Disparate risks of conviction under Michigan’s felony HIV disclosure law: An observational analysis of convictions and HIV diagnoses, 1992–2010

Punishment & Society

2015, Vol. 17(1) 73–93

! The Author(s) 2015

Reprints and permissions:

sagepub.co.uk/journalsPermissions.nav

DOI: 10.1177/1462474514561711

pun.sagepub.com

Article

Disparate risks ofconviction underMichigan’s felony HIVdisclosure law: Anobservational analysisof convictions and HIVdiagnoses, 1992–2010

Trevor Alexander HoppeUniversity of California at Irvine, USA

Abstract

Recent public debates on race and crime have reignited an interest in the criminology

literature on the effect of victim characteristics on criminal justice outcomes. This

article examines whether defendant and complaining witness demographic characteris-

tics are associated with disparate outcomes under Michigan’s felony HIV disclosure

statute, which makes it a crime for HIV-positive individuals to have sex without first

disclosing their HIV-status. Despite indications nationwide that the number of defend-

ants charged under HIV-specific criminal statutes (HSCS) is increasing, few empirical

studies have examined their application. This study of HSCS convictions (N¼ 51) retro-

spectively observes and compares the risks of conviction under Michigan’s HSCS for

particular HIV-positive subpopulations between 1992 and 2010. Overall, a compara-

tively greater risk of conviction was observed for black men with female partners and

white women overall. Contrary to expectations, a comparatively low risk of conviction

was observed for men with male partners as compared to men with female partners.

While the observational methodology employed in this analysis cannot identify the

causal mechanisms driving the observed disparities, the findings nonetheless suggest

an uneven application of Michigan’s HIV disclosure law deserving of further inquiry.

Keywords

complainant, discrimination, HIV/AIDS, public health, risk

Corresponding author:

Trevor Alexander Hoppe, University of California at Irvine, 2340 Social Ecology II Irvine, CA 92697-7080,

USA.

Email: [email protected]

by guest on January 27, 2015pun.sagepub.comDownloaded from

http://pun.sagepub.com/

Introduction

Socio-legal scholars have a longstanding interest in explaining discrimination in thecriminal justice system. While conceptual approaches to studying discriminationunder the law vary, criminologists have a longstanding interest in analyzing asso-ciations between a variety of criminal justice outcomes and the demographic char-acteristics of both defendants (see, for example, Koons-Witt, 2002; Steffernsmeieret al., 1995) and victims (see, for example, Baldus et al., 1983; Holcomb et al.,2004). This article builds on these insights by examining whether defendant andcomplainant characteristics are associated with disparate observed, population-level risks of conviction1 under Michigan’s felony HIV disclosure law, whichmakes it a crime for HIV-positive individuals to have sex without first disclosingtheir HIV-status.

Over 30 US states currently have similar HIV-specific criminal statutes (referredto hereafter as ‘‘HSCS’’) on the books (for a review, see Lehman et al., 2014). Themajority of these laws were enacted in the late 1980s and early 1990s when little wasknown about the disease, and they almost invariably criminalize a wide range ofbehaviors irrespective of the risk those behaviors pose for transmitting the disease(Galletly and Pinkerton, 2006). In Michigan, for example, a recent study docu-mented a 2009 case in which an erotic dancer was convicted of not disclosing herHIV-positive status to a client before his nose penetrated her vagina – a behaviorthat could not plausibly have transmitted the virus (Hoppe, 2014).

Notably, although the CDC (2013a) estimates that 15.8 percent of HIV-positivepeople do not yet know they are infected, an individual must have first beendiagnosed as HIV-positive by a medical provider before he or she could be prose-cuted under HSCS. The CDC (2013b) has required that state public health depart-ments maintain precise, names-based surveillance data regarding the remaining84.2 percent of diagnosed HIV-positive people since at least 2008. In many cases(including Michigan), state health officials began collecting names-based HIV sur-veillance data before the state’s HSCS was enacted.

These factors create a unique legal context, in which a great deal is known aboutthe narrowly delimited population of HIV-positive people who are governed byHSCS. This study takes advantage of this unique context in order to estimateand compare the observed, population-level risks of conviction for particularHIV-positive subpopulations by comparing two sources of data. First, I draw onan original dataset describing 95 percent of convictions (N¼ 51) under Michigan’sHSCS between 1992 and 2010 – including the defendant’s race, gender, and thegender of the complainant(s) in the case. I compare these data against MichiganDepartment of Community Health data documenting new HIV-positive diagnoses2

in the state’s Lower Peninsula during the study period – including the individual’srace, gender, and the gender(s) of their reported partners.

While scholarly interest in HIV disclosure laws and their application hasincreased in recent years, there is not yet a rich empirical literature analyzingtheir enforcement. For example, although critics have charged that these lawsare homophobic and racist (Strub, 2010), we do not actually know whether

74 Punishment & Society 17(1)



sexual and/or racial minorities are disproportionately convicted under the law.While this study’s observational methodology cannot identify the causal mechan-isms driving the disparate impact observed, this study nonetheless contributes toour understanding of the impact of HIV disclosure laws and the implications suchlaws may have for inequality.

Documenting and explaining disparate outcomes in thecriminal justice system

There is a long tradition in criminology of documenting disparate criminal justiceoutcomes between various groups – often distinguished as either ‘‘disparate treat-ment’’ or ‘‘disparate impact’’ studies (Lucas, 2008). Disparate treatment studies onrace examine racist practices that explicitly treat groups differently – such as racialbias in criminal sentencing (see, for example, Crawford et al., 2006; Doerner andDemuth, 2010; Kautt, 2009). Disparate impact studies, on the other hand, examinepractices that are not explicitly related to race, gender, or some other legallyprotected identity but nonetheless result in disparate outcomes, such as the well-documented disparate impact of the ‘‘war on drugs’’ on racial minorities (for areview, see Provine, 2011).

While racial discrimination is perhaps the most widely studied form ofdiscrimination, a wide range of social factors have come under the scrutiny ofsocio-legal scholars interested in discrimination. This has typically involvedexamining demographic characteristics of the defendant, such as their age(Steffernsmeier et al., 1995) and gender (Koons-Witt, 2002) and the interaction ofthese variables for black men in particular (Doerner and Demuth, 2010;Steffernsmeier et al., 1998).

