The petit jury in criminal and civil trials has been closely studied for several years (see reviews by Davis, Bray, & Holt, 1977; and Gerbasi, Zuckerman, & Reis, 1977). Empirical data have been gathered from surveys of court records (e.g., Bermant & Coppock, 1973; Kerr, Harmon, Graves, & Sawyers, Note 1), mock trials (cf., Bray & Kerr, 1979; Davis, Bray, & Holt, 1977), courtroom observers (e.g., Diamond & Zeisel, 1974), and posttrial interviews (e.g., Broeder, 1965; Kalven & Zeisel, 1966). Nonetheless, there remain too many important questions and proposals concerning existing procedures to be answered by the existing data base, or by data likely to be available.
This problem is not confined to the courtroom, or even to the criminal and civil justice systems. Attempts to engineer social institutions often encounter the suggestion to “wait for a study to be conducted.” While engineering of all kinds, whether social or physical, necessarily relies upon empirical data, the physical engineer has the opportunity to obtain preliminary results from theory—sometimes well-established theory—or even to examine the behavior of a small-scale simulation or system mockup. Such a tradition is generally lacking in the case of social engineering, or where evaluation of existing social groups and institutions is at issue. Both time and expense are formidable impediments to a similar strategy in evaluating social change, especially since the latter must often be reckoned in human as well as financial terms. Yet, changing a social institution that has evolved over long periods of time is not undertaken lightly, and is usually difficult to undo should the well-intentioned alteration prove to be a mistake.
Empirical data play an indispensable role in any social engineering endeavor. But we should like to argue that precise, formal theory is equally indispensable, especially for the complex phenomena likely to be encountered in the case of social institutions. The fundamental goal is to evaluate the likely consequences of each of several policies or procedures being considered, even though few or no directly relevant data are available. Several illustrations are available, in connection with the criminal jury, of the general approach in which data guide theoretical simulations or “thought experiments.” (See Davis, 1980, Vollrath & Davis, in press, and Nagao & Davis, Note 2, for examples and more comprehensive discussions of this general use of thought experiments.)
Other evaluations of recent procedural changes in criminal juries (namely, in allowable size and decision rule for establishing a verdict) have been undertaken. Both empirical and theoretical studies followed decisions by the Supreme Court (e.g., Williams v. Florida, 1970; Apodaca et al. v. Oregon, 1972) which allowed juries of less than 12 persons and decision rules of less than unanimity. The Court has recently established the minimum size of the criminal jury at six (Ballew v. Georgia, 1978), and Saks (1978) has noted the significance of the Court's unique citation of such research studies in the decision. (See Vollrath and Davis, in press, for a review and detailed discussion relevant to the questions of size and decision rule in petit juries.)
However, there is a somewhat different use of data-guided thought experiments which has much to recommend it: the a priori consideration of social policies. For example, the original Court decisions on size and rule were evaluated by social researchers after they became law; the recent Court decision cited social research as having been useful in the process of deciding. The next step might be the use of social research, empirical and/or conceptual, prior to the event—at the time legislative, executive, or judicial policy is being formulated or debated. We consider now just such an example in connection with grand jury reform.