¹ The numbers themselves will be of interest to students of courts as institutions, for they show how the business and users of courts vary over time. But those who have this interest are ordinarily also interested in these patterns as indicators of the place of courts in society.

² My debt to Geertz (1983) for the phrase “local knowledge” is obvious, as is the fact that I am not using the phrase precisely as he did. For Geertz, “the law is local knowledge” (ibid., p. 218). Law is necessarily in some measure idiosyncratic to time, place, class, and attitude because it is constitutive of the social world as it exists at particular locations. I use the term “local knowledge” not in an effort to make sense of the law-involved actor but to advise the would-be objective observer. Knowledge of the local—that is, the local court and legal cultures—is necessary to understand those cultural artifacts like court records that are the most concrete residue of court activity. If Geertz is correct in characterizing the law as local knowledge, we cannot understand our legal past or changes in legal activity over time without local knowledge in my sense of the term. Rather than assuming that the meaning underlying court records is constant across localities or over time, we must be aware that the records are products of different cultures, and that surface similarities and differences do not necessarily reflect parallel similarities and differences in underlying dynamic or meaning. To give a simple example, similar per capita rates of tort litigation in two locales or at two points in time will not reflect a similar propensity to litigate (“litigiousness”) if the rate of potentially actionable behavior differs in the locales or if alternative means of achieving the ends of tort litigation, like the presence of a socialized health care system, differ across the locales.

³ At one time appeals of the board's decisions were almost nonexistent. In recent years appeals from the eviction board to the Authority's Board of Commissioners have regularly occurred, but appeals to the circuit court are still exceedingly rare.

⁴ For two years it was composed of three HHA officials, then it was composed of five lay people, then two tenant members were added, and finally a second, seven-member panel was created. Along with the last change came a change in the types of people appointed to the board. At about the same time the Authority's system for prosecuting evictions became centralized and professionalized (Lempert, 1989).

⁵ Eviction files usually include information not only about the cause of action (predominantly nonpayment of rent) but also about family size, composition, age, income, occupation, and welfare status. If the tenant is represented by counsel, this will be indicated as well as the presence at the hearing of witnesses for or against the tenant; there is often information about a tenant's “project citizenship” as well. In addition, for the board's early years and for occasional cases thereafter, full transcripts of the hearings are available.

⁶ I also might have thought that the absence of eviction hearings at one point in the 1970s was a culmination of the trend that began in 1969. However, I was alerted by several informants to an important court case in which the Authority was involved, and a search for information regarding this case brought to light several memoranda cancelling for a period of months all eviction hearings until certain matters relating to the case were resolved. Since this case was settled before trial, there is no case report. Had I been focusing on a period in the more distant past, I would no doubt have missed the case entirely.

⁷ Note that the degree to which difficulties like those I describe are problematic varies with the questions one seeks to illuminate. For some purposes, particularly when the focus is on courts in a narrow institutional perspective, docket data may not pose special difficulties of interpretation. For example, in a study seeking to understand why intermediate appellate courts develop, the number of appeals filed is an obvious and not especially problematic exogenous variable. And just as qualitative investigations may condition the interpretation of docket data, so may docket data call into question interpretations based on more qualitative evidence. Thus Galanter (1983) relies in part on studies that report docket data in his important article questioning the “hyperlexis” hypothesis.

⁸ For example, stricter decisions in nonpayment cases in the mid-1980s are associated with a decline in the proportion of such cases that have legal representation. The obvious interpretation is that when lawyers are less involved in the eviction process, tenants fare less well. Conversations with legal aid attorneys indicated that the causal direction was the reverse. Legal aid responded to the Authority's stricter policies in nonpayment cases by refusing to represent nonpayment tenants, since to do so would be to waste scarce resources (Lempert and Monsma, 1988).

⁹ Note I say “aim.” It is impossible to avoid some degree of objectification because the social scientist is always imposing meaning on traces left by others, whether the data are quantitative or qualitative.

¹⁰ Missing the initial causes of enduring patterns of activity can lead to mistaken interpretations because, as Joe Sanders pointed out in a seminar we co-taught many years ago, in any ongoing social or cultural system the initial and continuing causes of a pattern may differ. Without information about initial causes, a pattern may be attributed to continuing causes that are in fact effective only after a pattern has begun. For example, plea bargaining may begin in a jurisdiction because of caseload pressure. Practitioners may then be socialized into disposing of cases by guilty pleas, and thus plea bargaining may continue even after caseload pressure has diminished. An analysis that lacks specific knowledge about the forces that first led to plea bargaining but that finds no direct correlation between caseload pressure and plea bargaining rates will mistakenly conclude that caseload pressure has no causal relationship to the existence of plea bargaining.

¹¹ For example, if contextual conditions are markedly different at two adjacent sounding points, the researcher might be able to identify a transition point between the sounding points and look for effects associated with the transition. If the transition involves an abrupt change and the soundings have permitted sufficiently precise identification of the point at which its effects occurred, interrupted time series designs can provide a good way of testing for effects of the change in conditions. If the change is not abrupt, one may be able to incorporate a variable that captures the incremental contributions of the continuing change in a longitudinal model. Without having taken deeper soundings at discrete points, the researcher might never have been alerted to the existence or potential importance of the particular change and thus may never have thought to test for possible effects.

¹² Some of these advantages apply to research in the recent but not more distant past. Informants, for example, can explicate changes in data collection routines, and the recent increases both in our capacity to generate and store “hard data” as well as an increase in our obsession with social statistics mean that for many variables (e.g., crime or accident rates) data are available for recent decades that are not available or are unreliable when we go further back.

¹³ Such team enterprises have existed to study other problems like intergenerational mobility and changes in health over time.

¹⁴ For a nice example of the richness that is possible when local knowledge and quantitative models are used in tandem, see the complementary articles by Berk et al. (1983) and Messinger et al. (1985).

¹⁵ In arguing for the importance of local knowledge in court docket research, I do not mean to suggest that researchers in this area never go beyond docket data. Indeed some have been acutely aware of the importance of the kinds of cultural understandings that local knowledge allows. Friedman and Percival (1981) and Kagan (1981) have written books to do justice to the complexities of the cultures they were studying. Press accounts, letters, and interviews have figured in the work of scholars such as Munger (1986a, 1987a) and Daniels (1985), and they and others have recognized the need for knowing more about the context in which litigation is embedded. Thus I do not mean to appear as if I were the first to recognize the importance of the considerations I describe. Nevertheless, I would not have written these comments if I did not feel that many court docket studies suffer from insufficient attention to the kinds of factors that can be uncovered only through a quest for local knowledge.

I also do not mean to suggest that the pursuit of local knowledge is the only methodological strategy that can enhance our ability to gain knowledge about the court-society relationship from court docket research. For example, a very different strategy that has promise in this respect is to test quantitative models that have been precisely specified a priori on well-defined theoretical grounds. If the data accord with previously specified, well-grounded models, there may be few plausible explanations for the fit other than the ones specified in the theory. It is, however, my hunch that the generation of plausible well-defined models will require considerable local knowledge.

Finally, the relationship between quantitative and qualitative data is a two-way street. Quantitative data may question or condition conclusions that qualitative sources suggest. Qualitative analyses in their own way should be rigorous, but it is no accident that the term is usually associated with quantitative investigations. I could give examples from my Hawaiian research to support these contentions, but that would be another paper; one that would be less responsive to the shortcomings of the studies of court records that have been done to date.