LSS Newsletter Extension of Remarks

Introduction: Political Science and Divided Government

Political scientists have often implicitly or explicitly endorsed unified party control of government. The appeal of unified party government for political scientists appears straightforward. Cutting through the clutter of governmental institutions, unified government concentrates authority and accountability in a single party. It also helps clarify and diametrically oppose the interests of the competing parties, providing a clear choice for voters.

While others (Mayhew 1991, Jones 1994, Quirk and Nesmith 1995, Krehbiel 1996, and Cameron, Howell, and Adler 1997) have recently challenged whether legislation in unified government is actually more “productive” than in divided government, in this abridged argument I’ll be focusing on another aspect of partisan activity: Communication. This paper argues that political scientists have overestimated the value of party government, particularly for the presidential party’s ability to communicate effectively with American voters. In particular, I argue that:

- patterns of partisan communication are more damaging to the presidential party’s standing with voters in unified than in divided government;

and

- the non-presidential party is generally effective in their communication with voters in both unified and divided government.

I begin this process by developing a typology of partisan messages in the news, determining the likely effects of these messages based on their general content and credibility. Next, I predict how the proportions of these types of messages vary systematically across unified and divided control of government, and how these changes might affect public opinion. Third, I test these predictions using aggregate public opinion.

Credibility and Partisan Communication

In this section, I assume that the evaluative statements of partisans break down into four basic categories:

Statements attacking the other party (cross-party attacks)
Statements supporting one’s own party (intra-party praise)
Statements supporting the other party (cross-party praise)
Statements attacking one’s own party (intra-party attacks)

If parties are competing with each other for votes, they should generally prefer to broadcast messages from categories 1 and 2, and to avoid presenting messages from categories 3 and 4. However, in determining each message type’s effect on viewers, it is important to note not just the content of the message itself, but also the credibility of the message or its speaker. Party messages are not “injected” into the public, but rather are processed and can be accepted or rejected based on their credibility (Asch 1952, Sniderman, Brody, and Tetlock 1991, Zaller 1992, Kuklinski and Hurley 1996, and Druckman 2001).

One source of credibility for a message is the belief that the speaker and listener have common interests (Crawford and Sobel 1982). Putting common interest into partisan terms, we can therefore assume that statements by a listener’s own party will be regarded as more credible than those of the opposing party, all else equal (Partisan Credibility Assumption).¹

Such credibility is based solely on the characteristics of the speaker of the message and is independent of its content. However, another important source of credibility derives from the interaction of source and message: whether the message being spoken is costly to the speaker (Spence 1973). Messages that are viewed as harming the interests of the speaker are regarded as more credible than those that impose no costs (so-called “cheap talk”).² In the context of partisan messages, we can state that messages by a partisan speaker that appear to hurt their own party or help the other party are regarded as being more credible than messages that help their own party or hurt the other party (Costly Credibility Assumption). Moreover, such messages should be credible regardless of the party affiliation of the listener.

Table 1 summarizes the relative credibility of different political messages based on their partisan and costly credibility. The table also italicizes messages that are damaging to Party A.

Table 1:
Credibility of Partisan Messages where Listener is a Member of Party A

	Partisan Credibility
Costly Credibility	High	Low
High	A attacks A A praises B	B praises A B attacks B
Low	A attacks B A praises A	B attacks A B praises B

Consistent with the intuition of the costly credibility assumption, Table 1 shows us that the messages a party least prefers actually include the messages most likely to influence the beliefs of party members in the electorate.

Persuasion in United and Divided Government

Now that we have a broad sense of how changes in partisan messages might affect persuasion in the electorate, we turn to examining how the distribution and effects of these messages might vary systematically depending on institutional variables. In particular, we take up the issue of how the presence of unified or divided government can systematically alter a party’s ability to avoid delivering damaging and credible information to the electorate.

We begin this process by examining partisan messages presented in President Clinton’s first two years in office (the last instance of unified government). Table 2 highlights considerable problems for the presidential party in unified government. During this time, messages regarded as damaging to the presidential party (italicized in the table) make up well over half of the total partisan evaluations. Moreover, fully one in five of the messages being received at this time are from the upper-left quadrant of the table—the most credible and damaging quadrant for members of the presidential party.