While far less numerous, studies have also examined whether the characteristicsof crime victims – rather than just defendants – are also associated with criminaljustice outcomes. In particular, numerous studies have examined how the genderand race of violent crime victims impacts the imposition of the death penalty,finding that black defendants accused of killing white victims receive the deathpenalty at higher rates than others (see, for example, Baldus et al., 1983;Holcomb et al., 2004). Similar studies of victim characteristics in non-capitalcrimes and their adjudication are less numerous. Examining a broader setof crimes, Curry (2010) finds that the trend of harsher punishment for homi-cide cases involving white victims did not extend to sexual assault and robberycases.

Evidencing the causal mechanisms underlying these varied, disparate outcomeshas been more problematic; socio-legal scholars have noted that interpreting suchtrends as racist often conflict with conventional notions of racial prejudice that areimplicitly understood to be intentional (Murakawa and Beckett, 2010). Myriadstudies have documented various mechanisms that could be driving disparateoutcomes. These include discriminatory attitudes and practices among police(Beckett et al., 2006; Bernstein and Kostelac, 2002; Petrocelli et al., 2004);

Hoppe 75



prosecutors (Radelet and Pierce, 1985), judges (Curran, 1983), and juries (Mazellaand Feingold, 2006; Schuller and Hasting, 2002). Other studies have examined‘‘supply-side’’ factors that result in fewer criminal complaints. For example, studieshave shown that gays and lesbians are less likely to report crimes to the police thanheterosexuals (Miles-Johnson, 2013) and that crime is less likely to be reported tothe police when the offender is white or when the victim is male or white (Hart andRennison, 2003).

The examples cited in this brief review represent but a fraction of the vastliterature on the subject and are intended to familiarize the reader with currentconceptual approaches to explaining discrimination in the criminal justice system.In this article, I extend the insights of these scholarly efforts by evaluation whetherdefendant characteristics and partner gender (that is, the gender of the sexualpartner in cases of non-disclosure of HIV-positive status) are associated with dis-parate risk of conviction under Michigan’s HIV disclosure law.

HIV disclosure laws in the United States, Michigan

Criminal laws in 24 US states currently impose misdemeanor or felony penalties onHIV-positive people who have sex without first disclosing their HIV-positive statusto their sexual partners. Including states that have sentence enhancement policiesfor other crimes when HIV risk is involved (for instance, rape or prostitution),32 states currently have an HSCS on the books. While these laws vary intheir specifics, none requires that the virus be transmitted and most do notdifferentiate sexual contact that poses no or low risk of transmission from contactthat poses a significant risk (Bernard and Nyambe, 2012; Center for HIV Law &Policy, 2010).

Of the 24 states that classify non-disclosure as a felony or misdemeanor, themajority (14) enacted their HSCS during the mid-1980s and early 1990s in thecontext of high HIV/AIDS mortality rates, relatively low levels of public know-ledge about the disease and a general anxiety about how it was transmitted,and before life-saving medications known as antiretrovirals were introduced in1995 (Burris et al., 1993; Galletly and Pinkerton, 2006). Based on their historicalcoincidence, legal scholars have suggested that the President’s Commission onthe HIV Epidemic’s final report issued in 1988 and the Ryan White Care Act of1990 prompted state legislatures to enact these statutes (Lahey, 1995; McArthur,2009). Enacted before either of these policy efforts, Michigan’s felony HIV disclos-ure law was voted into law by legislators in 1988 and specifically states:

1) A person who . . . knows that he or she is HIV infected, and who engages in sexual

penetration with another person without having first informed the other person . . . is

guilty of a felony. 2) As used in this section, ‘‘sexual penetration’’ means sexual

intercourse, cunnilingus, fellatio, anal intercourse, or any other intrusion, however

slight, of any part of a person’s body or of any object into the genital or anal openings




of another person’s body, but emission of semen is not required. (Michigan Compiled

Law [MCL] § 333.5210)

Thus, the statute does not require evidence of transmission and criminalizes a widerange of sexual practices that, according to records from a conviction as recently as2009, involved an HIV-positive woman convicted in 2009 for allowing a man to‘‘penetrate’’ her vagina with his nose without first disclosing (Hoppe, 2014).Additional evidence suggests that local health officials in Michigan employ epi-demiological surveillance technologies to assist in identifying HIV-positiveclients who are not disclosing, and have enacted policies that can facilitate theirprosecution under the criminal law (Hoppe, 2013).

US policy debates over criminal prosecutions for non-disclosure of HIV infec-tion have tended to focus on atypical cases involving multiple partners who areunderstood to be particularly vulnerable. Perhaps the most widely cited caseinvolved Nushawn Williams, a black male defendant accused in New Yorkof infecting at least nine women (some of whom were underage and white).Without a criminal HIV disclosure law on the books, New Yorkprosecutors instead prosecuted Williams for having sex with minors (Shevory,2004). As Shevory argues, the frenzied media spectacle surrounding the casereflected social anxieties not just about HIV, but also about race and crime moregenerally.

A handful of studies have analyzed attitudes toward HIV disclosure laws and,specifically, whether not disclosing one’s HIV-positive status should be a crime(Galletly et al., 2012a, 2012b; Horvath et al., 2010). These studies have tended todraw convenience samples of HIV-positive residents and/or men who have sex withmen, and their specific questions have varied, thus making it difficult to collectivelyinterpret their findings. While these studies have consistently found high levels ofsupport for criminalizing non-disclosure in at least some scenarios, no representa-tive study has yet attempted to analyze whether demographic characteristics areassociated with attitudes toward HSCS and, more specifically, one’s willingness topursue criminal charges against potential offenders. As I suggest later in the article,such variations (if they exist) might help to explain the disparate impact ofMichigan’s disclosure law.

While large-scale, representative studies analyzing the enforcement of HSCS donot yet exist, Galletly and Lazzarini (2013) examine arrests under Tennessee’sHSCS in Nashville courts between 2000 and 2010. Notably, Tennessee’s HSCS isbroadly defined to criminalize any ‘‘HIV exposure’’, including non-sexual expos-ures such as biting and spitting. Most individuals arrested for HIV exposure weremen (74 percent) and white (56 percent). Of the 15 cases involving adult sexualrelationships (as opposed to spitting and biting cases), 12 (80 percent) involved menor women with opposite-gendered partners. Notably, despite the disparate lawenforcement contexts, many of these findings echo those presented in this articleand are thus highlighted.

