LSE - Small Logo
LSE - Small Logo

Blog Admin

August 7th, 2015

Americans don’t need to agree with elected officials in their districts, they just need someone in government to represent them

3 comments | 1 shares

Estimated reading time: 5 minutes

Blog Admin

August 7th, 2015

Americans don’t need to agree with elected officials in their districts, they just need someone in government to represent them

3 comments | 1 shares

Estimated reading time: 5 minutes

Jeff Harden 80x108Chris Clark 80x108Modern electoral politics typically produces two narratives: competition between candidates in individual districts and competition for overall control of government. Which of these matters more for voters? Is it most important to have their preferred candidate win in their own districts, or would voters accept an electoral loss locally if their political preferences were represented in the legislature as a whole? In new research, Jeff Harden and Chris Clark find strong evidence supporting the latter perspective. It is much more important to Americans that their political preferences are represented by a group of legislators, even if that group does not include their own representative.

A common perspective on political representation suggests that legislators should do all they can to make constituents in their districts happy—from learning their policy preferences and introducing relevant legislation to helping people in their districts who have problems with government agencies. In this view of representation, or what we refer to as a dyadic representation relationship, constituents feel represented based upon the political behavior of a single elected official. In new research, we show evidence that dyadic representation is not the only form of representation that is important to Americans; we find that if forced to choose, Americans would forgo dyadic representation from someone with whom they agree for collective representation, which refers to a person’s interests being represented by a group of legislators who represent other districts.

The empirical evidence that supports our findings comes from two versions of an experiment embedded in surveys of American adults. The experiments presented respondents with short vignettes of text describing hypothetical state legislative election results. We used two different traits known to be politically important—race and party affiliation—to randomly manipulate whether respondents reading the text would be represented collectively, dyadically, both, or neither if they were constituents in the hypothetical district.

For example, we administered the race version of the experiment to a random sample of 871 African American adults. The vignette signaling collective and dyadic representation read as follows:


Maurice Allen won the race in district 15 with 53% of the vote, becoming the first black candidate to win in that district. Additionally, the state as a whole elected the largest contingent of black legislators in the last decade. 20% of the seats in the state legislature will be filled by black representatives when the next session begins.

Compare that with the vignette that signaled little collective and no dyadic representation:


Colin Hunter won the race in district 15 with 53% of the vote, which means the district will go at least another four years without ever being represented by a black legislator. Additionally, the state as a whole elected the smallest contingent of black legislators in the last decade. Only 5% of the seats in the state legislature will be filled by black representatives when the next session begins.

The other two vignettes interchanged text from these examples to signal collective representation but no dyadic representation and dyadic representation but little collective representation. The party version of the experiment, which we administered to a random sample of 1,000 American adults, contained similar wording. After they read one vignette at random, respondents rated whether they would feel represented in government based on the election results. Respondents gave their answers on a 0-100 scale, with 0 corresponding to “not represented” and 100 corresponding to “represented.”

Figure 1 below shows the results of both experiments. The bars signify the average ratings from respondents who viewed each vignette: (A) collective/dyadic, (B) collective/no dyadic, (C) little collective/dyadic, and (D) little collective/no dyadic. The black brackets indicate the margin of error associated with each average.

Figure 1- Average values of the representation scale in each experiment

Harden Fig 1

The results show that people have a strong preference for collective over dyadic representation. In both versions of the experiment, the collective representation vignettes (A and B) produced higher ratings than the vignettes with little collective representation (C and D). In the race version, ratings of the collective representation vignettes were, on average, 21 points larger than those from the two vignettes with little collective representation. That same number jumps to an average difference of 30 points in the party version of the experiment, corresponding to nearly one third of the scale.

Examining vignettes B and C within each experiment further supports this finding. In those two vignettes, respondents only “received” one type of representation (collective in B, dyadic in C). Thus, comparing responses to those two vignettes is like asking the question “if you could only have one type of representation, which would you choose?” In both versions, vignette B produced an average rating about 11 points larger than vignette C. People clearly prefer a group of legislators who represent their interests over just one like-minded representative in their districts.

At first glance this finding might seem fairly obvious; a group of many legislators representing a voter rather than just one representative seems like it should always be better. However, we maintain that these results are still somewhat surprising for a few reasons. First, by preferring collective representation over dyadic, respondents in our samples showed willingness to give up some political power. With dyadic representation, people have the ability to affect who represents them through their vote choice, and they can punish those who are out of step with their interests. However, people have far less ability to influence the overall makeup of a legislative body (i.e., who wins in several other districts). Additionally, a great deal of research shows that the connections that legislators make with people in their districts are important to constituents (for recent examples of this work, see here and here). Thus, it is surprising that collective representation—a form of representation where people do not personally know each legislator—was that much strongly preferred over dyadic.

There are many important implications of these findings for American politics. First, they are relevant to discussions about the utility of majority-minority districts. On one hand, majority-minority districts are important because they virtually guarantee representation of black interests, at least in some districts. However, our results show that blacks value black legislators even if they do not reside in those legislators’ districts. Thus, majority-minority districts are not just important for providing black citizens with a black representative in their district; they are also important for representing any and all blacks in the electorate (for more on this point, see here).

On the other hand, research also shows that packing many racial minorities into a single district may result in blacks failing to receive policy representation because it often means more Republicans are being elected in surrounding districts, increasing that party’s strength (see, for example, here and here). Thus, majority-minority districts allow for some black dyadic representation, but may actually depress black collective representation.

Our research also has implications for discussions about district magnitude, or the number of representatives in a district. In the United States, many (though not all) state legislative districts and all districts in the U.S. House of Representatives have a single member. By definition, multimember districts allow for a greater number of people to represent a constituent, which in turn provides a better chance that legislators will possess different traits that appeal to their constituency. We hope that our findings may spur discussions about the usefulness of increasing the number of multimember districts in the United States.

Finally, in showing that collective representation is valued over dyadic, this work also lends some support to the notion that dividing constituencies based on geographic communities may be problematic. As Andrew Rehfeld contends, removing territorial boundaries may produce other benefits to the quality of representation, such as “forcing representatives to justify themselves before a heterogeneous population” (p. xii). Our results do not directly support or refute that assertion, but they do suggest that collective representation is much more vital to Americans than political scientists previously thought.

This article is based on the paper, ‘A Legislature or a Legislator Like Me? Citizen Demand for Collective and Dyadic Political Representation’, in American Politics Research. This article is available toll-free until September 30, thanks to Sage publications.

Featured image credit: the justified sinner (Flickr, CC-BY-NC-SA-2.0)

Please read our comments policy before commenting.          

Note:  This article gives the views of the author, and not the position of USAPP– American Politics and Policy, nor of the London School of Economics.

Shortened URL for this post:


About the authors

Jeff Harden 80x108Jeff HardenUniversity of Colorado Boulder
Jeff Harden is an assistant professor in the Department of Political Science at the University of Colorado Boulder. His research focuses on political representation, American state politics, and quantitative methodology. See his faculty webpage for more information.


Chris Clark 80x108Chris ClarkUniversity of North Carolina at Chapel Hill
Chris Clark is an assistant professor in the Department of Political Science at the University of North Carolina at Chapel Hill. His research focuses on race and electoral representation in the United States, with a particular focus on black representation in state legislatures. See his faculty webpage for more information.

About the author

Blog Admin

Posted In: Chris Clark | Elections and party politics across the US | Jeff Harden


LSE Review of Books Visit our sister blog: British Politics and Policy at LSE

RSS Latest LSE Events podcasts

This work by LSE USAPP blog is licensed under a Creative Commons Attribution-NonCommercial 3.0 Unported.