politicians gain information from the democratic process that allows them to correct these misperceptions. We find little evidence that politicians do learn from the democratic process. After the election campaign and the November election itself, we re-contacted politicians to appraise whether their perceptions of their constituents had grown more accurate. Strikingly, politicians’ perceptions of public opinion after the campaign and the election itself look identical to their perceptions prior to these events, with little evidence that their misperceptions had been corrected.
...With the rise of public polling in politics and arithmetic education we believe nearly all of our respondents – the vast majority of whom have college educations – would have no difficulty expressing their perceptions in percentage terms. Butler (2013) also shows that state legislators are exceptionally accurate when it comes to describing district demographic information in percentage terms, suggesting that our results are unlikely to contain a great deal of measurement error due to elite innumeracy.
...Put differently, for the typical politician to believe that these policies command majority support, it appears that public support would need to pass a threshold of close to 60%. In this regard our findings echo surveys of other political elites suggesting that elites systematically overestimate the conservatism of the mass public (McGarrell and Sandys 1996; Kohut, Palmer, Sonner, Flemming and Donovan 1998; Kull and Ramsay 2002) and have important implications we will consider shortly.
...Lest one be tempted to attribute these findings to the particular issues we have examined, Figure 6 shows the distribution of support for the third and final question on which we asked politicians to estimate their constituents’ opinions – the statement “Abolish all federal welfare programs.” Although we do not have a sufficient sample size to reach sufficiently accurate district-by-district estimates, the Figure shows that the same striking conservative bias exists on this question. Very few Americans agree with this extreme statement – a complete dismantling of the entire American welfare state – yet American politicians believe that support for this sentiment is extremely widespread. Indeed, even liberal politicians typically believe that support is well above 25% in their districts, while conservative politicians typically perceive opinion on this issue closer to 40%. The true national mean is 13%, considerably more liberal than politicians of all stripes believe. Conservative politicians again overestimate support for conservative positions by about 25 percentage points on this issue, while liberals substantially overestimate support for conservative positions as well.
...Yet inferring congruence from the strength of the relationship in the tails of the distribution is misleading (e.g. Achen 1977) when one considers that in the preponderance of districts – where voters support these policies at rates of between 50% and 60% – congruent candidates have at most a 60% chance of winning while incongruent candidates have at most a 40% chance of defeat. Constituencies’ preferences do certainly express themselves in the electoral process to some extent, but in accounting for which policies their representatives will espouse they represent only a small part of the story in places where collective opinion is short of unambiguous
...Factors beyond constituency opinion appear to account for most of the variation in what representatives do on behalf of their constituencies. Be these additional influences on elite behavior national party platforms (e.g. Ansolabehere, Snyder and Stewart 2001), primary voters (e.g. Burden 2001), or narrow interest groups (e.g. Bawn et al. 2012), on the basis of our investigation it seems alternative dynamics such as these are the ones principally worthy of investigation for understanding how representatives make public policy in the United States."
A nifty paper, but it seems badly marred by a failure to ask or test the most obvious explanation for persistent incongruency: that their assessments do not reflect their 'constituents' but does reflect the voters. A half-hearted gesture in that direction at the very last paragraph does not make up for this sustained silence & failure, especially when their existing dataset seems like it could check at least one possibility: whether greater levels of incongruency correlate with greater risk of not being elected (if it doesn't, that's evidence that the views are congruent with voters).
"We reexamine prospects for constituency control in American politics with original data describing nearly 2,000 state legislative candidates’ perceptions of mass opinion in their districts and recent advances in public opinion estimation that allow us to determine actual district-level opinion with precision. Actual district opinion explains only a modest share of the variation in politicians’ perceptions of their districts’ views. Moreover, there is a striking conservative bias in politicians’ perceptions, particularly among conservatives: conservative politicians systematically believe their constituents are more conservative than they actually are by over 20%, while liberal politicians also typically overestimate their constituents’ conservatism by several percentage points. A follow-up survey demonstrates that politicians appear to learn nothing from democratic campaigns or elections that leads them to correct these shortcomings. Electoral selection has a limited impact on whether the chosen representative is congruent with the majority of her constituents...For perspective, 20 percentage points is roughly the difference in partisanship between California and Alabama. Most politicians appear to believe they are representing constituents who are considerably different than their actual constituents.