Hoppe 77



Expectations

This study examines a series of hypotheses to understand the impact ofvarious social factors on the risk of conviction observed under Michigan’s HIVdisclosure law. Based on the myriad findings demonstrating the vulnerability ofracial minorities to incarceration in the USA, the first hypothesis tests the inde-pendent effects of race on the observed risk of conviction:

Hypothesis 1: The risk of conviction under Michigan’s HIV disclosure law

between 1992 and 2010 was greater for HIV-positive black people as compared to

whites.

A second hypothesis aims to test whether the independent effects of gendersimilarly follow from expectations that men are more likely to face criminal pros-ecution than women:

Hypothesis 2: The risk of conviction under Michigan’s HIV disclosure law between

1992 and 2010 was greater for HIV-positive men as compared to women.

Based on findings that black men and women in particular face a disparateimpact under the law as compared to their white counterparts, the third hypothesistests the joint effects of race and gender:

Hypothesis 3a: The risk of conviction under Michigan’s HIV disclosure law between

1992 and 2010 was greater for HIV-positive black men as compared to other race–

gender groupings.

Hypothesis 3b: The risk of conviction under Michigan’s HIV disclosure law between

1992 and 2010 was greater for HIV-positive black women as compared to other race–

gender groupings.

Informed by critics of these laws who suggest they are motivated by anti-gaybias, the fourth hypothesis tests the effect of the partner’s gender for HIV-positivemen.3


1992 and 2010 was greater for HIV-positive men with male partners as compared to

HIV-positive men with female partners.

Building on these expectations, the fifth hypothesis tests whether the partner’sgender modifies the effect of race on men’s risk of conviction:


1992 and 2010 was greater for HIV-positive black men with male partners as com-

pared to HIV-positive black men with female partners and other groupings.




Study design, data, and analysis

I obtained public records of convictions from the Michigan State PoliceInformation Center, which include the county, sentencing date, and court dispos-ition of the 48 convictions reported to the state between 1992 and 2010.4 In order toanalyze each case, including the social identity of those convicted of violatingMichigan’s HIV disclosure statute, I attempted to identify the defendant in eachcase by visiting county libraries in which local newspaper archives were housed. Isearched for any mention of a case in newspaper records within a two-week periodof the sentencing date indicated in the state police data. I successfully identified thedefendant in 29 cases this way. I identified the defendant in an additional 29 casesusing county clerks’ records and internet sites documenting criminal records –including 13 cases not documented in the state police records.5 In total, I identifiedthe defendant in 58 (95 percent) of the 61 known convictions between 1992 and2010. These 58 cases included 51 discrete defendants, as several defendants wereconvicted multiple times. I obtained court records that revealed the basic facts foreach case, including the defendant’s demographic characteristics and the complain-ant’s gender. This research design was reviewed by the University of MichiganHealth Science and Behavioral Sciences Institutional Review Board and deter-mined to be exempt from review.

In order to examine differences between the demographic characteristics of con-victed defendants and HIV-positive residents generally, I requested data from theMichigan Department of Community Health detailing individuals diagnosed asHIV-positive any time between 1992 and 2010 in all Lower Peninsula6 counties.Notably, in cases in which the health department data indicated ‘‘>5’’ HIV diag-noses instead of specifying a definite number of individuals for a particular risk-group, I substituted either an indirectly determined precise figure or an averageestimator. For example, if the data indicated that seven HIV-positive women werediagnosed between 1992 and 2010 in a particular county and that five of thoseHIV-positive women were white and zero were ‘‘other’’, then necessarily two wereblack because these categories were mutually exclusive. In other instances where itwas not possible to deduce the precise figure, an average of the possible values wassubstituted. For example, if the data indicated that eight HIV-positive womenwere diagnosed in a particular county, and that four of them were white, <5were black, and <5 were ‘‘other’’, I used the average of the minimum and max-imum possible values as an estimator. In this example, since the sum of the possiblevalues could not exceed four, the estimate substituted would be the average of 1and 3 or 2.0. Of the 352 data points for the 22 jurisdictions analyzed, it wasnecessary to substitute 25 such estimates. Because the estimators were necessaryonly in cases with very small numbers, this strategy has a negligible effect on thevalidity of the overall estimates for HIV-positive diagnoses in these jurisdictions.

For the purposes of this analysis, population-specific risks of conviction arepresented both with and without data from Wayne County (which includesDetroit). Wayne County is an exceptional case for many reasons, and presentingthe data without distinguishing its effects muddles efforts to interpret the findings.

Hoppe 79



While Wayne County represents 62.19 percent of all HIV-positive diagnoses duringthe study period, it represents only 15.52 percent of convictions under the disclos-ure law. Thus, diagnosis in Wayne County is clearly negatively associated with thedependent variable in this analysis (the risk of conviction). Further, 82 percent ofthe individuals diagnosed as HIV-positive in Wayne County between 1992 and2010 are black, compared to 36.4 percent of those diagnosed as HIV-positive inthe remainder of the state. In this sense, diagnosis in Wayne County is also asso-ciated with the independent variable of race used in this analysis.

Because it is associated with both an independent (e.g. race) and the dependent(e.g. risk of conviction) variables in this analysis, diagnosis in Wayne County canbe understood as a spurious, ‘‘lurking’’ variable that confounds the associationsobserved in some of the analyses in this article. While it is not possible to ‘‘control’’for these effects using the methods employed in this article, analyzing data exclud-ing Wayne County allows for its effects to be disentangled from other factors in therest of the state. As the number of convictions in Wayne County on its own is sosmall (n¼ 9), an analysis of its figures alone would be extremely crude and unlikelyto yield meaningful results. As such, each analysis includes figures both includingand excluding Wayne County. The populations for the analyses including WayneCounty include 51 convicted defendants (representing 58 criminal cases) and 15,659HIV-positive diagnoses. The populations for the comparisons excluding WayneCounty include 42 defendants (representing 49 convicted cases) and 5921 HIV-positive diagnoses.