Table 2:
Proportion of Messages, Unified Government (1993,94),
where PP is Presidential Party and OP is the Non-Presidential Party³

	Partisan Credibility for Presidential Party (PP) Viewers
Costly Credibility	High		Low
High	PP attacks PP PP praises OP	18% 3%	OP praises PP OP attacks OP	5% 3%
Low	PP attacks OP PP praises PP	10% 27%	OP attacks PP OP praises OP	28% 6%

Table 3 repeats this analysis for two years of divided government. In it, we find a more desirable situation for the presidential party.

Table 3:
Proportion of Messages, Divided Government (1992,95),
where PP is Presidential Party and OP is the Non-Presidential Party

	Partisan Credibility for Presidential Party (PP)
Costly Credibility	High		Low
High	PP attacks PP PP praises OP	6% ¯ 2% ¯	OP praises PP OP attacks OP	3% ¯ 6%
Low	PP attacks OP PP praises PP	30% 12% ¯	OP attacks PP OP praises OP	26%¯ 16%

While harmful (italicized) messages still make up about half of all messages, the vast majority of these harmful messages are now less credible to rank-and-file members of the presidential party. And while the presidential party is less able to praise themselves in the news, they are more able to savage the opposing party, presumably helping to shore up their standing with their partisans.

The non-presidential party, as we shall see below, is less affected by changes in control of government. Table 4 breaks down the same unified governmental messages as Table 2, but does so according to their relative harm and credibility for non-presidential party viewers.

Table 4:
Proportion of Messages, Unified Government (1993, 94),
where PP is Presidential Party and OP is the Non-Presidential Party

	Partisan Credibility for Non-Presidential Party (OP) Viewers
Costly Credibility	High		Low
High	OP praises PP OP attacks OP	5% 3%	PP attacks PP PP praises OP	18% 3%
Low	OP attacks PP OP praises OP	28% 6%	PP attacks OP PP praises PP	10% 27%

Table 4 shows that less than half of the messages in unified government are considered harmful (italicized) to the non-presidential party, and of those messages, the vast proportion are of low credibility. Conversely, over half of all messages are beneficial to the non-presidential party, and all such messages are at least somewhat credible. Table 5 repeats this analysis for divided government.

Table 5:
Proportion of Messages, Divided Government (1992, 95),
where PP is Presidential Party and OP is the Non-Presidential Party

	Partisan Credibility for Non-Presidential Party (OP) Viewers
Costly Credibility	High		Low
High	OP praises PP OP attacks OP	3% ¯ 6%	PP attacks PP PP praises OP	6% ¯ 2% ¯
Low	OP attacks PP OP praises OP	26% ¯ 16%	PP attacks OP PP praises PP	30% 12% ¯

On balance, moving to divided government does not change the overall proportions of beneficial or harmful messages for the non-presidential party.⁴

The next section begins testing the impact of such messages on actual public opinion. In so doing, I use national public opinion polls to test the contention that unified government disproportionately harms the standing of the presidential party.

Evidence from Aggregate Public Opinion⁵

In this section, we will be comparing the results of presidential approval polls in unified and divided government during Clinton’s first term. This analysis will rely on comparative statics drawn from two observations highlighted in the previous section: First, that presidential party members in the electorate received proportionately more damaging and credible information about the president and his party in unified than in divided government; and second, that members of the non-presidential party received substantially similar messages about the president and his party in both unified and divided government.

If we assume that increases in credible, damaging messages about the president should decrease his approval (all else equal), we can make the following predictions about partisan approval of the president in unified versus divided government:

H1: Disapproval of the president by members of his own party should be higher in unified than in divided government. Conversely, disapproval of the president by members of the non-presidential party should be unchanged in unified or divided government.

H2: Uncertainty regarding approval of the president by members of his own party should be higher in unified than in divided government. Conversely, uncertainty regarding approval of the president by members of the non-presidential party should be about the same in unified versus divided government.⁶

Table 6 uses four years of Gallup presidential approval polls to test Hypothesis 1 by examining disapproval of President Clinton by partisans during his first term in office.