...We first investigate the general relationship between elite perception of public opinion and actual public opinion and show that it is remarkably linear on average – an increase in support for same-sex marriage of 10 percentage points in a district is associated with an equally-sized 10 percentage point increase in politicians’ perceptions of their districts’ opinions on average. The same holds true for districts’ support for universal healthcare and politicians’ average perceptions of this support. Yet although politicians’ perceptions do vary directly with their constituencies’ actual views on average, the correlations between public opinion and politicians’ perceptions of it are at best modest – 0.43 for universal health care and 0.51 for same-sex marriage.
...We exploit the timing of our survey, August 2012 – prior to the most intense period of the fall election campaign, and prior to when the first election itself had occurred in newly drawn legislative districts – to examine whether politicians gain information from the democratic process that allows them to correct these misperceptions. We find little evidence that politicians do learn from the democratic process. After the election campaign and the November election itself, we re-contacted politicians to appraise whether their perceptions of their constituents had grown more accurate. Strikingly, politicians’ perceptions of public opinion after the campaign and the election itself look identical to their perceptions prior to these events, with little evidence that their misperceptions had been corrected.
...With the rise of public polling in politics and arithmetic education we believe nearly all of our respondents – the vast majority of whom have college educations – would have no difficulty expressing their perceptions in percentage terms. Butler (2013) also shows that state legislators are exceptionally accurate when it comes to describing district demographic information in percentage terms, suggesting that our results are unlikely to contain a great deal of measurement error due to elite innumeracy.
...Put differently, for the typical politician to believe that these policies command majority support, it appears that public support would need to pass a threshold of close to 60%. In this regard our findings echo surveys of other political elites suggesting that elites systematically overestimate the conservatism of the mass public (McGarrell and Sandys 1996; Kohut, Palmer, Sonner, Flemming and Donovan 1998; Kull and Ramsay 2002) and have important implications we will consider shortly.
...Lest one be tempted to attribute these findings to the particular issues we have examined, Figure 6 shows the distribution of support for the third and final question on which we asked politicians to estimate their constituents’ opinions – the statement “Abolish all federal welfare programs.” Although we do not have a sufficient sample size to reach sufficiently accurate district-by-district estimates, the Figure shows that the same striking conservative bias exists on this question. Very few Americans agree with this extreme statement – a complete dismantling of the entire American welfare state – yet American politicians believe that support for this sentiment is extremely widespread. Indeed, even liberal politicians typically believe that support is well above 25% in their districts, while conservative politicians typically perceive opinion on this issue closer to 40%. The true national mean is 13%, considerably more liberal than politicians of all stripes believe. Conservative politicians again overestimate support for conservative positions by about 25 percentage points on this issue, while liberals substantially overestimate support for conservative positions as well.
...Yet inferring congruence from the strength of the relationship in the tails of the distribution is misleading (e.g. Achen 1977) when one considers that in the preponderance of districts – where voters support these policies at rates of between 50% and 60% – congruent candidates have at most a 60% chance of winning while incongruent candidates have at most a 40% chance of defeat. Constituencies’ preferences do certainly express themselves in the electoral process to some extent, but in accounting for which policies their representatives will espouse they represent only a small part of the story in places where collective opinion is short of unambiguous
...Factors beyond constituency opinion appear to account for most of the variation in what representatives do on behalf of their constituencies. Be these additional influences on elite behavior national party platforms (e.g. Ansolabehere, Snyder and Stewart 2001), primary voters (e.g. Burden 2001), or narrow interest groups (e.g. Bawn et al. 2012), on the basis of our investigation it seems alternative dynamics such as these are the ones principally worthy of investigation for understanding how representatives make public policy in the United States."