While the number of convicted defendants is small, both defendant data andhealth department data can be treated analytically as populations because theyrepresent nearly 100 percent of all possible cases during the study period. Theconviction data include 58 out of 61 cases (95.08 percent) known to have beenconvicted in the state between 1992 and 2010. Notably, the number of convictionswas fairly consistent over the study period: there were 13 convictions between 1991and 1995; 10 between 1996 and 2000; 18 between 2001 and 2005; and 17 between2006 and 2010. The demographic characteristics of these cases also did not varysubstantially over time. The cases were geographically diverse, representing 28 ofthe 68 counties in Michigan’s Lower Peninsula. For a more detailed description ofthese cases, see Table 1 in Hoppe (2014).

Regarding the health department data, the law requires that informationon anyone who is diagnosed as HIV-positive be reported to the state. This hasbeen the case for the entire study period. This includes individuals who werediagnosed first out of state and later moved to Michigan, as any HIV lab-workdone after their arrival would fall under the same legal reporting requirementsas a new diagnosis. While there are individuals who were HIV-positive but notdiagnosed as such during this time, they are not relevant to this analysis becauseone must be diagnosed as HIV-positive in order to be prosecuted for failing todisclose.

Because both sets of data include nearly the entire population, statistical signifi-cance tests are unnecessary because the probability that the differences observed




between the two data sets were the result of random error (or ‘‘chance’’) due tosampling is negligible. Statistical significance figures for the analyses conducted inthis article would estimate the probability that the differences observed betweengroups were the result of random error due to sampling rather than actual differ-ences in the population. Because these two sets of data represent entire populations(or, in the case of the conviction data, 95 percent of the entire population), stat-istical significance tests are inappropriate.

Findings

The independent effect of race

The first hypothesis predicted that a higher risk of conviction would be observedfor blacks as compared to their white counterparts. As demonstrated by the datapresented in Table 1, this hypothesis is easily rejected for the analysis excludingWayne County. Overall, this analysis estimates a risk of conviction of approxi-mately 71 convictions under the HIV-disclosure law per 10,000 HIV-positivediagnoses during the study period. The differences between racial groups in thisanalysis are negligible: 70and 72 for black and white HIV-positive diagnoses,respectively.

As described earlier, diagnosis in Wayne County confounds this relationship.Because of the spurious effect of Wayne County, including its data would seem tosuggest that whites – not blacks – faced a greater risk of conviction than othergroups. Indeed, the conviction risk observed for whites is three times greaterthan that observed for blacks (60 convictions versus 19 convictions per 10,000HIV-positive diagnoses). However, this apparent association between race and

Table 1. Independent effects of race, 1992–2010

Race

Convicted defendants HIV-positive diagnosesConvictions per 10,000

HIV+ diagnosesaFrequency Percenta Frequency Percenta

Excluding Wayne County

White 23 55 3209 54 72

Black 15 36 2155.5 36 70

Other 4 10 555.5 9 72

N 42 5920 71

Including Wayne County

White 27 53 4485 29 60

Black 19 37 10142.5 65 19

Other 5 10 1030.5 7 49

N 51 15658 33

Note: aFigures are rounded to the nearest whole number for clarity.

Hoppe 81



risk of conviction is spurious and entirely explained by diagnosis in Wayne Countywhere the risk of conviction among its large, mostly black population is compara-tively very low.

The independent effect of gender

The second hypothesis predicted that a higher risk of conviction would be observedfor HIV-positive men as compared to their female counterparts. As demonstratedby the data presented in Table 2, this hypothesis is rejected for both comparisons.Excluding Wayne County, this study estimates a risk of conviction during the studyperiod of 70 convictions per 10,000 HIV-positive male residents, and finds a similarrisk for female residents of 75 convictions per 10,000 HIV-positive female residents.The estimated risk of conviction including Wayne County is lower overall(33 convictions per 10,000 HIV-positive diagnoses), but does not differ greatlyby gender. Thus, gender does not seem to have an independent effect on theestimated risk of conviction for HIV-positive Michiganders.

The joint effects of race and gender

While no clear evidence exists that race and gender on their own are associated witha disparate impact under Michigan’s HIV disclosure law, the third hypothesispredicted that these factors would be jointly associated with disparities. In particu-lar, I expected to observe a heightened risk of conviction for black men as com-pared to other groupings. Based on the analysis excluding Wayne County, there isevidence to support this hypothesis (see Table 3). While this analysis estimates arisk of conviction for white men of 57 convictions per 10,000 diagnoses outside ofWayne County, the risk for black men is more than 60 percent greater at 94. Just asin the previous analysis of the independent effects of race, diagnosis in Wayne

Table 2. Independent effects of gender, 1992–2010

Gender




Male 33 79 4714.5 80 70

Female 9 21 1205.5 20 75

N 42 5920 71


Male 40 78 11903.5 76 34

Female 11 22 3754.5 24 29

N 51 15658 33





County confounds this relationship, resulting in estimates that would suggest ahigher risk of conviction for white men at 46 convictions per 10,000 diagnosesversus 25 for black men.

However, the parallel hypothesis that predicted black women would also face ahigher risk of conviction is not just rejected but inverted whether or not WayneCounty data are included. While this study observed a risk of conviction for blackwomen of 15 convictions per 10,000 HIV-positive diagnoses (three when includingWayne County), white women faced a risk more than 10 times greater at 172convictions per 10,000 diagnoses (151 when including Wayne County). Despiterepresenting 23 percent of all Wayne County HIV-diagnoses, not a single blackwoman is known to have been convicted under the HIV disclosure law in WayneCounty during the study period. While this is consistent with the general trendtoward a lower risk of conviction within Wayne County, it is nonetheless note-worthy (especially given that two white women are known to have been convictedin Wayne County and yet they represent less than 2 percent of all Wayne CountyHIV-positive diagnoses). In both analyses, white women were observed to havefaced the greatest risk of conviction of any group analyzed, suggesting that theymay face a particular burden under the HIV disclosure law that deserves furtherattention.