Table 6:
Test of Hypothesis 1:
Gallup Disapproval of President Clinton by Partisans, 1993-1996:

	Disapproval by Republican Respondents	Disapproval by Democratic Respondents
Average in Unified Government (n=37)	71.3%	17.2%
Average in Divided Government (n=56)	72.2%	13.5%
t statistic	0.59	-4.34
		* p<.01

Table 6 shows us that Republican evaluations of Clinton were similarly negative in both unified and divided government. Democrats, on the other hand, were significantly more likely to disapprove of Clinton in unified than in divided government. Both of these results are predicted by Hypothesis 1. However, it is worth noting that the results in Table 6 might also reflect a change in Clinton’s performance in office following the 1994 election. In other words, it is unclear whether Democrats’ decreased disapproval reflects changes in the messages sent by Democratic elites or some actual change in Clinton’s job performance.

Table 7 addresses this issue by looking at “don’t know” responses by partisans regarding approval of Clinton’s job performance.⁷ Hypothesis 2 predicts that the transition from unified to divided government in 1994 should have decreased the indecision of Democrats, but not Republicans. If Democrats were receiving larger doses of credible and damaging information from Democratic elites regarding Clinton’s job performance in unified government, one would expect their indecision to decrease once the Republicans took over Congress in 1995. Conversely, if Republicans were able to transmit consistently negative messages regarding Clinton in both unified and divided government, one would expect their indecision to remain relatively constant.

Table 7:
Test of Hypothesis 2:
Gallup Partisans Who "Don't Know" Whether They
Approve of Clinton, 1993-1996:

	"Don't Know" by Republican Respondents	"Don't Know" by Democratic Respondents
Average in Unified Government (n=37)	7.1%	8.0%
Average in Divided Government (n=56)	6.7%	6.7%
t statistic	-0.51	-2.12
		* p<.05

As Table 7 shows, both Republicans and Democrats were at least somewhat more willing to express an opinion of Clinton’s job performance in divided government.⁸ Consistent with the predictions of Hypothesis 2, the Republican change is statistically insignificant, while Democrats were statistically more likely to answer “don’t know” in unified than in divided government.

Notes

1. Note that this does not imply that a voter will regard their own party’s statements as unvarnished truth, but rather that they would be less willing to believe the same statement coming from a member of the opposite party. This assumption is also consistent with observations of partisan perceptual screens, such as those discussed by Zaller (1987).

2. A related line of inquiry is research in the social psychological literature into the influence of “incongruous” stances, or ones that appear to be inconsistent with a speaker’s best interests (Walster, Aronson, and Abrahams 1966, Koeske and Crano 1968). This perspective is also reinforced by further findings that “disconfirming” messages (i.e. messages in which a speaker takes an unexpected position) are more persuasive than “confirming” messages. See Eagly, Wood, and Chaiken 1978, Eagly 1981).

3. Proportions are based on Center for Media and Public Affairs coded evaluations with presidential or congressional partisans as their sources and targets.

4. While there has been an increase in the proportion of “damaging” messages, almost all of the gain has occurred in the least-credible lower-right quadrant. Conversely, the significant drop in infighting by the presidential party is largely offset by the non-presidential party’s increase in credible self-praise.

5. Note that the full version of this paper includes additional evidence on this point drawn from on-line split ballot experiments.

6. Conversely, the null hypothesis would be that Republicans and Democrats should display similar levels of disapproval and uncertainty in their evaluations regardless of the institutional setting.

7. “Don’t know” responses are taken to indicate “ignorance, indecision, or uncertainty about the meaning of the question asked” (Sanchez and Morchio 1992). The first two causes of “don’t know” answers should equally impact Democratic and Republican respondents in the case of Gallup’s poll. Thus, changes in don’t knows in this situation should arguably be attributable to changes in the final component: uncertainty.

8. Since Clinton’s unified government took place when he was beginning his first term in office, it is intuitive that there might be somewhat less uncertainty two years later in divided government.

Tim Groeling is an assistant professor of communication studies at UCLA. He recently completed his dissertation, When Politicians Attack: The Causes, Contours, and Consequences of Partisan Political Communication, at University of California - San Diego. His articles have appeared in the Journal of Politics and in Polarized Politics (CQ Press, 2000). Email address: Groeling@ucla.ed