Distribution and Correlates of Self-Reported Crimes of Trust

Abstract

This study examines the distribution and correlates of a special class of property crimes, crimes of trust, using longitudinal and cross sectional self-report data from a national sample. We begin by defining crimes of trust and consider their conceptual relationship to “conventional” property crimes, which we here characterize as crimes of stealth, and to white collar crimes, which are defined in terms of the social status of the perpetrators. Crimes of trust are here defined as property crimes that typically involve deliberate contact with the victim or, where there is more than one victim, with at least one or more victims, in which there is typically more of a focus on concealing the fact that a crime has been committed than on concealing the identity of the perpetrator (as is the case in crimes of stealth), without regard to the socioeconomic status of the perpetrator (thus including but not limited to white collar crimes). The focus here is on crimes of trust committed by individuals (as opposed to corporate crime). We first examine their distribution by sociodemographic characteristics, then examine the correlation of crimes of trust with other types of illegal behavior, using data from the National Youth Survey Family Study, including (1) longitudinal self-report data from a nationally representative panel of individuals who were 11–18 years old in 1976–77 and who were followed through early middle age (ages 36–44) in 2002–2003, plus (2) cross-sectional data on these individuals plus their parents, spouses, and children age 11 and older in 2002–2003 (total age range 11–88). The results suggest that crimes of trust have a different age-crime curve from conventional crimes, and that they are not as strongly correlated with problem substance use, gender, and other socioeconomic indicators as conventional crimes.

Distribution and Correlates of Self-Reported Crimes of Trust

There is a recognition in criminology that it is useful to distinguish among different types of crime. Causes and consequences may be different for violent crimes, property crimes, substance use, and for public order offenses not related to substance use. Crimes may also be distinguished by who commits them: individuals or corporations, and at the individual level, individuals of higher or lower socioeconomic status. The research presented here is designed to examine the extent, nature, and correlates of three types of crime, occupational, avocational, and contrepreneurial crimes. These offenses are, in a sense, doubly marginalized within the field of criminology, central to the study neither of “conventional” crimes involving individual acts of violence or theft, nor of “mainstream” white collar crime, with its typical focus on corporate actors and, frequently, on offenders of higher socioeconomic status. Our approach blends the more inclusive scope of occupational, avocational, and contrepreneurial crimes (some of them marginal to the study of white collar crime), suggested by Friedrichs (2002), with the suggestion by Shapiro (1990) that the focus be on the nature of the crimes and not the criminals. In particular, Shapiro suggested the study of “violations of trust” encompassing civil as well as criminal offenses. Here we focus specifically on the criminal rather than the civil element of Shapiro’s concept, crimes of trust, that are most clearly the appropriate concern of criminology and the criminal (as opposed to civil) justice system. Crimes of trust include, but are not limited to, crimes perpetrated by individuals of higher socioeconomic status (but the focus here is on the characteristics of the crime, not of the criminal); and they include, but are not limited to, individual level offenses commonly included in the study of white collar crime.

There are two main reasons for paying special attention to crimes of trust in general and to occupational, avocational, and contrepreneurial crimes more specifically. First, by further developing the concept of crimes of trust, it may be possible to provide a classification that preserves the original intent of Sutherland’s (1940) focus on crimes characteristically committed by individuals of higher social status, as well as the distinction between corporate and individual crime, by recognizing typical white collar crimes as crimes of trust, but also recognizing that not all crimes of trust are really white collar crimes in the sense originally intended by Sutherland. Second, as noted above, these (occupational, avocational, and contrepreneurial) crimes have been marginal to the focus of most criminologists, and have thus been under-studied. In particular, to the best of our knowledge, this study represents the first examination of these offenses using nationally representative data spanning much of the life course, from adolescence to middle adulthood for a focal group of respondents and, in a more exploratory analysis, spanning the life course from adolescence to advanced old age. Attention to the life course has become increasingly prevalent in criminology (e.g., Benson 2002; Moffitt 1997; Piquero and Mazerolle 2001; Sampson and Laub 1993). In this latter context, changes in prevalence of white collar crime over the life course may be of particular interest. As noted by Moffitt (1997), among others, the typical pattern for most criminal behavior is a gradual increase from childhood to a peak in adolescence, followed by a decline thereafter. For occupational, avocational, and contrepreneurial crimes, however, opportunities are likely to be more limited in adolescence and to increase with age, and thus for these offenses, we would expect a different age-crime curve than might be the case for more conventional violent, property, substance, or public order crimes (Benson 2002). The present study advances the study of crime and paves the way for future research by clarifying the distinction between crimes of trust and other types of crime; examining the relationship between crimes of trust and other types of crime; and examining age and other sociodemographic correlates of these offenses in a nationally representative longitudinal sample.

Conventional Classifications of Crime

A conventional division of types of crime in many criminology textbooks and much criminological research distinguishes among (1) crimes of assault, or violent crimes, (2) crimes of theft and vandalism, or property crimes, and (3) public order offenses, or victimless crimes. In this last category, distinctions may be made (particularly in the literature on juvenile delinquency) between (3a) vice crimes and other victimless crimes that are illegal for adults, including public disorder, public drunkenness, illicit drug use, gambling, and prostitution, and (3b) status offenses, which involve behaviors that are illegal for juveniles but legal for adults, including underage alcohol and tobacco use, truancy, and runaway. Not all crimes fit uniquely into a single category. Robbery, in particular, although it is most often classified as a violent crime, involves both violence and theft, and thus straddles the first two categories. The term “victimless” is often disputed when applied to some public order offenses such as prostitution or illicit drug use. Another distinction found particularly in the victimization literature, and applying to the second category of offenses (property), is the distinction between larceny with contact (for example, pocket picking or purse snatching) and larceny without contact (for example, theft from a public place). Also, most often within the second category, a distinction is sometimes made between “crimes in the streets” and “crimes in the suites” or between conventional (property) crimes and individual level or corporate white collar (property) crimes. Although some white collar crimes may result in physical harm (illness, injury, or death as a result of environmental pollution, falsification of test results, or faulty construction or manufacturing practices), there is substantial agreement concerning the idea that the concept refers to crimes committed in a “legitimate occupational context, … motivated by the objective of economic gain or occupational success, … not characterized by direct, intentional violence, … committed by offenders who do not have a criminal self-image, [and refers to crimes that have] inspired a more limited criminal justice response than is true of conventional, street crime” (Friedrichs 2007, p. 4–5).

White Collar Crime

Since its earliest use by Edwin H. Sutherland, the definition of the term “white collar” crime as “a crime committed by a person of respectability and high social status in the course of his occupation” (Sutherland 1940; Sutherland 1949: 9) has been the subject of much debate—for an exceptional in-depth discussion defining white collar crime see Geis (2007). Sutherland’s definition brought attention to the fact that great social harm results from the behaviors of individuals outside the circle of poverty and outside of disadvantaged neighborhoods. However, it may be argued that his definition is limited, by modern standards, in the sense that it did not cover identical acts by those not of high social status or behaviors not committed through one’s occupation (Shapiro 1980, 1990). Opportunity structures for white collar types of offending were different during Sutherland’s day, and today people of all ranks commonly have opportunities to commit some of the same financially based violations of trust that Sutherland explored in the 1940s. This is in part because of the evolution of the marketplace in the information age (Weisburd, Waring, and Chayet 2001: 11), part of a longer term trend in which the structure of occupations in the United States has shifted, with declines in primary (extractive industries such as mining, farming and fishing) and secondary (manufacturing industries involving primarily blue collar jobs) to tertiary (service industries including professional and technical as well as sales and clerical) positions (Caplow, Hicks, and Wattenberg 2001), the last group (tertiary) more characteristically being entrusted with access to cash, personal information, and other opportunities consistent with white collar crime, even in lower status positions. Still, Sutherland’s definition has received much attention in recent years (e.g., Geis 1992; Coleman 1992; Wheeler 1983; Weisburd et al. 2001), and his focus on crimes of the elite continues to be of interest.

