【正文】 s (1986) argument that the evidence is contradictory, depending on pleas and prior judicial action. The trend of favorable treatment for women is strong enough to make it pertinent to ask if the criminal justice system is typically chivalrous or paternalistic towards women, or unduly punitive towards men, or if it appropriately recognizes the dependency demands on most women thereby protecting innocent dependents and balancing social costs (Daly, 1987). Alternatively, as Naffine argues, sentencers may simply operate from the assumption of an abstract person, a standard individual abstracted from any particular set of social circumstances. If this is so, then sentencers are most in need of patterns to guide their search for relevant case details when they deal with offenders who do not fit the image of the ideal legal person who is masculine, autonomous, rational, independent. Consequently, the very young, the old, and the poor would join women as problematic drinkdrivers to sentence. Alternatively, for a mon, rec。s work space is constrained and outes are quantifiable, it is possible to be explicit about individual differences rather than to rely upon global descriptions of magistrates as idiosyncratic (McFatter, 1986, p. 150) or stylistically tough or lenient (Homel, 1983a). In addition, by sampling magistrates and cases from two large city courts that were contrasted on overall severity, court could be used as a further, supraindividual factor. We expected that the prehensive modeling process would allow us to explain more of the sentencing variance than is possible using general magistrate factors. Since the interpretive sentencing process is dependent on the adequacy of information made available to sentencers by attorneys and wit nesses with different agendas and abilities (Lawrence, 1984, 1991。s overall severity in penalizing drinkdrivers (see Homel amp。s age, alcoholic dependence, and actual or potential cause of a collision。 tariff versus individualized sentencing goals。 varying positions on: the seriousness of drinkdriving as an offense。s age, standing as a person of good character, and susceptibility to alcohol dependency. From this and many other studies using archival data (see Homel, 1982), we generated a set of orientations to classify empirically magistrates39。 propensities to use tariff or individualized approaches to sentencing。s (1983a, 1988) extensive research, and we could specify different positions on those orientations. Analyses of over 15,000 drinkdriving cases allowed Homel to infer that penalties were in fluenced by the magistrates39。s case, working back and forth between the ining information and stored patterns of how different types of cases hang together (Lawrence, 1988a, 1991). Little sense can be made of the mass of case information that may affect sentencing outes, unless we understand how that information is filtered, interpreted, and clothed with meaning by the individual sentencer. If experienced judging shares the characteristics of expertise over many domains (Chi et al., 1988) the orientations sentencers apply to cases are likely to be task (offense) specific, and sensitive to different details and their sources. For example, a sentencer seeking to rehabilitate alcoholdependent offenders may pay careful attention to how a drinkdriver came to police notice, and how much alcohol he had consumed on this and other occasions. The sentencer matches these details with his or her mental image of the typical alcoholic drinkdriver, Another magistrate with a tariff approach may simply slot blood alcohol concentration (BAC) into an offenThere is little doubt that personal and social characteristics color sentencers39。 Lawrence, 1984). models, adding normative values and rules as well as affective markers to the knowledge used for interpreting information. Personal orientations and case information are brought together in the sentencer39。 courts, it is reasonable to expect that an experienced magistrate is able to call upon stored patterns of typical cases as the next case is presented in court, and of course, these stored patterns are influenced by that magistrate39。 abilities to pile and organize their knowledge and belief structures to allow them to construct working images or mental models of each new task (Chi et al., 1988。 Lawrence and Homel, 1987). The work of the sentencer is to select, weigh, and apply evidence to par ticular cases. Essentially, it is an information management activity and the sentencers39。 Softley, 1980), there is limited value in attempting to predict penalties from sentencer factors, if sentencerrelated influences are not examined in terms of their responsiveness to the different configurations of case features such as the actual offense category under which an offense is classified (Douglas, 1989). Sentencer factors may be once removed from the courtroom task, and simply function as the backdrop to the sentencer39。 attitudes, goals, and role definitions to sentencing outes (., Gibson, 1978。s interests and mitments, with consequent possibilities of variations in explanatory power. For instance, theoretical assumptions of stable personal traits and attitudes are likely to lead to analyses that do not look for intrasentencer variability in response to different contexts (Douglas, 1989。 Hogarth, 1971. McFatter, 1986), although analytic procedures for encapsulating these interactions are no simpler than the explanations they seek to supply. For example, Grossman (1966), Green (1961), and Hood and Sparks (1970) agree about the futility of seeking onetoone associations between a ju