Religious Programs, Institutional Adjustment, and Recidivism
among Former Inmates in Prison Fellowship Programs{1}

BYRON R. JOHNSON
Lamar University

DAVID B. LARSON
National Institute for Healthcare Research
Duke University Medical Center

TIMOTHY C. PITTS
Morehead State University

JUSTICE QUARTERLY, Vol. 14 No. 1, March 1997
© 1997 Academy of Criminal Justice Sciences

This study examines the impact of religious programs on institutional adjustment and recidivism rates in two matched groups of inmates from four adult male prisons in New York State. One group had participated in programs sponsored by Prison Fellowship (PF); the other had no involvement with PF. PF and non-PF inmates are similar on measures of institutional adjustment, as measured by both general and serious prison infractions, and recidivism, as measured by arrests during a one-year follow-up period. However, after controlling for level of involvement in PF-sponsored programs, inmates who were most active in Bible studies were significantly less likely to be rearrested during the follow-up period.

As long as there have been prisons, religious education and training have been offered to prisoners. Religious programs for inmates are not only among the oldest but also among the most common forms of rehabilitative programs found in correctional facilities today. This high prevalence of use is confirmed by the U.S. Department of Justice (1993), which reports representative data on American's prison population. After admission to prison, 69 percent of inmates report having working assignments, 45 percent report participating in some form of academic education, and 31 percent report attending vocational training.{2} Among all other types of personal enhancement programs offered in prison, religious activities attracted the most participation: 32 percent of the sampled inmates reported involvement in religious activities such as Bible studies and church services, 20 percent reported taking part in self-improvement programs, and 17 percent reported that they had been involved in counseling. This national survey verifies what many correctional practitioners and volunteers have observed for years: Many inmates continue to attend and participate in religious programs.

These percentages are quite revealing, nearly one inmate in three is involved in religious programs. Yet despite these figures, only a handful of published studies (Clear et al. 1992a, 1992b; B. Johnson 1984, 1987a, 1987b) have examined the influence of religion and religious beliefs or practices on key prison predictor and outcome measures such as inmates' adjustment and recidivism.

In contrast to the corrections literature, the broader criminological topic of religion and deviance has received ongoing, although still sporadic, scholarly attention (see Evans et al. 1995). Researchers in the latter area usually consider the influence of religious commitment, as measured by frequency of religious practice such as church attendance or by the importance of one's religion, on some indicator of deviance. For example, researchers have studied single items of religious commitment and their associations with an array of delinquent activities such as alcohol abuse, drug use, and sexual activity (B. Johnson 1987).

The scarcity of research about prisoners and the influence of religious variables on inmates' adjustment and recidivism can be attributed to potential problematic biases held by both religious workers and scientific researchers (Larson et al. 1986; Larson et al. 1995; Larson, Sherrill, and Lyons 1994; Post 1995). Many chaplains, ministers, and religious volunteers who work in religious programs have been reluctant or have lacked the skills to undertake publishable research. This reluctance had been fueled by a broader historical skepticism about the relevance of religion held by many in higher education, and at best by university researchers' ambivalence in studying spirituality or religion (Jones 1994, Larson et al. 1994).

RELIGION AND REHABILITATION

Tocqueville (1833) was correct when he observed that rehabilitation and reformation are two different things: We can measure rehabilitation, but we cannot measure reformation. Though we can observe that some inmates unquestionably may have changed in positive ways, there is no way of observing whether they have "repented." The former inmate still may be a "very bad" person "on the inside," but as long as he does not commit illegal acts when he leaves prison, as he did previously, he is "rehabilitated." This intriguing distinction warrants research.

Other than anecdotal accounts, the literature contains relatively little research linking the possible influence of religion to its potential beneficial, neutral, or even harmful influence on the rehabilitation of offenders. This deficit is somewhat surprising because most of the published studies have found inverse and essentially beneficial relationships between measures (usually single-item measures) of religiousness and various measures of crime or delinquency. In a recent, important study, Evans et al. provide perhaps the most comprehensive treatment of the topic and conclude that "among our religiosity measures, participation in religious activities was a persistent and noncontingent inhibitor of adult crime (1995:195)." This finding held even after controlling for secular constraints, religious networks, and social ecology. The results of that study are noteworthy because some research has shown that various measures of religiosity or religious commitment lose their significance in the presence of secular controls (Albrecht, Chadwick, and Alcorn 1977; Cochran, Wood, and Arneklev 1994; Elifson, Peterson, and Hadaway 1983). Evans et al. (1995) succeeded in resolving some of these issues, however, by testing the religion-crime relationship with a comprehensive crime measure and three separate dimensions of religiousness, not merely with a single item. Their results are consistent with other published research showing that religion has similar benefits in other areas of impulse control or deviance, such as drug and alcohol abuse (Gartner, Larson, and Allen 1991; Larson et al. 1994).