The focus of research on white collar crime ranges from crimes of the power elite (e.g. Geis 1992; Simon 2006; Sutherland 1949) to corporate or organizational crimes (e.g. Bilimoria 1995; Braithwaite 1984, 1985; Piquero et al. 2005; Pontell and Calavita 1992; Reichman 1992; Simpson 1992; Vaughan 1982) to white collar criminality at the individual level (e.g. Benson and Moore 1992; Benson and Kerley 2000; Collins and Schmidt 1993; Croall 1989; Daly 1990; Holtfreter 2005; Lemert 1953, 1958; Schoepfer and Piquero 2005; Wheeler et al. 1982; Weisburd et al. 2001; Weisburd, Wheeler, Waring, and Bode 1991; Williot et al. 2001). The common thread appears to be agreement that the study of white collar crime constitutes economically based, non-conventional criminal behavior; some violation of trust; some form of personal gain which is usually either directly or indirectly financial in nature (direct acquisition of money or property, or possibly occupational advancement that results in higher income or other financial rewards); and the absence of direct or overt violence (Friedrichs 2007; Shapiro 1990).

Most white collar crime criminologists would also agree that it is possible to distinguish between different subtypes of white collar crime. At a minimum, it is customary to distinguish between occupational crime and organizational or corporate crime. Occupational crime refers to “illegal or harmful financially driven activity committed within the context of a legitimate, respectable occupation,” while corporate crime deals with “illegal and harmful acts committed by officers and employees of corporations to promote corporate interests” (Friedrichs 2007:7; see also Coleman 2002). Finally, there are some crimes that have the appearance of white collar crime, or are related to them, but are not at the core of what most researchers would consider white collar crime. Friedrichs (2002:7) refers to them as “contrepreneurial crimes” (“swindles, scams, and frauds that assume the guise of legitimate businesses”) and “avocational crimes” (“illegal, but nonconventional criminal acts committed by white collar workers outside a specifically organizational or occupational context, including income tax evasion, insurance fraud, loan/credit fraud …”) as originally defined by Geis (1974).

Crimes of Trust

We suggest an approach to categorizing crime that acknowledges the original intent of Sutherland, to highlight the crimes committed by individuals of higher social status; but which at the same time incorporates insights from Friedrichs (2002) and Shapiro (1990) and recognizes that individual level white collar crimes have much in common with, and may be regarded as a subset of, a broader categorization of crimes, which we call crimes of trust. Crimes of trust may be distinguished from conventional or “street” crimes of theft and vandalism in the following way. Conventional crimes of theft and vandalism involve stealth, the avoidance of detection by concealing one’s presence and avoiding direct or detectable contact with one’s victim: stealing a purse or wallet when someone’s back is turned, burglarizing a house when the residents are away, spraying graffiti on a fence when no one is looking. To the extent that contact is involved (purse snatching, which may constitute strong-arm robbery if the crime is detected and resisted as it is occurring; armed robbery), a crime of theft may also be considered a crime of violence (use of force or threat of force), a type of crime in which contact with the victim is more common. If there is contact between victim and offender in conventional crimes of theft or vandalism, it is likely to be incidental, and the perpetrator is unlikely to prolong it any longer than necessary. In crimes of trust, by contrast, direct contact with at least some of the victims of the crime is deliberately sought out, and may be ongoing rather than incidental.

Whether committing occupational crimes such as embezzlement or workplace theft; avocational crime such as check, credit card, tax, or insurance fraud; or contrepreneurial crimes such as selling worthless goods; the attempt is made not to avoid detection of the perpetrator, but the detection of the offense, at the time the offense is being committed. The perpetrator deliberately (a) continues to show up at work while embezzling, (b) files the false tax or insurance forms or signs and presents the forged check or credit card slip, or (c) interacts with the purchaser; and attempts to gain and maintain the trust of the victim (the company or more specifically the supervisor at work; the government agency, insurance agency, or merchant; the buyer). Ideally, the improper use of funds, the fact that the forms have been falsified, the fact that the goods are worthless, is never detected in a crime of trust, in contrast to conventional thefts and vandalism (crimes of stealth), where the loss or damage of property is readily detectable almost immediately after it occurs, and it is the identity of the perpetrator, not the occurrence of the crime, that the perpetrator is most concerned with concealing.

A more detailed contrast between crimes of trust, as the term is being used here, and other types of crime may be helpful. Compared to crimes of violence: although crimes of trust may ultimately result indirectly in physical harm (injury from faulty merchandise, suicide resulting from financial loss), crimes of trust as such do not involve the deliberate threat or infliction of violence. Compared to predatory property crimes: In crimes of stealth like automobile theft, burglary, and larceny without contact, the offender typically avoids making contact with the victim; and even in larceny with contact (for example, pickpocket; as opposed to robbery, a crime of violence as well as theft), perpetrators will attempt to avoid making their presence known; but in crimes of trust, the perpetrator may interact with the offender on a regular and frequent basis (employee theft, embezzlement; one might also include thefts targeting family members here), or deliberately seek interaction with at least one of the victims with no attempt to conceal the presence of the perpetrator (selling something worthless, insurance fraud), although there may be an attempt to conceal the true identity of the perpetrator (forgery, credit card fraud). Compared to public order crimes: Some public order crimes are not committed for financial gain (illicit drug use, disorderly conduct), but crimes of trust are motivated by profit; and other public order crimes involving provision of illegal services (illicit drug sales, prostitution) do not, as such, involve deceptive claims (it is only when the promised services or goods are not delivered, for example, stiffing a john or selling oregano as marijuana, that the nature of the crime shifts from being a public order crime to a crime of trust). Compared to white collar crimes: staying with Sutherland’s classic definition, white collar crimes are necessarily committed by individuals of higher social status; crimes of trust may be committed by individuals of any social status, and hence may or may not be white collar crimes; and in general, we would not expect sales frauds, credit card frauds, or employee thefts, for example, to be committed exclusively or even primarily by individuals of higher social status. Conceptually, then, crimes of trust represent a constellation of crimes that share important characteristics with one another and are not adequately covered in other existing categories of criminal behavior.

Research on White Collar Crimes and Crimes of Trust

To the best of our knowledge, the present research is the first to take Shapiro’s (1990) concept of violations of trust, here more specifically focusing on criminal violations of trust, or crimes of trust, and to examine this category of crimes both in its own right and in contrast to other types of crime. Typically, crimes of trust are either lumped together with other crimes of theft, or with white collar crime, or sometimes split between the two (e.g., Siegel 2003, who argues for a definition of white collar crime that includes blue collar criminals). To the extent that there is research on crimes of trust, however, they are most often considered as more or less marginal forms of white collar crime (and the actual white-collar status of the offenders, consistent with the approach in Siegel 2003, is ignored, contrary to Sutherland’s original conceptualization). In reviewing research on crimes of trust, therefore, some focus on white collar crime research is unavoidable. Individual level white collar crimes generally are crimes of trust, and the research on crimes of trust (including crimes of trust perpetrated by blue-collar offenders) is generally presented, albeit marginally, in this context.

