James P. Scanlan, Attorney at Law

Massachusetts Criminal Justice Disparities

(Oct. 25, 2015)

This is one of ten subpages (as of October 22, 2021) of the Criminal Justice Disparities of page of jpscanlan.com.  As explained on the main Criminal Justice Disparities page, it and its subpages principally address the mistaken belief that generally reducing adverse criminal justice outcomes tends to reduce (a) relative racial differences in rates of experiencing the outcomes (as commonly presented in terms of the ratio of the Black rate to the white rate) and (b) the proportion Blacks make up of persons experiencing the outcomes (compared with the proportion Blacks make up of population potentially experiencing the outcomes).  In fact, the opposite is the case.

That is, as I have explained in scores of places with respect to any favorable or adverse outcome since 1987, when two groups differ in their susceptibility to an outcome, generally reducing the outcome, while tending to reduce relative differences in rates of avoiding the outcome (i.e., experiencing the opposite outcome), tends to increase relative difference in rates of experiencing the outcome itself.  Correspondingly, reducing the outcome, while tending to increase the proportion the more susceptible group makes up of persons avoiding the outcome (thus reducing all measures of difference between the proportion the group makes up of the population and the proportion it makes up of persons avoiding the outcome), tends also to increase the proportion the group makes up of persons experiencing the outcome itself (thus increasing all measures of difference between the proportion the group makes up of the population and the proportion it makes up of persons experiencing the outcome). 

This page addresses the way innumerate observers are leading the state of Massachusetts to believe it has comparatively large racial disparities in adverse criminal justice outcomes without consideration of the extent to which comparatively large relative differences in adverse outcomes are a function of the prevalence of such outcomes in Massachusetts.  The page focuses on a September 2020 study by the Criminal Justice Project of Harvard Law School (HLS) titled “Racial Disparities in Massachusetts Criminal Justice System,” which was brought to my attention by February 23, 2021 article in the Harvard Gazette titled “Solving Racial Disparities in Policing.”  I will discuss both items further below after addressing how innumeracy in Massachusetts generally colors perceptions about racial and other demographic differences in the state. 

While having some of the world’s leading educational institutions, Massachusetts has long suffered from a serious innumeracy problem reflected in the failure to understand the ways measures of differences between outcome rates of advantaged and disadvantaged groups tend be affected by the prevalence of an outcome.  One manifestation of this innumeracy involves the failure to understand that because it is an affluent and otherwise advantaged state where adverse outcomes are comparatively uncommon, it tends to have comparatively large relative racial/ethnic differences in adverse outcomes, though comparatively small relative racial/ethnic differences in the corresponding favorable outcomes.  This was the particular subject of my “The Mismeasure of Health Disparities in Massachusetts and Less Affluent Places,” Quantitative Methods Seminar, Department of Quantitative Health Sciences, University of Massachusetts Medical School (Nov. 18, 2015) (Abstract) (UMMS Seminar). 

The statistical patterns addressed in the seminar were also the subject of my methods workshop titled “The Mismeasure of Group Differences in the Law and the Social and Medical Sciences,” Institute for Quantitative Social Science at Harvard University (Oct. 17, 2012) and in letters to arms of Harvard University of October 9, 2012, and October 26, 2012.  Pages 343-344 of my “Race and Mortality Revisited,” Society (July/Aug. 2014), are especially critical of Harvard both for the innumeracy of its research and guidance respecting health and healthcare disparities research and its obstinacy respecting that innumeracy. 

An irony of the situation in Massachusetts is that its many renowned universities contribute to the affluence of the state and the state’s predominant political philosophy, both of which tend to keep certain adverse outcome rates comparatively low, with corresponding comparatively large racial/ethnic differences in rates of experiencing the outcomes, while the innumeracy of those universities also contributes to mistaken beliefs about the meaning of those comparatively large relative differences.  In the case of health and healthcare disparities, the state’s exceptional (if largely innumerate) healthcare facilities contribute to the state’s comparatively low adverse health and healthcare outcome rates, with corresponding comparatively high relative racial/ethnic differences in rates of experiencing those outcomes.  Certain of these factors, however, may also tend to reduce racial differences, properly measured, which thus tends to reduce all measures of racial disparity, and which thus may sometimes cause the state to have comparatively small relative differences in adverse outcomes notwithstanding the tendency for the comparatively low adverse outcome rates to drive up those differences (as illustrated, for example, in Table E.4(b) (slide 106) of the UMMS Seminar).

