Focally Concerned About Focal Concerns: A Conceptual and Methodological Critique of Sentencing Disparities Research

Abstract

“Focal concerns” is the predominant theoretical framework in criminology for explaining disparities in sentencing outcomes. While the framework has generated a large body of empirical scholarship, its postulates remain inadequately tested in the criminological literature. In this paper, I offer a conceptual and methodological critique of focal concerns as it is being deployed in a large body of sentencing research. I first trace the genealogy of the “focal concerns” concept and detail its current articulation. I then describe the body of work that has reduced “focal concerns” to a commonsense psychological construct, and illustrate the fallacies of logic and paucity of direct theory development and testing that weaken the explanatory value of the framework. I conclude by building on Ulmer’s recent call to treat criminal courts as “inhabited institutions” to assess approaches that are more social scientifically robust and empirically testable for understanding how sentencing disparity is produced.

Keywords:

• Focal concerns
• sentencing
• disparities
• inequality

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 Very early on, Miller himself published about this research in social work (Miller, 1959) and psychiatric (Miller, Geertz, & Cutter, 1961) journals, which may have opened up an audience in the psy-fields.

2 A 1976 Ohio State University doctoral dissertation by Ramon Priestino, (1976) also appears to have adapted Miller’s “focal concerns” typology chart into a “focal concerns scale” which he used as an individual differences measure in a study of parole agents and aides. He credits Miller, (1958) for the scale itself, but Miller neither provides a such a scale, nor does he really speak of focal concerns at the level of individual differences.

3 Indeed, in the 1972 article, Hoffman suggested that decision-making could be made less biased by directing decision-makers to consider individual “focal concerns” in assessing defendants’ risk (Hoffman, 1972, p. 122).

4 I located one indirect cite to Hoffman’s use over the next two decades in a Wayne State Law Review article on the rise and fall of indeterminate sentencing (Zalman, 1977). A 1978 LEAA report on a federal sentencing guidelines’ feasibility study also used the Hoffman and DeGostin version of the “focal concerns” concept (Wilkins, Gelman, Kress, & Calpin, 1978).

5 In two publications that lay out the origins of focal concerns theory (Hartley, 2014; Hartley et al., 2007), Richard Hartley and colleagues assert that focal concerns as a concept was first used by Steffensmeier in 1980 (Steffensmeier, 1980), however that non-empirical article does not use the focal concerns terminology whatsoever, and instead evaluates various potential explanations for gender differences that range from the perceptual to the cultural to the structural. Steffensmeier and colleagues (1998, 766) marked their first development of the focal concerns framework with the Steffensmeier et al., (1993) article. In later articles, however, Steffensmeier and coauthors (e.g., Demuth & Steffensmeier, 2004) suggest it was first articulated in the 1980 article. Because that 1980 article did not reference “focal concerns” and was primarily focused on evaluating the predominant explanations for gender differences in criminal justice outcomes, I am not considering it as the original articulation of the framework.

6 They did include a number of county contextual variables as controls in their preliminary analyses, but those were highly correlated with court size resulting in “severe multicollinearity” so those analyses were not presented (Steffensmeier et al., 1998, p. 774). Even this approach, though, is not adequate to capture the strong influence of local jurisdiction on criminal case outcomes, which a multi-level model would be better able to capture.

7 It is often framed in the court communities literature that also influenced Steffensmeier, et al. (1998).

8 An excellent alternative approach was used by Alexes Harris, (2009) to examine juvenile case outcomes. She coded case fitness reports to directly measure the attributions made about youth being considered for transfer to adult court.

9 I would add defendants in that mix as well, since they have the ultimate power to take or reject plea offers.

10 This assertion seems especially problematic in the federal court context, at least outside of the high-volume border districts, since judges and other actors have huge amounts of information about defendants and their lives, and they have ample time to review material and determine sentence. The standard period after conviction for pre-sentence investigation, filing and vetting of pre-sentence report, parties’ responses and filing of sentencing positions, etc. is 80-90 days during which a considerable volume of information is produced and shared among all parties about the defendant. See generally, Rule 32, Federal Rules of Evidence; see also Ulmer and Bradley, (2006) calling this into question in serious, violent cases.

11 Attribution theories dominated social psychology from the late 1950s through the 1970s, but have waned in influence since then (Weiner, 2008). In any case, this body of work is much more complex and nuanced than the usage of “attribution” in focal concerns research.

12 Focal concerns sentencing research has also used state-level data from Pennsylvania (e.g., Steffensmeier & Demuth, 2001; Ulmer & Johnson, 2004), Florida (e.g., Crow & Bales, 2006), Texas (e.g., Fernando Rodriguez, Curry, & Lee, 2006), and the SCPS data from 75 large U.S. counties (e.g., Demuth & Steffensmeier, 2004).

13 There are now methods for linking the federal data across stages of criminal justice process, although it is not a perfect system (Johnson, 2018; Rehavi & Starr, 2014).

14 This too is a common feature in available administrative datasets from courts.

15 Several scholars have made the case that at a conceptual level, specifying demographic features including race, in particular, as predictor variables raises a number of interpretation problems, in that it seems to specify demographic identity as a causal force. For example, Martin and Yeung, (2003, 522-523) suggest that “the use of race as a term in causal models may reinforce the idea that race is itself a ‘cause’ of differences in outcomes between say, Whites and Blacks. Not only does this invoke an idea of causality that makes little statistical sense…but it may lead us to consider that we can answer questions as to the reason for certain differences… simply by invoking the ‘fact’ of race, as opposed to considering this itself as a problem to be explained.” Zuberi, (2000, p. 179) contrasts the use of these kinds of regression approaches to the experimental paradigm to make the point that these kinds of attributes can never be causal in inferential modeling: “Causal statements involving unalterable characteristics are really statements of association between the values of an attribute and a response variable across individuals in the population.”

16 For two more carefully worded approaches that clearly acknowledge the descriptive nature of such disparity studies, see Mustard, (2001) and Everett and Wojtkiewicz, (2002).

17 It also highly endogenous with the imposed sentence dependent variable, which makes its utility as control variable suspect (Fischman & Schanzenbach, 2012).

18 It is not limited as a practice to focal concerns literature, but for my purposes it is important to highlight since it has become a standard practice within this literature. It also aptly illustrates the issues that plague this literature in regard to theorizing and modeling disparate outcomes.