Research directed at explaining economic crimes ranging from the white collar offenses of middle-class individuals to the most highly publicized corporate crimes of the power elite has been slow in development compared to research on conventional crimes (Sutherland 1940; Hirschi and Gottfredson 1987; Friedrichs 2007, p. 31). One underlying reason for this slow development has been a lack of representative data. This problem has limited the external validity of many findings and has constrained modern statistical analyses, leaving many questions about white collar crime unanswered (Benson 2002, p. 134; Weisburd et al. 2001, p. 135). Although some white collar crime scholars may disagree on the definition of white collar crime, it is clear that “non-conventional” property crimes, which are not included in more restrictive definitions of white collar crime but are included as crimes of trust, are quite prevalent and result in much social harm, making them worthy of focused attention (Benson 2002; Friedrichs 2007; Shover and Hochstetler 2006; Sutherland 1940:8; Weisburd et al. 1991).

Much of the offense-based individual level research on crimes of trust or white collar crime to date has relied on either case studies or official arrest or court records. Case studies show the diversity of behaviors commonly associated with white collar crime or crimes of trust, but generalizability is severely limited (Shover and Hochstetler 2006, p. 13; Shapiro 1983). Official data on the other hand have provided for greater sophistication in analysis and have had a substantial impact on what we know about the criminal justice response to crimes of trust and about the individuals who are convicted of a limited number of federally defined white collar offenses (see for example Benson and Kerley 2000; Benson and Moore 1992; Collins and Schmidt 1993; Daly 1989; Weisburd et al. 1991; Weisburd et al. 2001). Such findings, while important, must be interpreted with caution due to the nature of the data and the potential that many offenders may obtain effective legal counsel, thus minimizing sanctions and reducing the chance of conviction (Weisburd et al 1991; Benson and Moore 1992; see also Weisburd et al. 2001). More problematic is that such data only account for federal felony convictions in a limited number of jurisdictions dating back several decades (Weisburd et al. 1991, pp 17–20; Benson and Moore 1992; for an exception see Weisburd et al. 2001, pp 15–17). Civil cases, administratively handled incidents, and those cases tried in state courts are not included in such analyses.

Two major data sources have set a benchmark for offense-based white collar crime research. These include data collected for the Yale Studies on White Collar Crimes (referred to as the Wheeler data; see Wheeler et al. 1982) and data collected by Forst and Rhodes (N.D.). The Wheeler data have been used extensively and have been the basis for a number of books and research articles (e.g., Daly 1989; Langton and Piquero 2007; Wheeler et al. 1982; Weisburd et al. 1991; Weisburd et al. 2001). The Forst and Rhodes data have been researched exclusively by Michael Benson and his associates (1992, 2000). Both come from federal court records that originated in the 1970s and many analyses have been based on court-provided pre-sentence investigation (PSI) reports. These data have allowed for both quantitative and qualitative exploration on white collar offenders and provide the bulk of information on what is presently known about the correlates of white collar offending and individual level crimes of trust. In fact, Weisburd et al. (2001) greatly expanded upon the Wheeler data by gathering FBI rap sheets on offenders for a twelve year follow up period on the original sample. The Wheeler data covers antitrust violations, bribery, securities fraud, tax fraud, false claims, credit fraud, mail and wire fraud, and postal embezzlement. Our focus here is on individual level crimes of trust, overlapping with these data in the areas of embezzlement (not, for us, limited to postal or bank embezzlement), tax fraud and tax evasion, and false claims; but consistent with our concern for the broader range of trust, not limited to these offenses.

The most significant limitation to these data sources is that, as noted above, they cannot account for the “dark figure” of unreported white collar crime or crimes of trust. Official statistics fail to detect many (for most offenses, the majority of) offenses and many of the perpetrators who commit them (Elliott 1995). Although official data and victimization surveys have shown a trend toward convergence in their estimates of illegal behavior over the past decade, victimization surveys still indicate much higher rates of violent and property crime than official statistics, and official statistics and victimization surveys often disagree about differences and trends in crime, whether particular types of crime are higher in one location or another, or whether they are increasing or decreasing over time (Lynch and Addington 2007; Menard and Covey 1988). Similarly, self-report data, which are generally consistent with victimization data, consistently indicate much higher rates of offending than official statistics, particularly for crimes other than the Federal Bureau of Investigation Part I or “Index” offenses (Menard 1987; Wells and Rankin 1995). This is especially a concern for crimes of trust, in which concealment of the fact that a crime has been committed is one of the characteristics of this type of offense. As noted by Benson and Moore (1992:258), such crimes may be more difficult to detect compared to conventional crimes, and a single offense may be “geographically and temporally scattered.” The ability to provide information on unreported crimes is a well established strength of self-report data (see for example the review in Elliott, Huizinga, and Menard 1989:4–9) and a documented shortcoming of white collar crime research to date (Benson 2002:142; Weisburd et al. 2001:19:135). The data to be used in the present research help address the need (Benson 2002:142; Benson and Moore 1992:258; Friedrichs 2007:37; Weisburd et al. 2001:135) to examine white collar crime or crimes of trust using nationally representative self-report data.

CORRELATES OF CRIMES OF TRUST: SELECTED OFFENSES

A substantial body of literature surrounds the particular offenses that are explored here. Here we limit our review of previous findings to studies involving the crimes of trust actually covered in the present study: occupational (embezzlement, employee theft), avocational (credit card fraud, personal check fraud, forgery, filing fraudulent insurance claims, tax evasion) and contrepreneurial (selling worthless goods—cheating someone) crimes of trust.

Federally Convicted White Collar Offenders

The correlates of white collar offending for offenders convicted at the federal level are fairly consistent across samples. Compared to conventional criminals, offenders convicted of white collar offenses are more typically male; older, entering the criminal justice system later in life than conventional criminals; white; less likely to have a prior history of arrest; more socially stable (home ownership, steady employment, steadily married); and more educated (Benson and Kerley 2000; Weisburd et al. 1991). They are less likely to have a history of trouble in school (Benson and Moore 1992), less likely to come from a troubled childhood or from poverty (Benson and Kerley 2000), less likely to have drug and alcohol problems (Benson and Moore 1992; Weisburd et al. 2001), more likely to participate in pro-social adult life activities (e.g. to be involved in community groups and attend church regularly) (Benson and Kerley 2000), and they commonly lead stable lives comparable to the population at large (Weisburd et al. 2001, p. 57). In summary, research on federally convicted white collar offenders suggests that on average they may be distinguishable from conventional offenders in many respects, but it remains to be seen whether these findings regarding convicted white collar offenders also hold for individuals who are self-reported crime of trust offenders (hereafter trust crime offenders). Are the same patterns found (as we might expect they would be if white collar offenses are a subset of crimes of trust, and if official data are consistent with self-reports) using offenses that (a) do not include the same range of white collar offenses as are present in previous studies and (b) include additional crimes of trust that would be excluded under more restrictive definitions of white collar crime?

Research Questions

The present study examines the prevalence and correlates of crimes of trust, with a comparison to conventional crimes. (1) First, we examine the annual prevalence of crimes of trust over the adult (and a little bit of the adolescent) life course, using both cross-sectional data that allow us to compare age groups ranging from age 18 to 88, and longitudinal data that allow us to trace the prevalence of crimes of trust for the same respondents from age 14 to 44. (2) Second, we examine the cumulative prevalence of crimes of trust over varying spans of the life course, including both prospective and long-term retrospective longitudinal data. (3) Third and finally for the present research, we compare the sociodemographic correlates of crimes of trust and conventional crimes, to see to what extent the patterns suggested by studies of white collar crime based on official data are informative for the crimes of trust studied here using self-report data. Are the correlates of self-reported crimes of trust consistent with the correlates found in previous studies of officially recorded white collar crime (Benson and Kerley 2000; Benson and Moore 1991; Croall 1989; Daly 1989; Weisburd et al. 1991; Weisburd et al. 2001), despite the differences in both method (official data vs. self-report) and measurement (the specific offenses examined)? The purpose here is not (yet) to develop a causal model of crimes of trust, but to lay a foundation that will better enable us to develop causal models in future research. Before we can explain the social facts (Durkheim 1938 [1895]) regarding crimes of trust, we must first establish what those social facts are, a task not yet adequately addressed in previous research.