Table 3. Joint effects of race and gender, 1992–2010

Race–Gender




White Men 16 38 2802 47 57

White Women 7 17 407 7 172

Black Men 14 33 1492.5 25 94

Black Women 1 2 663 11 15

Other Men 3 7 420 7 71

Other Women 1 2 135.5 2 74

N 42 5920 71


White Men 18 35 3887 25 46

White Women 9 18 598 4 151

Black Men 18 35 7237.5 25 25

Black Women 1 2 2905 3 3

Other Men 4 8 779 5 51

Other Women 1 2 251.5 2 40

N 51 15658 33


Hoppe 83



The modifying effect of the partner’s gender

I hypothesize in this article that the gender of the complainant in HIV disclosurecases may be associated with disparities conviction outcomes. Based on activistcritiques charging that these laws are driven by anti-gay prejudice, I expected tofind a heightened risk of conviction among men with male partners as opposed tomen with female partners. As shown in Table 4, this hypothesis can be uniformlyrejected. HIV-positive men with female partners face a higher risk of conviction inboth analyses – nearly four times as high when excluding Wayne County(110 versus 30) and more than double when including it (44 versus 18). In bothanalyses, the estimated risk of conviction for men who have sex with women(MSW) is higher than the mean while the estimated risk for men who have sexwith men (MSM) is consistently lower.

As shown in Table 5, this association extends to the impact of race on the risk ofconviction. The previous analysis revealed that the partner’s gender was associatedwith a disparate impact under the HIV disclosure law. In this analysis, this differ-ence is found to be consistent across all three racial categories and is consistentwhether or not Wayne County is included in the comparison. The risk of convic-tion is higher for black, white, and other MSW than for their MSM counterpartsboth when including and excluding Wayne County – although the estimated dif-ference is most dramatic for black men. For example, whereas 155 convictions areestimated for every 10,000 HIV-positive black MSW when excluding WayneCounty, only 11 convictions are estimated for every 10,000 HIV-positive black

Table 4. Modifying effect of partner’s gender on male convictions, 1992–2010

Partner’s

Gender

Convicted male defendants HIV-positive male diagnosesConvictions per 10,000

HIV-positive diagnosesaFrequency Percenta Frequency Percenta


Female 23 70 2084 38 110

Male 10 30 3348 62 30

N 33 5432b 61


Female 26 65 5900 43 44

Male 14 35 7765 57 18

N 40 13665b 29

Notes: aFigures are rounded to the nearest whole number for clarity.bThese analyses are based on risk group data from the health department. The number of total cases included

in these data is larger than in previous analyses (e.g. a total of 5432 cases are defined as male in the data

excluding Wayne County, whereas previous analyses indicated a total of 4714.5 male cases) because the

categories defining the gender of the partner are not mutually exclusive: some men have both female and male

partners. With this in mind, the figures presented here underestimate the risk of conviction for both groups.

However, because the overlapping cases are necessarily equally distributed among both groups analyzed here,

the estimated difference observed between the groups is not biased by the lack of interdependence.




MSM. Due to the association between Wayne County and the risk of conviction,these differences are smaller when including Wayne County (37 versus seven con-victions per 10,000 HIV-positive diagnoses), but they are nonetheless consistent.Among men outside of Wayne County, black MSM faced the least risk of convic-tion while black MSW face the greatest risk of conviction.

Discussion

As I have shown, there were far fewer HIV non-disclosure cases involving MSMdefendants in Michigan between 1992 and 2010 than would be expected given their

Table 5. Modifying effect of partner’s gender on race for male convictions, 1992–2010

Race–Partner

gender

Convicted male defendants HIV-positive male diagnosesConvictions per 10,000



White MSW 8 24 1031 19 78

White MSM 8 24 2214 41 36

Black MSW 13 39 840.5 15 155

Black MSM 1 3 880.5 16 11

Other MSW 2 6 212.5 4 94

Other MSM 1 3 253.5 5 74

N 33 5432b 61


White MSW 8 20 1434 10 56

White MSM 10 25 3040 22 33

Black MSW 15 38 4059.5 30 37

Black MSM 3 8 4270.5 31 7

Other MSW 3 8 406.5 3 74

Other MSM 1 3 454.5 3 22

N 40 13665b 29

Notes: aFigures are rounded to the nearest whole number for clarity.bWhile these analyses are based on the same denominator for HIV-positive diagnoses as the previous set, the

overlapping cases (e.g. men who reported female and male partners at diagnosis) in these analyses are not

necessary equally distributed across racial groups. In the first analysis excluding Wayne County, for example,

while 24.12 percent of men diagnosed overall reported male and female partners, 23.38 percent of white men,

27.10 percent of black men, and 18.45 percent of other men diagnosed as HIV-positive between 1992 and

2010 reported both male and female partners. The proportional distribution is similar for the analysis

including Wayne County (25.42 percent overall; 22.69 percent, 27.32 percent, and 21.37 percent for

white, black, and other, respectively). Because of this, the estimated differences observed between the

racial categories in both analyses may be biased. Specifically, the estimated number of convictions for black

men is slightly overestimated as compared to the white and other estimates, while the estimate for other men

is slightly underestimated as compared to the white and black estimates. However, the differences observed

between MSM and MSW within racial groupings are unbiased by this distribution.

Hoppe 85



vast overrepresentation among those diagnosed as HIV-positive during that sametime. Similarly, the inverted racial dynamic for men and women observed suggeststhat black men and white women are at a comparatively higher risk of convictionunder the law than white men and black women, respectively. While these observedpopulation levels are suggestive of disparities under the law, this article does notexplain how individual cases come before the court or why they are adjudicated in aparticular fashion. Rather, the observational approach to studying risk is useful fordetermining whether there are population-level patterns that might point to under-lying factors shaping the application of the law. Before it is possible to concludethat the findings observed in this analysis are actually evidence of discriminationunder the law, more research is necessary.

With these limitations in mind, this study provides little evidence to suggest thatrace and gender are independently associated with disparate risks of convictionunder the state’s HIV disclosure law. There is some evidence to support the viewthat these factors are jointly associated with disparate risks of conviction, but thesefindings also do not readily conform to expectations. In particular, while theheightened risk of conviction observed for HIV-positive black men (outside ofWayne County) is in line with the vast literature documenting discriminatory pat-terns against black men in the US criminal justice system, the dramatically low riskof conviction observed for black women (both when including and excludingWayne County), and the comparatively high risk for white women, defyexpectations.

As I argue in this article, accounting for complainant characteristics may help toexplain disparate outcomes by bringing the identity of those who file charges intothe analysis. While advocates have expressed concerns that criminal HIV disclosurelaws could be used in a discriminatory way against gay men, this study observed asubstantially lower risk of conviction for MSM during the study period as com-pared to MSW. While more research is necessary to understand this finding thatheld consistent across racial groups, partner gender appears to be associated withlegal outcomes for HIV-positive men in the state.

