The Annenberg reports on Civility in the House of Representatives undoubtedly contributed to the discussions at Hershey in several important ways.

First, the initial 1997 report placed the issue of civility within the House in a broader cultural, social, and historical context. In this respect, the report noted concerns about a deterioration in civility within American society generally. It also illustrated that Representatives have expressed concerns about civility in the House throughout our constitutional history.

Second, the report identified a variety of factors, both internal to the House and external, that may have contributed to the apparently widespread perception among Members that they had witnessed a decline in civility in the House’s proceedings, especially during floor debate. For example, the report pointed to the increasing use of one-minute speeches, and the greater likelihood that these statements would receive media coverage if they contained dramatic or what other Members might consider to be inflammatory language.

Third, the report documented a variety of instances in which Members have spoken or referred to each other during floor debate in terms that could well be interpreted as being insulting or offensive. For anyone who had not listened to House proceedings regularly, the report offered sometimes dramatic anecdotal evidence to explain why some Members thought it advisable to convene the two Hershey gatherings.

Fourth, the report included recommendations for improving civility in the House through changes in House rules and practices, as well as in the unofficial norms that Members choose to respect and follow. For example, the report suggested that Representatives of each party take more initiative to discourage their own party colleagues from engaging in rhetorical excesses. In this context, the report quoted former Speaker Gingrich in late 1995 as having suggested “to the Minority leader that each party take a more active role in enforcing proper floor behavior. I fear that if it is left solely to the majority party or the Chair, it will be viewed as a partisan act.”

Finally, all three Annenberg Center reports undertook the daunting task of developing data to help assess whether and to what extent there actually has been a decline during recent years in civility during the House’s floor proceedings.

I have been asked to examine these data and offer some comments on what they may or may not tell us. In an important sense, however, the data really do not matter very much. If there is a widespread perception among Members themselves that the level of civility in the House is a problem, then it is a problem, without regard to what the data may reveal about the state of civility in the past, and even without regard to any quantitative measures of civility and incivility in the House today.

If, hypothetically, the Members who were planning either of the Hershey programs had been told that the data actually revealed an improvement in the level of civility in the House, I doubt very much that they would have canceled their plans. Instead, I suspect they would have concluded that the data somehow were not capturing what they saw and felt.

One difficulty in attempting to measure breaches of civility in debate is that civility can gradually be eroded when Members question the motives of their opponents without even approaching the boundaries of what the House considers to be “unparliamentary” debate. Again hypothetically, if one Member were to say to another that the real purpose of the latter’s party was to “destroy the social security system” or to “confiscate the wealth of the American people,” the result would only be to fray nerves, strain tempers, and create ill will in ways that might not be reflected directly in measurable instances of incivility that Professor Jamieson and her colleagues have been able to identify and count.

This is not to disparage in any way the thoughtful, careful, and sometimes imaginative ways they devised to capture and measure instances of incivility. Any social scientist soon comes to appreciate that the most interesting questions can be the most difficult to study systematically, and that the most important phenomena can be the most difficult to quantify in any rigorous way. In those cases, you do the best you can, doing what is possible rather than ignoring the questions and phenomena because they can be investigated only by imperfect means.

Take, for example, Professor Jamieson’s sensible approach to identifying and categorizing words that, when used in debate, could indicate incivility even if they were not used in ways that actually violated applicable House rules and precedents. The goal was to identify words that could be (or probably were) used in an uncivil way, and to place them into one of several categories (or indices). I think it is worth quoting from the methodological discussion in the 1998 report (on page 10):

Is such reliance on “inter-coder reliability” a perfect way to measure how often Members used “pejorative words” or words that cast aspersions on others, or how often they engaged in hyperbole? No. But is there a perfect methodology? No. Could I have recommended a better method? No. Sometimes the nature of the question or the evidence available leaves social scientists no choice but to rely on imperfect methods. In such cases, you use those methods as carefully as possible, and then put only as much weight on your findings as your methods can reasonably support.

Yet almost any social science research strategy is subject to being second-guessed. If this were not so, the social science journals would be half as long (but probably much more interesting to read). So as this Committee and other Members of the House review the data and findings of the three Annenberg Center reports, there are a few methodological questions and qualms that could be taken into account.