DATA AND METHODS

Sample

Data for this study are taken from the National Youth Survey Family Study (NYSFS; formerly the National Youth Survey or NYS). The NYSFS study design involves a multiple cohort sequential design (Baltes, Cornelius, and Nesselroade 1979) with twelve waves of data over a 27-year period. The survey sample is based on a probability sample of households in the continental United States selected using a multistage, cluster sampling design. As described in Menard and Mihalic (2001) the sample was drawn in late 1976 and contained 2,360 eligible youth respondents aged 11–17 at the time of the initial interview. Of these, 1,725 (73%) agreed to participate in the study, signed informed consents and completed interviews in the initial survey. Parents of these youth respondents were also interviewed in this first wave. With respect to race, class, age, and residence (urban-suburban-rural), NYSFS respondents appear to be representative of the total 11 through 17 year old youth population in the United States as established by the U.S. Census Bureau for 1976, and an age, sex and race comparison between nonparticipating eligible youth and participating youth indicates that the loss rate from any particular age, sex, or racial group appears to be proportional to that group's representation in the population.

In the first four (annual) follow-up surveys, overall completion rates over waves 2–4 were above 94 percent; and 87% for wave 5. Completion rates were 87% for wave 6, 80% for wave 7; 83% for wave 8; and 78% for Wave 9. Data for waves 10, 11, and 12 were collected in 2002–2004. Additional deaths among the respondents were identified during the wave 10–12 data collection, reducing the total eligible sample to 1,677. Of these 1,677, we were able to interview 1,266 or 75% in wave 10 and 1,173 or 70% in wave 11.¹ In waves 11 and 12, we attempted to identify eligible (living and in sufficient physical and mental health to complete an interview) spouses and partners (either married or living together), parents, and original respondents’ children age 11 and older in 2003. Of the known eligible parents, spouses, and children, in wave 11 we interviewed 881 or 71% of the parents, 679 or 71% of current spouses or partners of the original respondents, 802 or 77% of their adolescent (age 11–17) children, and 464 or 66% of their 707 adult (age 18 and over) children. In wave 12, only the children of the original respondents were interviewed, and of the known eligible respondents, interviews were completed with 815 or 78% of the adolescent children and 491 or 70% of the adult children.

Two reservations need to be noted about the use of the parents and spouses of the original respondents in this analysis. Strictly speaking, the original respondents represent an age-specific general population sample, but the spouses and parents do not. The spouses and partners of the original respondents do not include individuals who are single, never married, and never in an intimate relationship. The parent sample likewise excludes individuals who have never been married and never in an intimate relationship, and also excludes individuals who have never had children. To the extent that involvement in crimes of trust and conventional crimes differ based on marital and parental status, this may limit the generalizability of our findings to individuals who are married or cohabiting and who have at least one child; but the disadvantage of this question about generalizability seems to us to be outweighed by the advantages of being able to examine crimes of trust across nearly the full life course, for individuals with ages ranging from 11 to 88; and using the original respondents as the focal sample, we may be able to get some indication of the extent to which the exclusion of individuals who are married or cohabiting and who have young or adult children might affect our conclusions.

Measurement of Variables

Respondents were asked to report on their involvement in a variety of delinquent and criminal behaviors during all waves of NYSFS data collection. For each wave of the study, subjects were thus asked to recall events from the past year and report a number of times they engaged in that behavior. In the first five waves of the study, this resulted in data from five consecutive years. Waves six to nine were collected at three-year intervals, and in each of those waves, three questions were asked. First, respondents were asked to report a number by recalling one year past as in all prior waves. Subjects were also asked to report on two years prior and three years prior, to fill in the years between data collections. For these two and three year recall items, subjects were not asked to report a specific number, but instead were asked to report the frequency of their participation using a 4 point scale: (1) never, (2) 1–2 times, (3) 3–11 times, and (4) 12 or more times. In Waves 10 and 11 for original respondents, and for each wave of parent, spouse, and child respondents, subjects were asked to provide the number of times in the past year they engaged in each of the presented behaviors as well as whether they had ever engaged in a behavior in their lifetime. For all waves and intervening years, prevalence of a behavior is indicated by any non-zero response to an item. It is evident that the longer-term retrospective data underestimate the prevalence of illegal behavior for the intervening years (Menard and Elliott 1990), but they are useful here to help reduce the error of underestimating cumulative prevalence of offending over the life course.

The NYSFS contains data on each of the crimes of trust described earlier: occupational crime (embezzlement and workplace theft), avocational crime (check, credit card, tax, and insurance fraud), and contrepreneurial crime (selling worthless goods). While not every variable was measured at every wave and for each respondent group, interview questions were presented in a standard format each time they were asked. This section will outline how each dependent variable was measured and will note which groups, original respondents (OR), parents (PA), spouse/partner (SP), adult offspring (AO), or youth offspring (YO), provide information for each type of crime at different ages and waves of the study. As noted above, depending on which respondent group an individual belonged to and which wave they were reporting on, respondents were asked to report either the frequency of offenses committed for the year prior to the interview or whether they had ever in their life committed a particular act.²

Three items, check fraud, credit card fraud, and sales fraud, were asked in waves 4–11 for the OR, wave 11 for the PA and SP, waves 11–12 for the AO, and wave 12 for the YO. For the check fraud item, respondents were asked to report whether they had ever “used checks illegally or used phony money to pay for something.” The measure of credit card fraud was whether they had “used or tried to use credit cards without the owner’s permission.” Sales fraud was measured by asking respondents to report whether they had “tried to cheat someone by selling them something that was worthless or not what you said it was.” Employee theft was reported by original respondents (waves 4–11) and their adult offspring (waves 11–12) only. Respondents were asked to indicate whether they had “stolen or tried to steal something from work.”

In waves 6 through 11 for the OR, 11 for SP and PA, 11–12 for AO, and 12 for YO, respondents indicated whether they had “embezzled money, that is, used money or funds entrusted to your care for some purpose other than intended.” In waves 9–11 for the OR, 11 for SP and PA, 11–12 for AO, and 12 for YO, questions were asked about forgery, false insurance claims, and income tax evasion. Forgery was measured by asking respondents if they had “forged or copied someone else’s signature on a check or legal document without their permission.” For insurance fraud, respondents were asked to report whether they had “made fraudulent insurance claims, that is, falsified or inflated medical bills and property or automobile repair or replacement costs.” Lastly, respondents were asked to report whether they had “intentionally underreported money earned or received, overestimated expenses or losses, or otherwise cheated on your state or federal income taxes.”

Data on gender and ethnicity are collected at each wave. For the original respondents, data were collected and verified on birth dates, and these data are standardized across waves (i.e., where there are variations in reported age at the time of the interview, the age based on the birth date is used). Data were collected on the highest grade completed in each year, and these data are aggregated into educational stages (elementary school; some high school; high school graduation; some college; college graduation; and postgraduate education) based on past research indicating that it is the successive levels, rather than specific years of education, that have apparent interval scale properties in terms of their relationships with other variables (Blau and Duncan 1967). This corresponds to the practical reality that the difference between two and three years of high school matters much less than the difference between three years of high school and high school graduation, and similarly for college. Occupational status for the original respondents during adolescence is operationalized as the occupational status of the parents of the original respondents; occupational status for the youth offspring of the original respondents is indicated by the occupational status of the parents of the youth offspring (i.e., the occupational status of the original respondents at the time the data were collected on the youth offspring); and occupational status of adults is operationalized by their own occupational status at the time the data were collected. Occupational status is measured using the Duncan Socioeconomic Index (SEI) scale of occupational prestige (Duncan 1961; Blau and Duncan 1967), a measure that allows us to maintain longitudinal consistency with the earlier waves of data. Data on whether the respondents (or, in the case of adolescent respondents, their parents) were receiving welfare payments, and on their marital status, were collected at each wave, and these data are also included in the present analysis.