Why might this be? Future research should examine the varied set of mechan-isms that could be shaping this disparate impact. Perhaps MSM are simply morelikely to disclose and are, thus, less likely to be prosecuted. However, the literatureon HIV disclosure suggests that it is relatively commonplace for HIV-positiveMSM to engage in sexual contact without first disclosing their HIV-status(see, for example, Holt et al., 2011; Kalichman and Nachimson, 1999; Prestageet al., 2001). Although no comparative study on HIV non-disclosure rates betweenMSM and MSW is known to the researcher, these findings suggest that variationsin non-disclosure are unlikely to entirely explain these differences.

However, while variation in behavior defined by the state as criminal may notexplain disparate rates of conviction, normative differences in how communitiesdefine non-disclosure may have an effect. For example, while it may be common forMSM to engage in sexual contact without disclosing one’s HIV-status, MSM maybe less likely to interpret that behavior as a criminal act deserving of punishment.




As they are often targeted for HIV prevention messages that encourage them totake responsibility for protecting themselves against HIV infection, they may beless likely to adopt the model of the victim requisite under the criminal justicesystem. Further, given the history of police persecution of same-sex sexualities,MSM may be less likely to report incidents of non-disclosure to law enforcementauthorities. Indeed, Amnesty International (2005: 3) concludes in a report on policebrutality against lesbian, gay, bisexual, and transgender (LGBT) Americans that‘‘in the U.S., LGBT people continue to be targeted for human rights abuses by thepolice based on their real or perceived sexual orientation or gender identity’’.Empirical studies back up these claims, suggesting, for example, that manypolice officers hold antigay attitudes (Bernstein and Kostelac, 2002); that manyLGBT victims of hate crimes do not report their crimes to police because they fearpolice mistreatment (Herek et al., 2002); and that reports of police misconduct andinaction in cases involving LGBT crime victims are persistent over time (Wolff andCokely, 2007). As such, MSM potential complainants in HSCS cases may not trustthe police to take their case seriously. Alternatively, they may be more likely tohave anonymous partners that they could not report to the police, even if theywished to do so.

The mechanisms underlying these trends may extend beyond ‘‘supply-side’’ fac-tors. There are a variety of discretionary factors that could be influencing theobserved trends. In particular, some readers may wonder whether the biases ofprosecutors, judges, and/or juries might result in fewer convictions in cases invol-ving gay male complainants because they view these cases as less deserving of legalintervention. This would echo findings that black men convicted of murderingwhite victims are more likely to receive the death penalty than those who areconvicted of murdering black victims. Although studies have documented discrim-inatory attitudes among law enforcement toward gays and lesbians, more researchis necessary to understand how that would impact the handling of HIV non-disclosure cases. While many plausible explanations exist, the evidence is clearthat Michigan’s HIV disclosure law is not disproportionately enforced againstMSM defendants.

The findings regarding white women are perhaps more vexing. Perhaps the factthat their partners may have been mostly white men may have motivated lawenforcement to punish defendants seen as endangering a ‘‘valuable’’ socialgroup. However, many of the nine white women convicted in these cases are espe-cially disadvantaged (for a detailed account of one such case, see ‘‘Sandra’’ inHoppe, 2014). Five were described in court as having substance abuse problems;three suffered from mental illness; three were engaged in sex work; and one washomeless. Only two of the nine women are not known to fit into any of thesecategories (by comparison, the lone black woman convicted under the disclosestatute in the study period was not described as suffering from any of theseissues). The vulnerabilities facing white female defendants in this study may havecontributed to their increased risk of conviction under the disclosure law. Many ofthe white women in this study had prior criminal records that would have made

Hoppe 87



plea bargains to lesser crimes less likely; their highly stigmatized practices (e.g.large number of sexual partners, engagement in sex work, etc.) may have madethem particularly attractive for prosecution and/or conviction to prosecutors,judges, and/or juries; and their engagement in other illegal activities may havesimply made interactions with the law more likely. This is especially true forthose engaging in sex work, since all three were originally picked up for engagingin sex work; HIV-related charges were added after their arrest.

Yet, while these vulnerabilities seem to in part explain the increased risk ofconviction for the white women in the study, we would expect the same vulner-abilities to afflict HIV-positive black women. How can the apparent protectiveeffect for black women against conviction be explained? This might in part bedue to the fact that their partners are most likely to be black men. Black menmay be less likely to report their partner’s non-disclosure to authorities, eitherbecause they are less likely to consider the failure to disclose a crime or becausethey are distrustful of law enforcement authorities. The evidence of mistreatment ofAfrican-Americans in the criminal justice system demonstrates a pervasivepattern of abuse and discrimination that has created a hostile and adversarialrelationship between law enforcement and black communities (see, for example,Alexander, 2010; Rios, 2011; Wacquant, 2010). Legal scholars argue that thesetrends extend beyond police and frontline law enforcement personnel to prosecu-tors who wield considerable discretionary power in deciding who to charge andfor what offense (for a review, see Davis, 1998). This combative legal con-text understandably discourages many black men from inviting the state intotheir lives.

All of these possible explanations deserve further inquiry. While this study wasonly able to evaluate the association between complainant gender and risk ofconviction, future studies might examine whether other demographic characteris-tics – such as their race or socioeconomic status (SES) – are associated withdisparate outcomes under the law.

Conclusion

In the wake of the fatal shooting of Trayvon Martin in Florida and his assailant’scontroversial acquittal under Florida’s ‘‘Stand Your Ground’’ statute, there hasbeen a renewed public interest in the relationship between victim characteristicsand criminal justice outcomes (see, for example, Hundley et al., 2012). Whilestudies on violent crime have demonstrated an association between victim–defend-ant characteristics and criminal justice outcomes, the enforcement of laws regulat-ing other non-violent forms of crime may also prove to be fruitful sites of inquiry.While the relatively smaller number of studies examining victim or complainantcharacteristics (versus simply defendant characteristics) is likely due to the fact thatstate agencies generally do not collect or report such information, it is possible tocollect such data independently and such efforts may prove fruitful for explainingtrends that otherwise elude explanation. While HIV disclosure cases are a unique




case in many ways, the methods employed in this analysis may be useful forunderstanding other statutes that are enforced largely through complainantreports.