Misperceptions in the state about its comparatively large relative racial difference (by both race and disability status) in school suspensions (which are also treated in Tables B1 and B2 (slides 67 and 68) of the UMMS Seminar) are addressed in Table 6 (at 7) of and in my “Measuring Discipline Disparities,” Written testimony for U.S. Commission on Civil Rights Briefing “The School to Prison Pipeline: The Intersection of Students of Color with Disabilities” (Dec. 8, 2017), and the Massachusetts Disparities, subpage of the Discipline Disparities page.  That subpage also addresses the way that failure to understand that reductions in absolute racial differences in suspension rates do not mean that relative racial differences decreased led to the belief that general reductions in suspension in Massachusetts were accompanied by decreased relative racial difference in suspensions when such differences in fact increased.[i]  See also discussion in "Race and Mortality Revisited" (at 337-339) and the October 9, 2012 letter to Harvard (at 21-24) regarding the way the reliance on absolute differences between rates to measure healthcare disparities in the healthcare disparities component of Massachusetts Medicaid’s the pay-for-performance program will tend to cause the component to result in increased disparities.

See also my The ‘Feminization of Poverty’ is Misunderstood,”  (Plain Dealer, Nov 11, 1987) and slides 20-22 of the UMMS Seminar (regarding the fact that because is an affluent state compared with Mississippi the feminization of poverty is greater among whites in Massachusetts than it is among Blacks in Mississippi even though the proportion female-headed families make up of the white families in Massachusetts is much smaller than the proportion such families make up of Black families in Mississippi[ii]) and my “United States Exports Its Most Profound Ignorance About Racial Disparities to the United Kingdom,” Federalist Society Blog (Nov. 2, 2017) (discussing the way that a report on racial/ethnic differences in criminal justice disparities in the United Kingdom cites a Massachusetts diversion program as a means of reducing relative racial/ethnic differences in incarceration while at the same time citing data indicating why the program will tend to increase such differences.[iii])

A few decades ago, at a time when I was rather less skeptical of the competence of academics regarding quantitative matters, I had some hope that Massachusetts’ institutions might have an important role causing demographic differences to be better understood.   For some years, commencing in the middle 1999s, my “The Perils of Provocative Statistics,” Public Interest (Winter 1991) – one of my early explanations of the pattern whereby the rarer an outcome the greater tends to be the relative difference in experiencing it and the smaller tends to be the relative difference in avoiding it – was required reading in a course titled “Quantitative Reasoning and Statistical Methods for Planning I” a course that currently seems to be a requirement at Harvard’s Graduate School of Design.   For at least a few of those years, students were required to read the article twice during the term.  But sometime in what I believe were the early 2000’s the article disappeared from the course’s list of required or recommended readings.  Thus, I assume that students who take the course today leave it not knowing that it is even possible for relative differences in a favorable outcome and relative differences in the corresponding adverse outcome to change in opposite directions, much less that it tends to occur systematically, and a great many of them leave mistakenly believing that reducing an adverse outcome will tend to reduce, rather than increase, relative differences in rates of experiencing the outcome.  That same, of course, may be said of almost anyone with a Ph.D. in statistics. 

Further, in 2005 a post on Harvard’s Social Science Blog “Misreading Racial Disparities –- Beware of Rate Ratios,” gave a reasonable account of my “Race and Mortality,” Society (Jan./Feb. 2000), which could be considered at updating of “The Perils of Provocative Statistics.”  And both the 2012 Harvard workshop and the 2015 UMMS Seminar were well attended by persons who ought to be able the understand the issues.  Nevertheless, to my knowledge (and the 2005 blog post aside) no analysis of demographic differences by researchers at Massachusetts universities has ever shown a recognition that generally reducing an adverse outcome would be expected to increase, rather than reduce, relative racial differences in rates of experiencing the outcome or that comparatively large relative differences in adverse outcomes are to be expected in Massachusetts simply because adverse outcomes are comparatively rare in Massachusetts.