To further examine the extent to which trust crime offenders are similar or different from other types of offenders, we also include measures of other types of offending as correlates of the sociodemographic and theoretical variables included here. For illicit substance use offenses, we include marijuana use and the polydrug use scale used in previous research involving the NYSFS. The polydrug use scale consists of use of any one or more of five substances: amphetamines, barbiturates, cocaine, hallucinogens, or heroin, and previous research using these data has indicated that polydrug users have almost universally used alcohol and marijuana as well (Elliott et al. 1989). For each substance, respondents were asked how many times in the past year they had used the substance. For nondrug offenses, we include the felony assault and felony theft scales from Elliott et al. (1989). Felony assault includes aggravated assault, sexual assault, and gang fights (or, in wave 9, gang fights are replaced by battery). Felony theft includes motor vehicle theft, burglary/breaking and entering, theft of something worth more than $50, and buying stolen goods. Respondents were asked how many times in the past year they had committed each offense. These offenses have been shown to be characteristically nontrivial offenses that involve actions likely to result in an official response if detected, in contrast to, for example, the minor theft and minor assault scales, which have been shown to include high proportions of offenses that might be considered too trivial to result in arrest (Huizinga and Elliott 1986). When comparing crimes of trust to these specific categories of conventional crime, we use a parallel division of crimes of trust into occupational, avocational, and contrepreneurial crime. When comparing crimes of trust to conventional crime more generally, we use a single index combining all seven types of conventional crime, and a parallel index combining all three types of crimes of trust.

Finally, we also include problem alcohol use and problem drug use as possible correlates of crimes of trust. Problem alcohol and drug use each consist of a six item scale reflecting negative social consequences of substance use by asking how many times in the past year (with responses ranging from never to more than six times) the respondent had experienced problems (trouble with friends, spouse/ boyfriend/ girlfriend, physical fights, physical health problems, trouble with police) as a result of their substance use. As noted earlier, official data on white collar offending suggests that there should be less of a correlation with problem substance use for crimes of trust than for conventional crimes; but illicit substance use that results in problems with friends, relatives, or the law may nevertheless increase one’s predisposition to commit crime in general, including crimes of trust as well as conventional crimes.³

In analyzing the relationship between crimes of trust and age for the five subsamples (OR, SP, PA, AO, YO), we use the Wave 11 NYSFS data collected in 2003, the only year for which we have data on all five subsamples. The data for Wave 11 are divided into eight age categories (seven of which are used below), and some explanation of those age categories is in order here. An adolescent age group, ages 11–17 (Youth Offender subsample), was deliberately chosen to parallel the ages of the original NYS respondents (OR) in the first wave of the survey, but is excluded from Table 1 because the YO subsample was not asked questions about crimes of trust in Wave 11. The transitional age group (ages 18–26) was likewise selected to parallel wave 6 of the NYS (ages 18–24), to correspond to a meaningful segment of the life course (the transition from adolescence to adulthood; this stage includes completion of high school, college and possibly graduate school for some respondents and serious entry into the labor market for others), and also to make a split that would be convenient for the definition of subsequent age groups. The young adult age group (ages 27–35) parallels the ages at wave 9 of the NYS (ages 27–33; we have no data on OR for ages 34 and 35), and represents the stage at which education is typically completed for all but a very few respondents, and full scale entry into the labor force has occurred for all who choose to participate in the labor market. In the present analysis, transitional and young adult respondents are drawn predominantly from the AO subsample. The number in the young adult category is relatively low, but should be sufficient (with due caution) for analysis. The early middle age group (age 36–45) is the first of four ten-year age groups; the ten-year span was used to provide sufficient cases for analysis in each age group. This group consists primarily of OR, with some SP as well. In later middle age (ages 46–55), the sample consists primarily of SP, with a few PA and AO (blended families) as well. The groups labeled pre-retirement (ages 56–65), early elderly (66–75), and later elderly (76–88; there were just a handful of respondents over age 85) consist almost entirely of the PA, with some older SP in the pre-retirement group. The numbers of cases for the later middle age and the later elderly groups are, as for young adults, relatively small, but should be adequate to give some idea of how the relationship between crimes of trust and age differs (cross-sectionally; this does not really represent change, for which longitudinal data on the same respondents at different times would be required) across the life course from early adulthood to old age.

Table 1.

Prevalence of Crimes of Trust by Age Group, Weighted Moving Averages (and Raw Prevalence) 2002 (NYSFS Wave 11)

Prevalence: Moving Average (Raw Prevalence)	Transitionals Age 18–26	Young adults Age 27–35	Early middle age Age 36–45	Late Middle age Age 46–55	Pre- retirement Age 56–65	Early elderly Age 66–75	Late elderly Age 76–88
	n=440	n=125	n=1,755	n=151	n=389	n=379	n=94
Embezzlement	.004 (.005)	.004 (.000)	.004 (.004)	.004 (.007)	.000 (.000)	.000 (.000)	.000 (.000)
Forgery	.021 (.011)	.020 (.056)	.025 (.020)	.021 (.060)	.017 (.010)	.008 (.008)	.006 (.000)
Fraud: check	.022 (.021)	.019 (.024)	.021 (.018)	.019 (.060)	.017 (.010)	.009 (.008)	.009 (.009)
Fraud: credit card	.013 (.005)	.054 (.040)	.008 (.003)	.005 (.040)	.008 (.003)	.000 (.000)	.000 (.000)
Fraud: income tax	.030 (.023)	.066 (.056)	.081 (.077)	.082 (.152)	.091 (.075)	.072 (.082)	.070 (.022)
Fraud: insurance	.003 (.002)	.006 (.008)	.008 (.007)	.007 (.026)	.010 (.000)	.007 (.013)	.013 (.011)
Fraud: sales (selling something worthless or worth less than claimed)	.025 (.027)	.014 (.016)	.012 (.011)	.010 (.020)	.011 (.003)	.007 (.011)	.010 (.011)
Total Crimes of Trust	.096 (.080)	.110 (.152)	.127 (.115)	.119 (.258)	.116 (.085)	.084 (.092)	.082 (.043)

RESULTS

Prevalence of Crimes of Trust and Conventional Crimes

Table 1 presents both raw prevalence data for each age group (in parentheses) and a weighted moving average for each age group (outside the parentheses), the latter constructed by taking the average prevalence of each category with the two adjacent categories (only one adjacent category for transitionals and late elderly), weighted by the number of cases in each category. Employee theft is omitted both as an individual item and as part of the composite measure in Table 1 because it is not available for most of the subsamples (only OR and AO were asked about employee theft at Wave 11). For the raw prevalence data, most of the crimes of trust appear to peak in late middle age, with the exception of sales fraud, which peaks in the transitional (18–24) age group. This pattern is reflected in the prevalence of total crimes of trust, the last line of Table 1.