As members of this Committee know, House Rule XVII provides recourse when any Member believes that clause 1 of that rule or the House’s precedents governing decorum in debate are being violated. When one Member objects under clause 4 of Rule XVII to words just spoken by another, those words are “taken down” and read. The Speaker then decides whether the words spoken were in order (unless the words are first withdrawn by unanimous consent). If the Speaker decides in favor of the objection (or point of order), the offending Member can proceed in debate only by unanimous consent or on motion.

Because invocations of this procedure indicate serious breaches of civility (or serious allegations of such breaches), the Annenberg reports look for trends over time in “words taken down.” The first two reports each present two figures: one tracing the number of instances in which demands that words be taken down led to rulings by the Chair; the second limited to instances in which the Chair ruled that the words in question were not in order. In the most recent report, however, I wonder whether one of the two figures has been changed in a way that mixes apples and oranges.

The figure on page 8 of the 1999 report purports to display data on “requests that words be taken down.” For the 1935-1984 period, however, the data presented in this figure, like the first of the two figures in each of the earlier reports, are on rulings of the Chair in response to demands that words be taken down. As the report itself notes in its methodology section (page 15), “[t]he counts for the years before 1985 only include requests that resulted in a ruling and not times when the words were withdrawn, the demand was withdrawn, the demand was too late, or a caution was given by the chair.”

For the 1985-1998 period, by contrast, the data are based on an online search of the text of the Congressional Record, and appear to include all cases in which Members demanded that words be taken down, without regard to whether or not that demand ultimately led to a ruling by the Chair. The original 1997 report shows 14 instances in 1995 and 2 in 1996 in which the Chair ruled on whether certain words that had been taken down were or were not parliamentary. The 1999 report, on the other hand, shows 19 instances in 1995 and 6 in 1996 of requests that words be taken down. Similarly, the 1997 report depicts 4 instances in 1992 in which the Chair ruled in response to demands that words be taken down, but the 1999 report shows 12 requests that words be taken down. I suppose that the differences represent instances in which, for one reason or another, the Chair made no ruling.

If that is the case, the figure and data on page 8 of the 1999 report combine data on rulings during 1935-1984 with data for subsequent years on requests that words be taken down. This juxtaposition could give an exaggerated impression of how incivility on the House floor may have increased since 1985. I should repeat that the corresponding figures in the first two Annenberg Center reports do not raise this question.

There are two other ways in which all the reports attempt to tap levels of incivility and disruptive behavior on the House floor, and both give me some pause.

First, the reports present data on the number of instances during the 99th-105th Congresses in which the Chair asked that the Member speaking “suspend.” These data show a modest increase during 1990-1994, then an explosion of instances in 1995, and a gradual subsidence in the last three years, though not back to the pre-1995 levels. The question is what to make of these numbers. It is true that the Chair may direct a Member to “suspend” when the Chair needs to admonish or remind the Member speaking about the limits of acceptable parliamentary discourse (for example, when a Member makes an inappropriate reference to the President or a Senator). However, the Chair also may ask Members to suspend simply because the Chair needs to ask other Members to stop talking among themselves so that the Member who has the floor can be heard. As the original 1997 report notes (on page 35), requests that Members “suspend” is “not an invariable predictor of tension.”

Unfortunately, the reports are ambiguous as to whether they present data on all instances in which Members were asked to “suspend” while speaking, or only on those instances in which the Chair directed Members to “suspend” because of what they had been saying. If the researchers actually examined all instances that their search of the Congressional Record revealed, and only counted those that were related to the content of Members’ speeches, then we would have helpful data on how often the Chair felt obligated to intervene at his or her own initiative to preserve or restore decorum in debate. On the other hand, if the reports present data on all the instances found, then these data are much more difficult to interpret, because they combine instances of inappropriate debate with instances of excessive noise in the House chamber.

These data raise the question of exactly what we mean when we talk about civility and incivility in the House. The same question is raised even more clearly by the final measure reported in two of the Annenberg reports: the number of instances in which the Chair called for the House to be in order. Such data appear in the first two reports, but not in the third.

