¹ What I have labeled the “consensus” theory has many theoretical roots. Most important is probably Durkheim (1964a).

² Consensus theory has led to two different schools of thought about the timing of decreased nonstate control and increased litigation. According to one school, economic development, population growth, and urbanization break down traditional customary relationships and lead to an ever increasing litigation rate as society develops. The second school hypothesizes that nonstate control is particularly weakened during an initial period of economic growth, called a “takeoff” period. Such expansion, it is argued, breaks down or mitigates against the emergence of regularized transaction patterns and customs, leading to an increased reliance on law to regulate social interaction. However, as time passes, regularized transaction patterns emerge, decreasing the need for legal recourse (Munger, 1988).

The first version of consensus theory sees a positive linear relationship between litigation and economic development. The second version sees a curvilinear relationship. However, they both agree that litigation functions to cope with increased social disharmony.

³ The Superior Court of Arizona is the state's general jurisdiction trial court. Its jurisdiction extends to all cases in law and equity not specifically vested by the constitution in other courts, the latter consisting of those few cases reserved by the constitution for the supreme court and those civil cases under $500 reserved for the justice courts. Cases involving from $500 to $1,000 were under concurrent jurisdiction of both the justice and superior courts. The superior court also deals with all domestic relations cases but not probate matters. While the superior court does have criminal jurisdiction, such cases were not included in this analysis.

⁴ For a discussion of the sampling method used, see Stookey (1986: 291–93).

⁵ In the analysis I also included a postwar variable on the assumption that there might be a “release” of pent-up disputes that were not litigated during the war.

⁶ For a description of the technique used see Lewis-Beck (1986). Simply, my approach was to conceptualize the beginning of each crisis as an intervention in the litigation time series. I determined the exact character of the intervention to reflect the likely impact of that intervention. In some instances, the intervention was simply a pulse function. For example, because the first depression was so short and because the expectation was that there would be a large, rapid, but brief increase, the depression years were coded 1, while other years were coded 0. For other interventions I included both an intercept change factor and a slope change factor.

The actual codes for each variable are given in Appendix 1. Some minor variation in the timing of the intervention results from my uncertainty about the difference in lag times associated with each litigation type. Therefore, for each intervention I would hypothesize a year and then check the statistical results for that year and for other years immediately before and after that year. The year selected is the one which best explains the data. While this approach biases the results in “my” favor, it seems an appropriate research strategy, given our primitive understanding of lags between social change and litigation.

⁷ Data from the policy analysis came from Arizona Session Laws and from the state's leading newspaper. With two student coders I analyzed all volumes of the Arizona Session Laws. We sought to identify all pieces of legislation that concerned matters which either might influence litigation or reflected litigation. Obviously a subjective element was involved in this selection, but resources precluded an analysis of all statutes, and a purposive approach seems more appropriate for this study than some type of random sample.

We carried out content coding of the Phoenix Gazette primarily for periods of crisis using random and purposive checks. Random checks were conducted for one day per month during depression and war years. Purposive checks were made for the entire week before and after an important piece of relevant legislation was passed, such as the foreclosure moratorium of 1933.

⁸ Because the economic cycle is so important to the analysis of crises and their impact on litigation, in each of the time series I have plotted an economic cycle. For a description of how it was calculated see Stookey (1986: 293).

⁹ In time series analysis autocorrelation may significantly inflate measures of statistical significance. In handling this problem, I follow the approach taken by Munger (1988: 83 n.32). Whenever the Durbin-Watson statistic indicated that autocorrelation was present, I have used the Cochrane Orcutt correction procedure. The data reflect such corrections, where appropriate.

¹⁰ Divorce cases were excluded from this analysis because the plaintiff is the victor in almost every such case. This fact, when combined with the large numbers of divorces, would totally mask the victory pattern for other I versus I cases.

¹¹ That farmers were organized is indicated by the large number who attended the legislative session and by the groups represented at the hearings. The day of the vote on the legislation, the Senate gallery was filled to capacity with farmers and both the director of the American Farm Bureau Federation and the president of the American Farm Bureau were present (Phoenix Gazette, 22 February 1933, pp. 1–2). On necessary preconditions for successful political organization and pressure, see Tilly (1978).