Extremity in Congress: Communications versus Votes

• Published date: 01 August 2016
• Date: 01 August 2016
• Author: Lindsey Cormack
• DOI: http://doi.org/10.1111/lsq.12126

LINDSEY CORMACK

Stevens Institute of Technology

Extremity in Congress:

Communications versus Votes

I propose a theory of legislator-to-constituent communication that describes a

relationship between the types of votes a legislator reveals and the partisan composition

of her constituency. To test this theory, I use an original data set of 40,000 official

communications containing 30,000 vote revelations from the 111th Congress. I find

evidence substantiating this theory; the extent to which a legislator endeavors to appear

more ideologically extreme in communications varies systematically with the relative

amounts of different types of voters in her district. This result is contrasted with an

analysis of voting extremism where I find that the ideological preferences of donors

better explain voting patterns.

If all we knew about a legislator was what she told us, would our

understanding of her ideology differ from what we would learn from

relying on direct observation of voting behavior? Do different factors

correlate with how legislators vote versus how legislators decide to selec-

tively communicate some of those votes to constituents? The answers to

these questions are important for our understanding of democratic

accountability as strategic communication techniques may alter voters’

ability to assess the ideological positions of their representatives. To bet-

ter hold legislators accountable, voters need an accurate understanding of

legislator ideology. If legislators attempt to shift perceptions of their

ideology by selectively revealing votes, voters may end up falsely feeling

more informed, even though this information may be strategically

skewed.

In this article, I present a theory of strategic vote revelation that

yields a testable hypothesis about how differently situated legislators

present their votes. Put simply, when deciding whether to reveal a vote

that splits a legislator’s constituents, a legislator faces a choice of whom

to alienate. A long line of research, starting with Cox and McCubbins

(1986), indicates that the relevant and most likely voters for each legisla-

tor can be divided into two camps: base and swing voters. Given this

division, I hypothesize those legislators who anticipate a higher marginal

LEGISLATIVE STUDIES QUARTERLY, 41, 3, August 2016 575

DOI: 10.1111/lsq.12126

V

C2016 Washington University in St. Louis

risk of vote abstention or defection associated with alienating base voters

will tend to omit votes in which they more likely sided with the preferen-

ces of swing voters and vice versa. The aggregate effect of this self-

censoring is for legislators to present an ideological version of self in

communications that may differ systematically from that suggested by

their voting behavior. I situate my theory of strategic communication

against literature analyzing ties between f‌inancial donors and legislator

voting behavior to assess likely sources for the potential divergence

of voting and communicating (Bartels 2008; Bonica, McCarty, and

Rosenthal 2013; Gilens 2009). I seek to provide answers to the following

questions: (1) Do legislators engage in systematic attempts to shift

ideological perceptions of their policy preferences in their communica-

tions? (2) If so, which factors are most related to voting strategy, and

which are most related to communication strategy?

To test my theory, I analyze an original data set of over 40,000

legislator-to-constituent communications sent during the 111th

Congress, using a methodological innovation building on the ideal-

point-estimation technique of Clinton, Jackman, and Rivers (2004).

1

During this period, 95% of Representatives and 85% of Senators sent

off‌icial electronic messages to constituents. Using the votes a legislator

reveals in her communications, I estimate a communicated ideal point

(CIP) and compare this to her behavioral ideal point (BIP), which is

based on her full voting history. It is useful to study these communications

because, unlike media reports, broadcast interviews, or f‌loor speeches—

all of which are subject to external constraints, i.e., different agenda set-

ters, specif‌ic interview questions, and time limits that may compel certain

types of speech—off‌icial messages are optional, and the scope of each

message is at the discretion of the sender.

2

Strategic vote revelation may

have consequences for voter knowledge as somewhere between 14 and

19% of US adults report they have subscribed to off‌icial congressional

communications at some point (Ansolabehere and Schaffner 2012).

I f‌ind that there is a positive correlation between CIP and BIP esti-

mates, but this relationship is not perfect, and, on average, the measures

are distinct. Second, I observe an overall leftward ideological shift, with

CIP estimates being more extreme for Democrats and more moderate for

Republicans than BIP estimates. Third, I f‌ind that the likelihood that a

legislator presents a CIP that is more extreme than her BIP is conditional

on district characteristics, most importantly the ratio of base-to-swing

voters. Fourth, I conf‌irm f‌indings that relate donor ideological extremity

to voting extremity. I conclude with a discussion of why different factors

likely contribute to voting strategy versus communication strategy and

the implications of this divergence.

576 Lindsey Cormack