The 1997 report states that “[c]alls for the House to be in order were sought out as a crude measure of the amount of noise and distractions in the Chamber at different periods.” For example, and as every Member of the House knows better than I, the Chair sometimes has to make several attempts to call for order in the chamber after a record vote because Members are milling about in the well and engaging in conversation so that “the amount of noise and distractions” prevents the House from resuming legislative business. No doubt these occasions are annoying to Members occupying the Chair and to those who are concerned with whatever business the House is scheduled to conduct next. I do wonder, however, if these are the kinds of situations that most Members have had in mind when they have expressed concern about a decline in civility in the House. Such behavior may be impolite, and under some circumstances there may be only a very indistinct line between a lack of politeness and a lack of civility. In the context of the House, however, and this Committee’s concern with civility, I think it is helpful to distinguish between what is thoughtless behavior at worst and what is offensive, even knowingly offensive, language used in debate.

Next I would like to comment briefly on an innovation in the 1999 report: the discussion of the correlation between productivity and civility. We learn on pages 10-11 that there is a reasonably impressive (by political science standards) inverse relationship between rates of “name calling” and the numbers of all measures and the number of joint resolutions that the House has passed per Congress. And we learn that there is an even more impressive positive correlation between the rates of “name calling” and the hours the House has been in session. What the data themselves cannot tell us is whether we are looking at mere correlation or whether there is a causal argument being implied.

It surely is plausible to argue that long hours make for raw nerves, so that the number of hours in session may provoke some Members on some occasions to use uncivil rhetoric in debate. But it is less clear whether we are to infer from the first two figures on pages 10-11 that a lack of legislative productivity is so frustrating to Members that they engage in “name calling” in reaction, or whether we are to infer instead that “name calling” depresses productivity, so that if Members behaved more civilly to each other, the House would be a more productive place. Or perhaps what we have here is correlation without causation. Although I have not done the calculations, we might well find a strong correlation between declining civility in the House during the 1990s and increases in the Dow Jones Industrial Average.

I certainly do not intend to imply, in making this last argument, that Professor Jamieson and her colleagues do not already know this better than I. I make the argument, however, because the manner in which these relationships are presented in the 1999 report does not, in my judgment, make sufficiently clear what we are to conclude from them.

Finally, let me offer a few observations about what we might conclude from all the data presented in the three Annenberg Center reports. For reasons I have touched on in this statement, the most solid data presented in the reports are on the number of rulings on words taken down and the number of instances in which words spoken were ruled out of order. Looking solely at these data, we can conclude that the House floor was a far more contentious place during the 1940s and the early 1950s than it was during the 40 years that followed. In fact, what may be most striking about these data is how rarely the House’s rules and precedents on decorum in debate gave rise to formal rulings during years when the House was debating, among other things, civil rights legislation and the Vietnam War. Yet then we observe a striking increase in both measures during 1995.

Taken alone, or taken in conjunction with some of the other data presented in the reports, does this recent development justify action by the House? Surely the Members who conceived the idea of the Hershey conferences and took the initiative to arrange them believed that there was a problem that needed to be addressed. And surely one effect of the Hershey conferences was to encourage Members to remind themselves how they are expected to speak and behave on the floor in accordance with the House’s existing rules and precedents.

If, however, this Committee contemplates proposing rules changes to promote civility, its Members might ask themselves two questions.

First, do we really know what has caused the problem we want to solve? The first Annenberg report identifies quite effectively a number of factors that may have contributed to a decline in civility in the House. I wonder if we really can predict with any confidence whether any particular change in internal rules or practices will make a significant difference in light of all the other possible contributory factors that are beyond the House’s direct control.

And second, are we confident that there is a persistent, long-term problem, and not what may prove to be a transitory or self-correcting phenomenon? I am sure that any Member who has examined the figures in the Annenberg reports will have noticed (as the 1997 report clearly points out) that the data spikes in their figures for 1995 compare only with the spikes for 1946, the year in which the long-time Republican minority regained control of the House. Many have characterized 1995 as a year of revolutionary change in the House, and as a year during which Members of both parties needed time to adjust to their changed situation in the House. Since 1995, the data generally have trended back toward the status quo ante. If the state of civility in the House was satisfactory before 1995, perhaps it soon will be so again, and without the need for any formal action by this Committee and the House.

But then again, perhaps not.

Back to Testimony Page