While the complexity of the causal relationships inherent to criminal justice willmake conclusively demonstrating the mechanisms underlying these trends difficult,future studies might attempt to examine – at least indirectly – the plausibility ofsuch mechanisms. For example, large-sample attitudinal studies could evaluatewhether there are differences between various demographic groups in attitudestoward HIV disclosure laws and more generally the criminality of failing todisclose. If such differences were found and reflected expectations from studiessuch as this one on conviction outcomes (namely, that MSM are less likely toperceive non-disclosure as a crime and/or report such an event to authorities),this would indirectly support the argument that variations in potential complainantattitudes helps to shape the disparate impact of HIV disclosure laws.

In regards to HIV disclosure laws, this study contributes to our understandingof how these laws are implemented on the ground. While previous cases studieshave suggested that race and gender play a role in shaping these cases and theirtrajectories, until now it was not clear whether demographic characteristics wereassociated with disparate legal outcomes. As this study’s findings are limited toMichigan and may not be generalizable to other states with similar statutes, futureresearch might examine the extent to which these patterns are consistent acrossjurisdictions and what implications such disparities might have for law and socialinquiry.

Acknowledgements

The author would like to thank the members of his dissertation committee for their insightand gracious help in finalizing this manuscript for publication: David Halperin, Renee

Anspach, Sarah Burgard and Sandra Levitsky.

Funding

This analysis was made possible by grants from the Social Science Research Council; theUniversity of Michigan Institute for Research on Women and Gender; the Rackham

Graduate School; the Center for the Education of Women; the Department of Sociology;and the Department of Women’s Studies.

Notes

1. Notably, this concept should be distinguished from ‘‘conviction rates’’, which gen-

erally refer to the proportion of cases convicted versus the total number prosecuted.

As I employ it here, conviction risk refers to the proportion of cases convicted versus

the total number of individuals governed by the law.

2. Because all new HIV-positive diagnoses must be reported to the state by law, these

data describe the population of potential offenders and are thus useful for

comparison.

Hoppe 89



3. This cannot be tested for women, since there were no women accused of not dis-

closing their HIV-status to other women, which reflects the fact that HIV prevalence

among women who have sex with women is exceedingly low. Indeed, the data

provided by the Michigan Department of Community Health does not distinguish

women infected with HIV by their partners’ gender.

4. While the state data did include some information on non-convictions, it was limited

to only those cases in which the defendant had a prior criminal record. As such, they

are excluded from this analysis.

5. Of these 13 cases, 10 were sentenced in 2008 or later. Of the 48 cases reported by the

state, zero were sentenced after 2008. This suggests a delay in reporting to the state

or in the state’s aggregation of those data.

6. There are hardly any cases of HIV to report in the Upper Peninsula of Michigan.

Indeed, the state health department does not even regularly report such figures.

There were also no reported convictions in Upper Peninsula counties.

References

Alexander M (2010) The New Jim Crow: Mass Incarceration in the Age of

Colorblindness. New York: The New Press.

Amnesty International (2005) Stonewalled: Police Abuse and Misconduct against

Lesbian, Gay, Bisexual and Transgender People in the U.S. New York: Amnesty

International.

Baldus DC, Pulaski CA Jr and Woodworth G (1983) Comparative review of death

sentences: An empirical study of the Georgia experience. Journal of Criminal Law &

Criminology 74(3): 661–753.

Beckett K, Nyop K and Pfingst L (2006) Race, drugs, and policing: Understanding

disparities in drug delivery arrests. Criminology 44(1): 105–137.

BernardEJ andNyambeM (2012)Criminal prosecutions forHIVnon-disclosure, exposure

and transmission: Overview and updated global ranking. Paper presented at the 19th

International AIDS Conference, Abstract WEAD0201, Washington, DC, 22–27 July.

Bernstein M and Kostelac C (2002) Lavender and blue: Attitudes about homosexuality

and behavior toward lesbians and gay men among police officers. Journal of

Contemporary Criminal Justice 18(3): 302–328.

Burris S, Dalton HK, Miller JL, et al. (1993) AIDS Law Today: A New Guide for the

Public. New Haven, CT: Yale University Press.

Center for HIV Law & Policy (2010) Prosecutions for HIV exposure in the United

States, 2008–2010. Available at: http://www.hivlawandpolicy.org/resources/view/

456.Centers for Disease Control and Prevention (CDC) (2013a) Monitoring selected

national HIV prevention and care objectives by using HIV surveillance data –

United States and 6 U.S. dependent areas – 2011. HIV Surveillance Supplemental

Report, 18(5). Available at: http://www.cdc.gov/hiv/library/reports/surveillance/

Centers for Disease Control and Prevention (CDC) (2013b) Surveillance brief: Terms,

definitions, and calculations used in CDC HIV surveillance. Available at: http://

www.cdc.gov/hiv/statistics/recommendations/terms.html.



http://www.sagepub.com/repository/binaries/pdfs/AnnotationGuidelines.pdf

http://www.sagepub.com/repository/binaries/pdfs/AnnotationGuidelines.pdf

http://www.adobe.com/products/reader.html

http://www.cdc.gov/hiv/statistics/recommendations/terms.html

http://www.cdc.gov/hiv/statistics/recommendations/terms.html


Crawford C, Chiricos T and Kleck G (2006) Race, racial threat, and sentencing of

habitual offenders. Criminology 36(3): 481–512.

Curran DA (1983) Judicial discretion and defendant’s sex. Criminology 12(1): 41–58.

Curry TR (2010) The conditional effects of victim and offender ethnicity and victim

gender on sentences for non-capital cases. Punishment & Society 12(4): 438–462.

Davis A (1998) Prosecution and race: The power privilege of discretion. Fordham Law

Review 67(1): 13–68.

Doerner JK and Demuth S (2010) The independent and joint effects of race/ethnicity,

gender, and age on sentencing outcomes in U.S. federal courts. Justice Quarterly

27(1): 1–27.

Galletly C and Lazzarini Z (2013) Charges for criminal exposure to HIV and aggra-

vated prostitution filed in the Nashville, Tennessee prosecutorial region 2000–2010.

AIDS and Behavior 17(8): 2624–2636.

Galletly C and Pinkerton SD (2006) Conflicting messages: How criminal HIV disclos-

ure laws undermine public health efforts to control the spread of HIV. AIDS and

Behavior 10(5): 451–461.