If religion can inhibit delinquent and criminal activity, why might it not facilitate the process as well as the outcomes of prison rehabilitation? Aside from complex and difficult theological discussion about the possible spiritual roles of religion, as well as the evidence demonstrating the mental and physical health benefits of religion (Bergin 1983, 1991; Gartner et al. 1991; Larson, Sherrill, and Lyons 1994; Levin and Vanderpool 1987, 1989), there are scientific reasons to predict that religion might effect behavioral and social change. Religion targets antisocial values, emphasizes accountability and responsibility, changes cognitive approaches to conflict, and provides social support and social skills through interaction with religious people and communities (Bergin 1991; Levin and Vanderpool 1987; Martin and Carlson 1988). Such emphases seem to be consistent with what many rehabilitation workers would call principles of effective treatment.{3}

THE RELIGION-PRISONER LITERATURE

Byron Johnson (1987a) studied the influence of three different religiosity indexes in explaining the institutional adjustment of Florida prison inmates. He operationalized the dependent variable, institutional adjustment, by creating an index of the amount of time served in disciplinary confinement as a result of violating institutional rules. The religiosity indexes were based on and derived from data contained in the files of 782 former inmates of the Apalachee Correctional Institution in Florida from 1978 through 1982.{4} One index was based on the inmate's self-reported religiousness; a second was based on the chaplain's perceptions of the inmate's religiousness; the third was based on the inmate's attendance at church-related activities in prison. The findings revealed that none of the three indexes were related in any significant way to institutional adjustment.

In a larger and more recent project, Clear et al. (1992a, 1992b) studied inmates in 20 prisons in 12 states, representing the different regions of the country as well as different security levels and classifications of inmates. Unlike Byron Johnson (1987a, 1987b), Clear and his colleagues were able to study religion in prison more comprehensively. They did so by using ethnographic research with a research design that included (1) administration of questionnaires to inmates, (2) focus-group interviews with religious and nonreligious inmates, and (3) interviews with chaplains, administrators, and correctional officers. From a previously developed questionnaire (Hunt and King 1971) they adapted a multifaceted religiousness measurement incorporating prison contexts to assess inmates' commitment to their religious beliefs as well as religious practice and attitude items. Clear and his colleagues found that religiousness was related both to improved adjustment and to fewer institutional infractions. After controlling for the influence of several key variables (age, race, and self-esteem), religion was no longer associated significantly with prison adjustment. However, the significant inverse relationship between religiousness and institutional infractions remained predictive and statistically significant.

Both Johnson (1987a, 1987b) and Clear et al. (1992a) recommended not only more research in this area but research that examines more reliably and more comprehensively the dynamic interplay between religion and prisoners' adjustment, recidivism, and postrelease success.

METHODOLOGY

In 1991 Prison Fellowship (PF), a nonprofit religious ministry to prisoners founded by former Nixon aide and Watergate figure Charles Colson, commissioned a study of PF programs. The study culminated in a final report to PF but was not published in a journal.{5} Subsequently we obtained permission from Prison Fellowship to reexamine the data by looking at institutional adjustment and postrelease arrests, in an effort to determine whether the various types of religious programs offered by PF changed former inmates' behavior.

The Sample

We selected four prisons in New York State for the study.{6} Though originally we had intended to study more prisons than these four, the original research team reported that consistent and detailed record-keeping was new to PF staff members and volunteers, and was difficult to implement fully in all of the state's prisons. Consequently, we chose these four prisons because the data on PF program participation for 1992 were far superior than for other New York State prisons offering PF programs. This data collection represented PF's first attempt to gather useful and accurate data on program participation. Though the mechanistic rationale for selecting these prisons is somewhat understandable, given the technical problems of data collection, the sample is not random; nor is it representative of inmates in general or of New York State inmates in particular.

