In the 1991 NES Pilot Study on the Consequences of War, respondents were asked about the position their representatives took on the resolution to allow the use of force in the Gulf. These items were repeated in the 1992 SES, and a parallel set of questions were asked about the vote to confirm Clarence Thomas as Justice to the Supreme Court. In some ways, these two issues are comparable. Both meet our basic criteria of high profile votes. The debate over the Persian Gulf War resolution was covered extensively in the print and broadcast media. Member agonized publicly over their positions. The President lobbied heavily for his position. The Thomas nomination similarly gripped the interest of the nation. Once accusations of sexual harassment surfaced, media attention, partisan posturing, and presidential lobbying grew intense. Only someone completely detached from public affairs could not have known that something was going on leading up to these two roll call votes.
In other ways, the votes are not comparable. First, and perhaps most important, the Persian Gulf war vote occurred nearly seven months before the Thomas vote, and 22 months before the SES hit the field in October, 1992. The Thomas vote took place less than a year before the 1992 SES. This alone ought to lead to lower levels of knowledge and accuracy for the Gulf War items. However, it is possible that the difference between 11 months and 22 months is irrelevant when we are asking respondents to recollect a specific roll call vote. Furthermore, 1992 was a campaign year, and both the Gulf War and the Thomas vote could have been an issue in some Senate races, reducing the knowledge gap. Second, as noted above, the Gulf War resolution was a foreign policy issue, one in which citizens may have been looking more to the President for leadership than the Senate. Neither race nor sexual harassment fall under the rubric ``domestic policy,'' but they are just as obviously not foreign policy. Finally, the Thomas vote dealt with a Presidential prerogative, appointment to the Court, as did the Gulf War (in some eyes). After the Hill revelations, however, Clarence Thomas's confirmation became a highly charged issue both within and outside the Senate. It was grist for tabloid television and talk radio. In addition, partisan rhetoric about Clarence Thomas was pointed, whereas, even if they disagreed with the use of force, most Senators agreed that something had to be done in the Gulf.
For all these reasons, time, race and sex, and the nature of the coverage, we think that public awareness of the Thomas vote ought to be higher than the Gulf War vote, and accuracy should be greater. The Thomas vote is a good case to test whether our results for the Gulf War, using 1991 data, are unique or are generalizable to other times and other issues.
In Table 
we present some simple descriptive statistics
showing how people answered both the Persian Gulf War Resolution and
the Clarence Thomas votes questions in the 1992 Senate Election
Study.
We provide in Table the percentages of respondents who
claim to know how their Senators voted on each issue, and whether they
recalled or guessed that their Senator voted yea or nay on both issues.
Table Goes Here
In our earlier study, where we used data from 1991, we found that 24% of respondents in that sample stated they knew the vote of their House representative on the Persian Gulf Resolution while 27% claimed to know the vote of their Senators on the same issue (Alvarez and Gronke 1996). Here, in Table 1, we find that the percentage of respondents in the fall of 1992 who claimed to know the vote of their Senators in January 1991 is strikingly consistent --- almost 25% of these respondents said they recalled the vote of their Senators.
A quarter fewer respondents claimed knowledge of the vote of their Senators on whether Clarence Thomas should be appointed to the Supreme Court. With almost 19% of respondents stating they knew of their Senators vote on this contentious and well-publicized issue, it is clear that the level of informedness about Senate action on Clarence Thomas is lower than the level of informedness about the Persian Gulf War Resolution. Thus, our expectations about these two votes were not met. Although in retrospect, we felt that the Thomas vote was more charged and salient, citizens remembered the Gulf War more frequently. Did they also do so more accurately?
Before we turn to this question, however, note that respondents were overwhelmingly of the opinion that their Senators had voted in favor of both the passage of the Persian Gulf War Resolution and the Clarence Thomas appointment. Roughly 70% of the people who said they knew the vote of their Senators on the Gulf War issue believed their Senator voted in favor of passage, while 69% guessed. The similar percentages are slightly lower on the Clarence Thomas vote, but even there, roughly 65% of respondents who either knew or guessed the votes of their Senators believed that these Senators had voted for passage of both pieces of legislation. Yet, both votes were relatively close. Clearly, false positives are rampant.
In Table , we report the accuracy of respondent recall and
guessing about the votes of their Senators on both these issues.
There we provide the percentages correct and incorrect, as well as the
percentages of false negatives and false positives for both those who
claimed to know the vote of their Senators and for those who guessed.
First, for both recallers and guessers, the percentages who were
correct is relatively high. Those who said they knew the votes of
their Senators were just under 60% for both the Gulf War and Clarence
Thomas votes; the percentages were slightly smaller for the guessers,
with almost 56% getting their Senator's vote on the Gulf War
Resolution correct and 51% getting the vote on the Clarence Thomas
appointment right.
Table Goes Here
These results do not compare very well with our earlier results from 1991. There we found that 76% of those who stated they knew the vote of their Senators on the Gulf War Resolution correctly recalled that vote, and that 59% of those who guessed got that correct (Alvarez and Gronke 1996). Given that the 1991 data was collected within about six months of the voting on the Gulf War Resolution, such high accuracy is notable. But the 1992 data were collected almost two years after the Gulf War Resolution vote; the 20% drop we see in recall accuracy, and the 8% drop in guessing accuracy, is most likely a reflection of the diminution in the ability of Americans to remember past voting behavior of their representatives. More worrisome, perhaps, if respondents simply guessed ``all yea'', they would have very nearly equalled these accuracy rates.
Is there any pattern in the ways in which respondents were inaccurate in their recollections of these two votes? Interestingly, those who stated that they knew the vote of their Senators, but who were incorrect, were somewhat more likely to err in the direction of inaccurately stating that their Senators had voted in favor of both measures. This tendency is slightly stronger in the Clarence Thomas case (47.8%) than in the Gulf War Resolution case (42.7%). Among the guessers, though, we do not see a consistent pattern --- those who guessed incorrectly were slightly more likely to guess that their Senator had voted against the Gulf War Resolution, but to guess that their Senators had voted for the Clarence Thomas appointment.
From these simple statistics, we draw three inferences about constituent knowledge of their representative's voting behavior. First, constituent knowledge seems to be consistent across these two salient Senate actions. We do not see great differences in the willingness of survey respondents to say they know how their Senators voted on these two, quite different, political issues; nor do we see great differences in how accurate these same respondents are in when they are asked to state how they believe, or to guess at, their Senators' votes.
Evaluating these percentages is difficult since we have no absolute standard against which to compare citizen knowledge. In general, respondents do seem to be more willing to state knowledge of these particular roll call votes than they do more general legislative behavior. In previous surveys, the NES has asked respondents if they recalled their House member's vote on any bill in the past few years. Just 11% of survey respondents in two successive NES studies (1988 and 199) were willing to say they could recall any vote. Notice that at least twice as many here --- and in our previous work using the 1991 data --- are willing to state they know the votes of their Senators on specific, if salient, issues.
Last, we do have evidence that citizen knowledge of representative legislative behavior decays with time. While we have no way of determining here how rapidly this decay occurs, or whether it eventually stops, we do have clear evidence that many fewer respondents were able to recall accurately, or even to guess accurately, the votes of their Senators almost two years after a significant legislative roll call vote.