Because the peak in crimes of trust occurs in one of the smaller age groups (n=151 for late middle age), the results using the raw prevalence data may be unreliable. The weighted moving average does present a somewhat different picture than that offered by the raw prevalence data. For the weighted moving average data, embezzlement has no real peak; it is relatively constant from age 18 to 55, after which it drops to zero, suggesting that retired individuals and individuals late in their career are less likely to be involved in embezzlement than younger respondents. Forgery peaks in early middle age, with prevalence rates dropping below 2% after age 55 (parallel to embezzlement). Check fraud is highest (just barely) for the youngest of the adult age groups, but remains close to two percent throughout the years normally associated with labor force participation (ages 18–65), then drops below 1% after age 65. Credit card fraud appears to peak in young adulthood, with prevalence rates below 1% after age 35. Insurance fraud, interestingly, appears to peak at the oldest ages, and to be least prevalent for the two youngest age groups in Table 1. Tax fraud reaches its peak in the pre-retirement ages (56–65), but remains higher among early and late elderly respondents than among the transitional and young adult age groups. Contrepreneurial sales fraud is the one offense in Table 1 that appears to follow the conventional crime pattern of peaking at younger ages, and is twice as high in the transitional age group as in the other age groups. For the aggregate measure of crimes of trust, the peak occurs in early middle age, and the three age groups with the highest prevalence of crimes of trust are early middle age, late middle age, and pre-retirement. Overall, the early and late elderly have the lowest prevalence of crimes of trust. From Table 1, the peak in the prevalence of crimes of trust appears to occur some time in middle age, between ages 36 and 55.

Table 1 uses cross-sectional data on differences as a proxy for true longitudinal data to examine the distribution of crimes of trust over the life course, in order to present data on a wider range of ages than would be possible with available longitudinal data. Figure 1 uses longitudinal data on the OR to examine the age-crime curve for crimes of trust over a shorter segment of the life course, but using data on the same subsample at each age. Figure 1 presents the weighted prevalence rates for crimes of trust for OR based on data from waves 4 through 11.⁴ For comparative purposes, the rates of conventional offending are also presented. Crimes of trust prevalence rates remain fairly stable from the teens to the early twenties and rise dramatically from the mid-twenties through the early thirties (unless employee theft is included, in which case the rise begins around age 21–22, reflecting the increased opportunity for employee theft as respondents enter the labor market). Data are not available for OR at ages 34 through 35 because of gaps in data collection. The rate of crimes of trust among OR diminishes during the late thirties. A similar trend is seen even when the offenses introduced at waves nine through eleven are excluded. The peak appears to occur a little earlier than in Tables 1 and 2. One possibility is that we simply have better data, including a larger sample, for the OR. Alternatively, it may be the inclusion of employee theft in Figure 1 but not Table 1 that accounts for the lower peak age in Figure 1. A third possibility is that the peak in crimes of trust is now occurring earlier than it used to, perhaps as a result of the changes in the labor force noted earlier. In any case, both longitudinally and cross-sectionally, the pattern of age-specific prevalence rates for crimes of trust differs dramatically from that for conventional offenses, consistent with expectations from the literature on white collar crime. Crimes of trust are clearly more prevalent during adulthood than conventional offending. In fact, crimes of trust appear to be more prevalent during the early to mid thirties than serious (felony level) conventional crimes are during teenage years.

Figure 1.

Age-Crime Curve: Crimes of Trust vs. Street Crimes (Original Respondents)

Table 2.

Cumulative Prevalence of Crimes of Trust by Subsample

	OR		SP		PA		AO		YO
	Waves 1–11 plus retrospective		Wave 11 plus retrospective		Wave 11 plus retrospective		Waves 11–12 plus retrospective		Wave 12
	n = 1,628		n = 843		n = 872		n = 634		n = 872
	%	n	%	n	%	n	%	n	%	n
Embezzlement	7.1	115	0.8	7	0	0	1.0	6	~	~
Employee Theft	20.3	330	~	~	~	~	6.5	41	~	~
Forgery	3.9	58	4.7	40	0.9	8	3.9	25	1.0	9
Fraud: Check	9.1	148	4.2	35	0.9	8	5.7	36	1.6	14
Fraud: Credit Card	3.0	49	1.9	16	0.2	2	2	13	0.2	2
Fraud: Income tax	16.3	239	12.5	105	7.7	67	3.8	24	~	~
Fraud: Insurance	2.3	34	1.9	16	0.7	6	0.3	2	~	~
Fraud: Sales	10.3	168	2.1	18	0.8	7	6.0	38	1.4	12
Total Crimes of Trust	38.2	622	21.2	179	8.9	78	1.9	126	3.8	33

Cumulative Prevalence of Crimes of Trust and Conventional Crimes

Annual prevalence rates identify individuals who commit crimes of trust in a given year, but if we rely only on annual prevalence, many individuals may be incorrectly identified as nonoffenders. To better examine the involvement of individuals in crimes of trust over a longer time span, the cumulative prevalence of crimes of trust was calculated for all respondent groups (OR, PA, SP, AO, and YO) using all available data, including long-term retrospective data described earlier. Individuals who have not been involved in crimes of trust in the past year (annual data) thus have the opportunity to report whether they have ever committed crimes of trust. In Table 2, the cumulative prevalence of each crime of trust, and general crimes of trust, is presented for each respondent group. The figures presented in Table 2 are based on the pooled cross-sectional time series (PCSTS) data. As shown, prevalence rates vary across subsamples and types of offenses, but crimes of trust in general are not as uncommon as one might think based only on results using annual data. Employee theft, tax evasion, and selling worthless goods were among the most commonly reported crimes of trust for each respondent group where data were available. These crimes represent concerns largely marginal to more restrictive definitions of white collar crime, and thus both these specific offenses and the more general composite measure of crimes of trust may yield different results than one would expect based on the literature using those more restrictive definitions of white collar crime.

More than one third (38.2%) of original respondents reported at least one crime of trust from the age of 14 to 43 (excluding ages 34 and 35). For spouses, whom we would expect to have similar cumulative prevalence of crimes of trust, the estimate is roughly half as high overall, and for those offenses available for both OR and SP, spouses report lower cumulative prevalence of all offenses except forgery. Because OR and SP are typically from roughly the same generation, and both are nearly equally divided between males (OR 50%, SP 46%) and females (OR 50%, SP 54%), we might expect that there would be little difference between the two, but recall that the OR include individuals who are neither married nor cohabiting, while the SP are by definition either married or cohabiting. The differences that do exist may, alternatively or additionally, reflect differences between prospective data for the OR and long-term retrospective recall data for the SP, and are comparable in magnitude to the differences in conventional crime for the OR when comparing prospective and long-term recall data (Menard and Elliott 1990). Differences between the OR and the PA reflect not only marriage but also childbearing (some of the married OR do not have children; all of the PA of course do), and may also reflect generational differences in opportunities for crimes of trust, again linked to changes in workforce composition. Differences between the OR and the other groups (PA, AO, YO) may also reflect differences between prospective reports and long-term retrospective reports, but they may also, particularly for the YO, reflect real differences at their specific stage in the life course in the extent of opportunity for and involvement in crimes of trust. Coupled with past findings on long-term recall data, these findings reinforce the concern with using short-term recall or prospective data whenever possible to minimize “false negatives” for the prevalence of crimes of trust, as for other types of crime (Menard and Elliott 1990). Even with these concerns it is clear that crimes of trust are highly prevalent, particularly for middle age respondents (OR, SP).