Prison Fellowship participation data pertained to 201 male inmates from the four prisons who had participated in at least one of three PF activities: Bible studies, in-prison seminars, or life plan seminars. Inmates could participate at least once in all three kinds of activities, or several times in only one. Therefore varying degrees of involvement in PF programs were possible, and in fact were borne out in the data.

The New York Department of Corrections, Division of Research and Planning, provided data on criminal history, prison adjustments, and recidivism for the entire cohort of inmates released from the New York State prison system between January 1, 1992 and April 30, 1993, almost 40,000 inmates. When the one-year follow-up period had elapsed, the New York Department of Corrections provided data on each individual in the study who had been rearrested.

The Matched Comparison Group

We selected a matched group of inmates by first excluding all known PF program participants from the release cohort of approximately 40,000 inmates. On the basis of a multivariate sampling method, seven variables most strongly predicted members of the PF groups: age, race, religious denomination, county of residence, military discharge, minimum sentence, and initial security classification.{7} The 201 former inmates most closely matching PF inmates on these seven variables were selected for inclusion in the non PF matched group; that is, each member of the PF group was assigned an individual non PF match. Table 1 displays descriptive data on selected variables in the form of frequencies and means for the PF and non-PF inmates. Though the two groups match closely on study variables such as denomination (8 percent Muslim, 41 percent Protestant, 39 percent Catholic), marital status (28 percent and 26 percent married, respectively), age (32 years), education (tenth grade), and race (47 percent black, 12 percent white, 40 percent Hispanic), Hispanics and Catholics represent a higher proportion of both samples than one might expect given their national distributions.

Religious Program Measures

In Prison seminars. The original purpose of the in-prison seminar (IPS) was to "disciple Christian inmates in their walk with Christ so that they, in turn, could have an impact on other inmates in the name of Christ" (Clower 1991:9). In recent years, however, the focus of the three-day IPS has been to introduce the unchurched inmate to the faith and then to provide teaching or discipling.

Life plan seminars. Life Plan Seminars (LPS) were implemented to meet the needs of prisoners who were soon to leave prison. The LPS was designed to help those inmates develop realistic attitudes and plans for their lives and thus to enhance their success after release. Therefore the objectives of LPSs differ considerably from the "spiritual growth" emphasis found in the longer-established PF Bible studies and IPS programs. The seminars, taught by PF-trained LPS instructors, last two to three days and target the building of inmate-mentor relationships within six to 12 months before release (Bence 1994).

The original purpose of PF Bible studies was to provide inmates an ongoing opportunity to study God's Word and to enjoy Christian fellowship....The Bible studies were seen as an opportunity to offer inmates a way to "grow in the Lord." Over time, Bible studies came to be seen as an excellent way to follow up on inmates who attended an In-prison Seminar (Clower 1991:8).

PF Bible studies have been available to the nonreligious, but are attended most frequently by those who are already committed in their faith. Bible studies typically meet once a week for one or two hours and include a time of "fellowship and sharing, scripture study, and praise and worship" (Clower 1991:9). Bible studies are led by PF volunteers; these volunteers usually come from local churches and, like IPS instructors, have taken a PF training course.

A series of multivariate analyses revealed that attendance at Bible studies was the most reliable mechanism for characterizing the level of PF participation (we return to this point later). PF inmates who did not attend any Bible studies were classified as low participants (49 percent); those attending one to nine Bible studies over the course of one year were considered medium participants (40 percent); those who attended at least 10 (or about one per month) were considered high participants (11 percent). We found a very low level of participation for many in the PF groups of inmates. Of the 201 PF inmates, for example, 49 percent had never attended a Bible study; another 17 percent had attended only one Bible study during a year. Similarly, 61 percent had attended an in-prison seminar but only 16 percent had attended a life plan seminar.

Measures of Institutional Adjustment

Many researchers have discussed the sociological and psychological uniqueness of the prison environment and how coping with such institutions consequently requires a great deal of adjustment by inmates (Clemmer 1958; Irwin 1970, 1980; Jacobs 1977; R. Johnson 1976; Johnson and Toch 1982; Sykes 1958; Toch 1975, 1977). Therefore correctional researchers often study offenders' adjustment to prison life. One of the most widely accepted approaches to this subject is to examine how often prisoners are in trouble or are caught breaking prison rules. The documented violation of prison rules provides a measure long associated with adjustment (Wolf, Frienek, and Schaffer 1966).