Galletly C, Glasman LR, Pinkerton SD, et al. (2012a) New Jersey’s HIV exposure law and

the HIV-related attitudes, beliefs, and sexual and seropositive status disclosure behav-

iors of persons living with HIV.American Journal of Public Health 102(11): 2135–2140.

Galletly C, Pinkerton SD and DiFranceisco W (2012b) A quantitative study of

Michigan’s criminal HIV exposure law. AIDS Care 24(2): 174–179.

Hart TC and Rennison C (2003) Reporting Crime to the Police, 1992–2000.

Washington, DC: Bureau of Justice Statistics. Available at: http://www.bjs.gov/

index.cfm?ty¼pbdetail&iid¼1142.

Herek G, Cogan JC and Gillis JR (2002) Victim experiences in hate crimes based on

sexual orientation. Journal of Social Issues 58(2): 319–339.

Holcomb JE, Williams MR and Demuth S (2004) White female victims and death

penalty disparity research. Justice Quarterly 21(4): 877–902.

Holt M, Rawstorne P, Worth H, et al. (2011) Predictors of HIV disclosure among

untested, HIV-negative and HIV-positive Australian men who had anal intercourse

with their most recent casual male sex partner. AIDS and Behavior 15(6): 1128–1139.

Hoppe T (2013) Controlling sex in the name of ‘‘public health’’: Social control and

Michigan HIV law. Social Problems 60(1): 27–49.

Hoppe T (2014) From sickness to badness: The criminalization of HIV in Michigan.

Social Science & Medicine 101: 139–147.

Horvath KJ, Weinmeyer R and Rosser S (2010) Should it be illegal for HIV-positive

persons to have unprotected sex without disclosure? An examination of attitudes

among US men who have sex with men and the impact of state law. AIDS Care

22(10): 1221–1228.

Hundley K, Martin ST and Humburg C (2012) Florida ‘‘Stand Your Ground Law’’

yields some shocking outcomes depending on how law is applied. Tampa Bay Times.

Available at: www.tampabay.com/news/publicsafety/crime/florida-stand-your-

ground-law-yields-some-shocking-outcomes-depending-on/1233133 (accessed 4

June 2013).

Hoppe 91


http://www.bjs.gov/index.cfm?ty=pbdetail&iid=1142




www.tampabay.com/news/publicsafety/crime/florida-stand-your-ground-law-yields-some-shocking-outcomes-depending-on/1233133

www.tampabay.com/news/publicsafety/crime/florida-stand-your-ground-law-yields-some-shocking-outcomes-depending-on/1233133


Kalichman SC and Nachimson D (1999) Self-efficacy and disclosure of HIV-positive

serostatus to sex partners. Health Psychology 18(3): 281–287.

Kautt P (2009) Heuristic influences over offense seriousness calculations: A multilevel

investigation of racial disparity under sentencing guidelines. Punishment & Society

11(2): 191–218.

Koons-Witt BA (2002) The effect of gender on the decision to incarcerate before and

after the introduction of sentencing guidelines. Criminology 40(2): 297–328.

Lahey KE (1995) New line of defense: Criminal HIV transmission laws. Syracuse

Journal of Legislation & Policy 85: 85–95.

Lehman JS, Carr MH, Nichol AJ, et al. (2014) Prevalence and public health implica-

tions of state laws that criminalize potential HIV exposure in the United States.

AIDS and Behavior 18(6): 997–1006.

Lucas SR (2008) Theorizing Discrimination in an Era of Contested Prejudice:

Discrimination in the United States. Philadelphia, PA: Temple University Press.

McArthur JB (2009) As the tide turns: The changing HIV/AIDS epidemic and the

criminalization of HIV exposure. Cornell Law Review 94(3): 707–742.

Mazella R and Feingold A (2006) The effects of physical attractiveness, race, socio-

economic status, and gender of defendants and victims on judgments of mock jurors:

A meta-analysis. Journal of Applied Social Psychology 24(15): 1315–1338.

Miles-Johnson T (2013) LGBTI variations in crime reporting: How sexual identity

influences decisions to call the cops. SAGE Open 3(2): 1–15.

Murakawa N and Beckett K (2010) The penology of racial innocence: The erasure of

racism in the study and practice of punishment. Law & Society Review 44(3–4):

695–730.

Petrocelli M, Piquero AR and Smith MR (2004) Conflict theory and racial profiling:

An empirical analysis of police traffic stop data. Journal of Criminal Justice 31(1):

1–11.

Prestage G, Van de Ven P, Grulich A, et al. (2001) Gay men’s casual sex encounters:

Discussing HIV and using condoms. AIDS Care 13(3): 277–284.

Provine DM (2011) Race and inequality in the war on drugs. Annual Review of Law and

Social Science 7: 41–60.

Radelet ML and Pierce GL (1985) Race and prosecutorial discretion in homicide cases.

Law & Society Review 19(4): 587–622.

Rios V (2011) Punished: Policing the Lives of Young Black and Latino Men. New York:

New York University Press.

Schuller RA and Hasting PA (2002) Complainant sexual history evidence: Its impact on

mock jurors’ decisions. Psychology of Women Quarterly 26(3): 252–261.

Shevory T (2004) Notorious HIV: The Media Spectacle of Nushawn Williams.

Minneapolis, MN: University of Minnesota Press.

Steffernsmeier D, Kramer J and Ulmer J (1995) Age differences in sentencing. Justice

Quarterly 12(3): 583–602.

Steffernsmeier D, Ulmer J and Kramer J (1998) The interaction of race, gender, and age

in criminal sentencing: The punishment cost of being young, black, and male.

Criminology 36(4): 763–798.




Strub S (2010) Criminalization 101. POZ Blogs. Available at: http://blogs.poz.com/

sean/archives/2010/11/criminalization_101.html.

Wacquant L (2010) Class, race & hyperincarceration in revanchist America. Daedalus

193(3): 74–90.

Wolff KB and Cokely CL (2007) ‘‘To protect and to serve?’’: An exploration of police

conduct in relation to the gay, lesbian, bisexual, and transgender community.

Sexuality and Culture 11(2): 1–23.

Hoppe 93


http://blogs.poz.com/sean/archives/2010/11/criminalization_101.html

http://blogs.poz.com/sean/archives/2010/11/criminalization_101.html


Documents

Disparate risks of conviction under Michigan’s felony HIV disclosure law: An observational analysis of convictions and HIV diagnoses, 1992–2010