Further, to my knowledge, the New England Journal of Medicine (NEJM), which is published by the Massachusetts Medical Society, has never shown an awareness that is possible for different measure of demographic differences in health and healthcare outcomes to change in opposite directions.  And I take for granted that, to the extent that NEJM articles have reflected a view as to how general reductions in adverse health outcomes should affect relative racial differences in rates of experiencing the outcomes, they have invariably reflected the mistaken view that reducing the outcomes would be expected to reduce those differences.  Similarly, to my knowledge, none the journal’s many discussions of subgroup effects has reflected an awareness that it is even possible for a factor that affects the outcome rates of two groups with different baseline rates for the outcome to cause a larger proportionate change in the outcome rate for one group while causing a larger proportionate change in the opposite outcome rate for the other group, even though that will almost always be the case when the baseline rates for the two groups differ substantially.[iv]

The principal focus of this page, however, is the way that, based on the size of relative differences in adverse criminal justice outcomes, the aforementioned HLS study reflected the view that Massachusetts had particularly large racial/ethnic incarceration disparities.[v]  In particular, at page 4 the study relies on a 2016 Sentencing Project study titled “The Color of Justice Racial and Ethnic Disparity in State Prisons,” for its ranking (on the basis of relative differences in incarceration rates) Massachusetts as having the largest Hispanic-white difference in incarceration rate and the 13th larger Black-white difference in incarceration rates.  This is the same Sentencing Project study that I had discussed in “Racial Impact Statement Laws in New Jersey and Elsewhere,” Federalist Society Blog (Mar. 20, 2017), with respect to the failure to understand the connection between low incarceration rates in New Jersey and high relative racial differences in incarceration rates (and New Jersey’s passage of a racial impact law while mistakenly believing both that the state had especially large racial disparities and that a racial impact law would tend to reduce, rather than increase, relative racial difference in incarceration rates.  (The link to the Sentencing Project study provided in the HLS study (at 4 n.8) and the March 20, 2017 Federalist Society Blog post currently takes one to a 2021 version of the study, which contains some discussion of recent events.)   The 2016 study is one of the many documents in which the Sentencing Project has promoted the mistaken belief that generally reducing adverse criminal justice outcomes will tend to reduce relative racial differences in rates of experiencing the outcomes.  I discuss another such Sentencing Project document in “The Pernicious Misunderstanding of Effects or Policies on Racial Differences in Criminal Justice Outcomes,” Federalist Society Blog (Oct. 12, 2017),

In the October 12, 2017 Federalist Society Blog post I qualified some of the observations I made in the March 20, 2017 post with respect to the correlation between New Jersey’s low incarceration rates and large relative racial difference in incarceration rates.  The relative difference between rates (or any other measure of difference between rates) is a function of the frequency of the outcome and the strength of the forces causing the adverse outcome.  That would apply to Massachusetts as well.  And I have not examined information that might be available to explore the extent to which the comparatively large relative racial/ethnic differences in incarceration rate in Massachusetts are functions of the frequency of incarceration in Massachusetts and the extent to which they may functions of other factors.

The fact remains, however, that it is difficult to say anything useful about demographic difference and the factors contributing them (or how policies affect them) without understanding the ways measures tend to be affected by the prevalence of an outcome and certainly not while mistakenly believing that policies will tend to reduce certain measures when they in fact tend to increase those measures.  Notably, the HLS study (at 18) relies on the smaller racial disparity in traffic stops during hours of darkness of evidence of the role of racial bias in traffic stops.  The drawing of inferences based on the comparative size of a relative difference (or a measure of disparity that is a function of the relative difference with respect to the comparative size of a disparity) disparity without understanding the effects of the prevalence of an outcome is the subject of the Drawing Inferences subpage of the Criminal Justice Disparities page. 

As reflected in the UMMS Seminar, the Harvard workshop, "Race and Mortality Revisited" and the references they cite, there are myriad types of data illustrating why relaxing a standard or otherwise reducing an adverse outcome will tend to increase relative racial differences in rates of failure to meet the standard at the same time that it reduces relative differences in rates of meeting the standard.  The data on recidivism risk presented on the Recidivism Illustration subpage of the Scanlan’s Rule page of jpscanlan.com nicely illustrate why relaxing the standard for receipt of any favorable criminal justice outcome (whether pretrial release, diversion, or a favorable type of sentencing), while tending to reduce relative racial differences in rates of experiencing the favorable outcome, tends to increase relative racial differences in the corresponding adverse outcome.  The same point can be illustrated with data in Figure A5 (at 94) of the HLS Massachusetts study showing Black and white rates of falling into various criminal history categories.