Cumulative prevalence of crimes of trust is of interest not only for its own sake, but also for comparison with conventional crimes. Table 3 presents the cumulative prevalence and frequency of the three categories of crimes of trust (occupational, including employee theft; avocational; and contrepreneurial) and seven categories taken from Elliott et al. (1989) of conventional crime for the OR, again using PCSTS data. Shown in Table 3 are the cumulative prevalence and frequency for OR for each trust crime grouping (contrepreneurial, avocational, and occupational) and each conventional crime outlined above. Mean frequencies were calculated based on non-zero reports, so they represent active offender rates rather than “incidence” as that latter term is used elsewhere (e.g., Elliott et al. 1989). Categorical data for the years between waves are excluded from the calculation of the mean frequencies, so the data are based only on actual frequency counts for each year. In an effort to account for variation that may be attributable to changes in measurement which began at wave 9 (the first wave at which forgery, insurance fraud, and tax evasion were included, reducing the coverage of avocational crime in the earlier waves), we compare the results using waves 9 through 11 with results using waves 6 through 11. Each crime type is lower in prevalence among respondents in waves 9 through 11 compared to reports from waves 6 through 11, as might be expected. Avocational crimes are the most prevalent crimes of trust among original respondents, possibly because of the larger number of indicators for avocational as opposed to contrepreneurial and occupational crimes. The prevalence of avocational crimes as measured here is similar in prevalence later in adulthood (i.e., waves 9 through 11) to the prevalence of offenses such as minor violence, marijuana use, and hard drug use. Regarding offense frequency, mean differences between the groupings appear not to be pronounced, with the exception of public order offenses, marijuana use, and minor violence.

Table 3.

Cumulative Prevalence and Frequency of Crimes of Trust and Conventional Crimes (OR Age 18 and Older)

	Prevalence OR		Frequency OR
	W6 – W11	W9 – W11	W6 – W11			W9 – W11
	% (n)	% (n)	Mean	S.D.	Reports	Mean	S.D.	Reports
Contrepreneurial	7.39 (117)	3.07 (45)	8.6	25.5	834	11.5	24.06	254
Avocational	23.04 (365)	22.53 (330)	8.8	37.5	2,614	7.8	36.5	1,958
Occupational	18.06 (286)	7.65 (112)	13.9	72.8	3,977	10.9	39.6	1,229
Serious property	14.90 (236)	5.39 (79)	9.0	29.9	2172	11.8	44.5	930
Minor property	21.15 (315)	8.67 (127)	16.0	69.5	5190	15	49.8	1,871
Serious violent	14.38 (227)	4.37 (69)	4.3	12.1	993	5.2	20.6	362
Minor violent	91.53 (1,446)	22.94 (336)	11.7	55.4	16,982	22.3	85.2	7.498
Marijuana	76.47 (1,209)	27.49 (403)	115	323.4	139,612	75.0	280.6	30,284
Hard drugs	41.20 (651)	23.40 (343)	2.0	1.4	1,090	1.3	0.63	461
Public Order	69.51 (1,101)	40.72 (645)	32.4	73.4	35,717	13.8	36.4	8,875

NOTE: Means based on non-zero reports.

Sociodemographic Correlates of Crimes of Trust and Conventional Crimes

In Table 4, we examine the relationship of different types of crimes of trust and conventional crimes to sociodemographic variables for the original respondents. Table 4 presents the bivariate correlations from the PCSTS data for each offense (logged frequencies) against sociodemographic data, excluding age, for waves 6 through 11, with data for waves 9 through 11 in parentheses (again noting that forgery, insurance fraud, and tax fraud are not covered in the data until wave 9). As in Table 3, only the one-year recall data on actual frequency of offending for each wave is used; the categorical longer-term retrospective data for years between waves is omitted. Each correlation was statistically significant at p < .05; however, for crimes of trust, the bulk of relationships are not substantively significant. Reporting trouble with alcohol stands out as being substantively and statistically significantly positively correlated with crimes of trust. Being male was also positively associated with crimes of trust, as was reporting trouble with drugs. For the later waves, being married was negatively correlated with crimes of trust and divorce was positively correlated with crimes of trust. Other relationships between crimes of trust and demographic variables were weak.

Table 4.

Sociodemographic Correlates of Crimes of Trust and Conventional Crimes (PCSTS)

		Male	White	Black	Hispanic	SES
Contrepreneurial		.11 (.05)	−.07 (.05)	.06 (.05)	.00 (−.02)	−.09 (−.03)
Avocational		.09 (.10)	.02 (.01)	−.03 (−.03)	−.02 (−.01)	−.02 (−.01)
Occupational		.10 (.04)	.08 (.04)	−.07 (.04)	.05 (−.02)	−.01 (.00)
Serious Property		.17 (.08)	−.07 (−.08)	.06 (.08)	.02 (−.03)	−.15 (−.10)
Minor Property		.15 (.07)	.00 (.04)	−.06 (−.01)	−.01 (−.02)	−.12 (−.08)
Serious Violent		.19 (.09)	−.09 (−.07)	.07 (.06)	.04 (.03)	−.20 (−.10)
Minor Violent		−.01 (−.06)	−.09 (.04)	−.06 (.00)	−.08 (−.07)	−.16 (−.04)
Marijuana		.15 (.11)	.04 (.06)	−.02 (−.04)	−.02 (−.03)	−.16 (−.15)
Hard Drugs		.04 (.01)	.04 (.00)	−.05 (−.02)	.01 (.00)	−.07 (−.06)
Public Order		.26 (.15)	.22 (.12)	−.21 (−.12)	−.05 (−.03)	−.11 (−.09)
	Education	Trouble with Drugs	Trouble with Alcohol	Married	Divorce	Welfare
Contrepreneurial	−.01 (−.02)	.11 (.05)	.17 (.15)	−.05 (−06)	.08 (.08)	.06 (.14)
Avocational	.04 (.03)	.08 (.07)	.18 (.13)	.01 (−.06)	.11 (.07)	.02 (.01)
Occupational	.02 (.02)	.08 (.07)	.17 (.09)	−.01 (−.06)	.04 (.03)	−.09 (−.03)
Serious Property	−.09 (−.04)	.22 (.20)	.32 (.24)	−.06 (−.14)	.10 (.11)	.12 (.10)
Minor Property	.01 (−.03)	.17 (.19)	.30 (.23)	.00 (−.10)	.10 (.11)	.08 (.06)
Serious Violent	−.10 (−.06)	.22 (.15)	.38 (.26)	−.04 (−.11)	.16 (.13)	.09 (.06)
Minor Violent	−.02 (−.06)	.11 (.10)	.22 (.16)	.09 (.00)	.14 (.09)	.09 (.08)
Marijuana	−.09 (−.09)	.12 (.09)	.36 (.27)	−.09 (−.15)	.12 (.15)	.10 (.06)
Hard Drugs	−.07 (−.08)	.17 (.16)	.31 (.22)	−.02 (−.11)	.15 (.13)	.09 (.08)
Public Order	−.01 (−.02)	.09 (.06)	.46 (.38)	−.04 (−.16)	.09 (.13)	.02 (.05)

Comparing the relationship of crimes of trust and conventional crime with the sociodemographic variables, there is little difference by gender, with both crimes of trust and conventional crimes being at best weakly correlated with being male (negatively in the single case of minor violent offending, which tends to include trivial offenses). For ethnicity, only public order offenses have correlations greater than .10, positively with being white and negatively with being African American (and none of the correlations with being Hispanic exceeds .10). Being married is very weakly negatively related to all except minor violent offenses (and note again that many of these offenses are trivial), no more or less so for conventional offenses (other than minor assault) than for crimes of trust. Being divorced is more strongly related to offending, particularly violent offending, illicit substance use, and avocational crime, and most weakly related to occupational, contrepreneurial, and public order offenses. These findings offer no more than mixed support for the expectation that crimes of trust are less likely than conventional offenses to be associated with marital status. Drug problems appear to be most closely associated with serious offending, both violent and property, also with minor property offending, and of course with hard drug use. Alcohol problems are even more strongly related to conventional offending than are drug problems, and the relationship of alcohol problems to crimes of trust appears to be substantially weaker than the relationship of alcohol problems to conventional crimes. Consistent with expectations based on the literature on white collar crime, then, drug and alcohol problems are less characteristic of trust crime offenders than of conventional offenders.