Table 1 show the rates at which Black and whites fall within or below (i.e., having a history that is equal to or worse than that in the category) three criminal history categories, along with the ratios of the Black rate of falling within each grouping and the ratio of the white rate of not failing into the grouping to the Black rate of not falling into the grouping.  And it show the usual pattern whereby the rarer the outcome (or, more precisely, the more it is restricted toward the end of the overall distribution) the greater is the relative difference in experiencing and the smaller is the relative difference in avoiding it.   The data thus show how increasingly restricting the adverse outcomes to defendants in with more serious criminal histories will tend to increase relative racial differences in the rates of experiencing the adverse outcomes while reducing relative racial differences in the corresponding favorable outcomes.[vi]  

 Table 1.  Black and white rates of falling into groupings of criminal history categories, with measures of difference (based on Figure A5 of Harvard Law School (HLS) study “Racial Disparities in Massachusetts Criminal Justice System”) 

 The Harvard Gazette article, which I think can be fairly regarded as promoting the mistaken belief that generally reducing adverse criminal justice outcomes with tend to reduce relative racial/ethnic differences in rates of experiencing the outcome even though it does not specifically state the view, contains several things that warrant mention.  The article discusses the work of one HLS professor that regards the criminalization of misdemeanors as a driver of racial disparity.  I leave the discussion of the role of misdemeanor to the Offense Type Issues (CJD) subpage of the Criminal Justice Disparities page, which discusses, inter alia, that it may be difficult to predict the impact of decriminalizing certain types of conduct on overall relative racial differences in adverse criminal justice outcomes even though the general reductions in adverse outcomes for any type of conduct increase relative differences in rates of experiencing adverse outcomes for that particular type of conduct.  The article also discusses another Harvard professor’s the view that implicit bias training has limited effect in reducing racial bias.  With respect to the matter, see the Implicit Bias Training subpage of the Criminal Justice Disparities page regarding the impossibility of evaluating the effects of policy or activity that might reduce all measures of racial disparity without understanding the way that general changes in the prevalence of an outcome are affecting each measure of disparity.  That is, for example, while reduction in racial bias tends to reduce relative racial differences in adverse outcome resulting from such bias, that effect may not be sufficient to counter the tendency for a general decrease in the outcome to increase the relative difference in rates of experiencing the outcome.

The article also discusses the same professor observations regarding the Department of Justice’s Ferguson report’s criticism of the city’s reliance on traffic fines to balance the city’s budget,  With regard to that matter, see “Usual, But Wholly Misunderstood, Effects of Policies on Measures of Racial Disparity Now Being Seen in Ferguson and the UK and Soon to Be Seen in Baltimore,” Federalist Society Blog (Dec. 4, 2019) regarding how generally reducing traffic stops – in this case, while specifically believing that doing so would tend to reduce relative racial differences in traffic stops –increased relative racial differences in traffic stops both in Ferguson and in the state of Missouri generally.  For one of the simplest explanations of the pertinent statistical pattern in the criminal justice context, see my “Things DoJ doesn’t know about racial disparities in Ferguson,” The Hill (Feb. 22, 2016).  See also the  Innumerate Decree Monitors subpage the of Criminal Justice Disparities page regarding the failure of the monitors of the City of Ferguson and Baltimore Police decrees to understand that the modifications to policies required by the decrees will tend to increase the measures of racial disparity on which the Department of Justice relied in bringing the underlying suits.

Finally, with respect to the general numeracy of lawyers from HLS or elsewhere, countless of whom have misled courts and the public about the effects of policies on measures of racial disparity, see Section B (at 12-14) of Response of J. Scanlan to Office of Management and Budget Request for Information “Methods and Leading Practices for Advancing Equity and Support for Underserved Communities Through Government” (FR Doc No: 2021-09109) (July 6, 2021).  The section addresses the failure or courts ever to understand that it is even possible for relative difference in rates of experiencing an outcome and relative differences in rates of avoiding the outcome to change in opposite direction, as well as the failure of both the majority and dissent in Brnovich v. DNC to understand that less stringent voter requirements tend to show larger relative racial/ethnic differences in rates of failure to meet the requirements than more stringent requirements.  The material is usefully read with reference to the Response’s Tables 3 and 4 (at 7-8), which show indisputably why relaxing income and credit score requirements to secure a loan or any other desired outcome will tend to increase relative racial differences in rates of failure to meet the requirements.  Yet, every lawyer dealing with fair lending issues, like every other person dealing with such issues, proceeds on the mistaken belief that relaxing lending standards will tend to reduce relative racial differences in rates of failing to meet the standards.  But see the Appendix to Memorandum of J. Scanlan to HUD September 22, 2020 Expert Panel (Sept. 19, 2020, updated Jan. 15, 2021) (which is among the attachments to Response on regulations.gov and is also available here) regarding the Department of Housing and Urban Development’s oblique recognition of the issue in its September 24, 2020 Final Rule on the Fair Housing Act’s Disparate Impact Standard.