Socioeconomic status (SES) is weakly negatively correlated with offending, most strongly for serious violent offending, followed by minor violent offending, marijuana use, serious property offending, minor property offending, and public order offenses. This finding is consistent with what we would expect from the literature on white collar crime: being of low socioeconomic status is not a characteristic of white collar crime, and insofar as white collar crimes are crimes of trust, crimes of trust are similarly less strongly correlated with socioeconomic status than conventional crimes. All of the correlations involving welfare status (receiving public assistance) are weak and, except for occupational crime, positive; but individuals who receive public assistance (and, correspondingly, are less likely to be employed) are less likely to be occupational criminals than individuals not receiving public assistance. This is consistent with what we would expect based on the literature on white collar crime. Note, too, that the strongest positive correlation with being on public assistance is with serious conventional property crimes. Education is not strongly correlated with any of the types of crime in Table 4, but it is weakly positively correlated for both Waves 6–11 and Waves 9–11 with avocational and occupational crimes (the correlations for Waves 6–11 and 9–11 are opposite in sign for minor property offenses). This result, too, is consistent with what we would expect from the literature on white collar crime as it pertains to crimes of trust. In greater detail, not shown in Table 4, increased level of education was found to be positively related to crimes of trust for OR and PA, but not for SP, AO, and YO. However, those relationships were still not substantively significant. To the extent that there are differences between conventional crimes and crimes of trust as correlates of indicators of social position (SES, Education, welfare status), the results appear to be consistent with expectations from the literature on white collar crime: crimes of trust are, as expected, less associated than conventional crimes with lower social position.

CONCLUSION

The findings presented here represent a first look at the correlates of several crimes of trust based on a nationally representative multigenerational self report survey. Several interesting findings emerged which contribute new information to what is known about these types of criminal offenses. Our results are somewhat supportive of those based on official data for white collar crime, but the data here go beyond the correlates of crimes of trust available in other data sets. In general, the correlates of crimes of trust appear somewhat similar to those commonly found for conventional offending. Though the directions of these relationships parallel one another, their magnitude is typically weaker for crimes of trust compared to serious conventional offenses. In some cases, that is precisely what we would expect from the literature on white collar crime: weaker relationships for crimes of trust than for conventional crimes with indicators of social position, divorce, and substance use. In a sense, trust offenders are just “less typical”: less typically male, African American, low SES, low education, and involved in substance use problems, than conventional offenders, particularly more serious conventional offenders; and the difference between crimes of trust and conventional crimes is even more pronounced when we consider the relationship of crimes of trust to age.

Conventional wisdom and much of the work in the life course perspective on crime, as described earlier, suggests that crime is a young person’s game. Conventional wisdom may work for conventional crimes, but the most striking difference between crimes of trust and conventional crimes is that crimes of trust peak not in adolescence or early adulthood, but in middle age. Of all the crimes of trust considered here, it is only contrepreneurial crime (sales fraud) that appears to be a young person’s crime, peaking at ages 18–26. That there might be a different age-crime curve for conventional crimes and crimes of trust was anticipated, as noted earlier, based on age-specific opportunities to commit certain types of illegal acts. More strikingly, however, some crimes of trust (here, in particular, insurance fraud) may be higher for the oldest of the old than for younger individuals, hinting not only at differences in illegitimate opportunity, but perhaps also in strain (e.g., Agnew 1992; Cloward 1959) across the life course. Prior research on crimes of the elderly using arrestee, parolee, and incarcerated samples have identified crimes of fraud as one of the more prominent offenses for which the elderly are exposed to the criminal justice system (Gewerth 1988; Phillips 2006; Shichor 1988). While some of the crimes for which the elderly are under the supervision of the criminal justice system may be crimes committed earlier in the life course, fraud appears to be an offense for which the elderly are arrested after they have become elderly, rather than before. We may speculate here that the relatively higher rates of insurance fraud among the elderly found in the present study may be connected to health care or related costs, but we lack the data to pursue this speculation in detail here.

One limitation of the present research is the specific crimes of trust that were available for study. We simply do not have data on some of the white collar offenses that are included in official record data sets that have been used to study white collar crime, and it would be interesting to compare more narrowly defined white collar crimes with the more marginal types of crimes of trust examined here. Another limitation is that questions about crimes of trust were not asked of the youngest respondents in the study, at ages 11–17 (we do have some data beginning at age 14). Many of the crimes of trust considered here could be committed by adolescents (for example, credit card fraud and employee theft). For the subsamples other than the OR, to examine cumulative prevalence of crimes of trust, it was necessary to rely on long-term retrospective recall data, which, as noted in previous studies, tends to underestimate prevalence; and that was evident in the comparison of the OR and SP subsamples on cumulative prevalence. Finally, although we were able to cover a range of ages from 18 to 88 (in some cases, 14 to 88), the PA and SP subsamples are not nationally representative samples of individuals in those age groups. Mitigating that concern somewhat, however, is the absence of any substantial relationship between being married and crimes of trust, the feature that most clearly distinguishes the SP and PA subsamples from a representative national sample.

This study sets a stage for future research aimed at exploring self reported crimes of trust, including but not limited to white collar crimes, as a category of offenses distinct from crimes of violence, conventional theft and vandalism, and public order offenses. Future studies of self reported crimes of trust should seek to evaluate in more detail how and whether conventional criminals differ from trust crime offenders. Of particular interest would be an exploration of situational factors as they interact with trust crime offending throughout the life course, as suggested by Piquero and Benson (2004) for white collar crime. It would also be interesting to go beyond an examination of correlates of crimes of trust to causal models providing more rigorous tests of existing theories of crime to see whether they are equally applicable to crimes of trust and other forms of offending.

Biographies

Scott Menard is a Professor of Criminal Justice at Sam Houston State University. He received his A.B. at Cornell University and his Ph.D. at the University of Colorado, Boulder, both in Sociology. He has been the principal investigator on the Evaluation of Bully-Proofing Your School project (NIJ-2004-IJ-CX-0082) and the National Youth Survey Family Study (NIAAA-AA11949 and NIDA-DA015983). His publications include books, monographs, and articles on quantitative methods and statistics, demography, neighborhoods and crime, youth gangs, criminal victimization, life course criminology, substance abuse, and the relationships among delinquency, substance use, and mental health problems.

Robert G. Morris is an Assistant Professor of Criminology at The University of Texas at Dallas. He received his Ph.D. from Sam Houston State University in criminal justice. His research is focused upon the etiology of technology driven crimes, frauds. He also researches contemporary issues in corrections research. His recent work has appeared in Criminal Justice and Behavior, Journal of Criminal Justice, and Deviant Behavior.

Jurg Gerber earned his Ph.D. at Washington State University and is currently a professor of criminal justice at Sam Houston State University. His research interests include white collar crime, drug policy, and comparative criminology. He has co-edited three books on drug policy and white collar crime, and has also published on comparative criminology, criminal justice education, and prison education. As part of his interests in comparative criminology, he has held a visiting professorship in Switzerland and has served as a Fulbright scholar in Kaliningrad, Russia. He has delivered a number of invited lectures in Poland, Germany, Korea, Thailand, China, and Japan.

Herbert C. Covey is an instructor at the University of Colorado at Boulder. He received his Ph.D. from the University of Colorado in Sociology. He served as the Vice-Chair of the State Juvenile Parole Board for 13 years. His publications include books and articles on methods, youth gangs, African American history, criminal victimization, crime measurement, methamphetamine use, history of aging and social gerontology.