Offenders committing institutional infractions have generally been viewed as adjusting poorly to prison life. In addition, records documenting frequent infractions have been identified by researchers as predicting postrelease recidivism (Flanagan 1983). Some researchers, however, have questioned the validity of this measure (Gottfredson and Adams 1981; Lipton, Martinson, and Wilks 1975; O'Leary and Glaser 1972). Despite the lack of consensus on the validity of infractions for predicting adjustment or recidivism, we believed that possible interplay between religious involvement and institutional misconduct deserved empirical inquiry. We measured infractions in two ways: (1) by examining those inmates who had committed at least one infraction (coded 1) and those who had committed no institutional infractions (coded 0); and (2) by examining the seriousness of the infractions (one or more serious infractions, coded 1; no serious infractions, coded 0).{8}

Recidivism Measure

Researchers have operationalized recidivism with a variety of measures including (1) all arrests and/or charges, (2) technical violations and/or revocations, (3) convictions, or (4) incarceration. Consequently findings on recidivism commonly yield different rates, depending on the measure (see Beck 1989; Greenfield 1985; Maltz 1984; Schmidt and Witte 1988).

We measured recidivism with a single factor: any arrest of a former inmate during a one-year postrelease period. We know the limitations of this definition, but the constraints of the available data made it impossible to collect more accurate measurements. We also know that the follow-up period of one year is less than optimal and that recidivism rates depend on the time window used (Allison 1984; Maltz 1984; Schmidt and Witte 1988, 1989). Though researchers obviously recognize that the likelihood of arrest increases with the size of the window, most post-release arrests occur within the first year of release (Bureau of Justice Statistics 1989; Joo, Ekland-Olson, and Kelly 1995; Minor and Courlander 1979).

Recidivism researchers commonly examine the relationship between inmates' risk levels and postrelease behavior (Joo et al. 1995; Turner and Petersilia 1991). In keeping with other research on recidivism rates, we developed a measure of risk to be applied to both the PF and non-PF groups. Our measure was based on three variables traditionally used in predicting postrelease behavior: (1) previous record of violence, (2) previous adult convictions, and (3) disciplinary record. We operationalized the index of risk as low, medium, and high.

FINDINGS

Institutional Adjustment

As shown in Table 2, the two groups are very similar with regard to committing any institutional infractions as well as serious infractions (e.g., assault, attempted escape). During the one-year study period preceding release from prison, 38 percent of the PF groups and 31 percent of the non-PF group committed any institutional infractions. PF and non-PF inmates are almost equally likely to commit (or not to commit) serious infractions while in prison: 8 percent of the PF group and 9 percent of the non-PF group committed such infractions during the study period. A chi-square test showed no significant difference between the PF and the non-PF groups.

Increased church attendance or religious practice has often been associated negatively with crime and/or delinquent activity (B. Johnson 1987a; Evans et al. 1995). Therefore we examined the level of participation in PF programs as a possible indicator of degree of commitment to ideals and values that might encourage moral decision making and reduce the likelihood of impulsive decision making. In keeping with much of the religion-crime literature, one could argue that inmates most strongly committed to religious activities, as measured by attendance and involvement in PF programs, might be (1) less likely to violate any institutional rules while incarcerated and (2) particularly less likely to commit more serious infractions.

Table 3 provides a breakdown of institutional infractions by level of participation for PF and non-PF inmates. As mentioned above, one might expect high-PF participants to be less likely to break institutional rules. However, a chi-square test showed no significant difference (at alpha = .05) between low, medium, and high participation groups with regard to committing institutional infractions, though high PF participants were slightly less likely than their non-PF counterparts to commit infractions.

Table 3 also shows no significant difference between PF inmates and non-PF inmates with regard to serious prison infractions. Somewhat unexpectedly, however, high PF participants (and their non-PF counterparts) were more likely than those in the low and medium PF participation categories (and in the non-PF sample) to have committed a serious infraction.

Because we do not know the sequence of institutional infractions and participation in PF programs, we cannot address this rather paradoxical finding. Do inmates get into trouble and then attend religious programs, or do they attend religious programs and then commit serious rule violations? We suggest the former sequence (getting into trouble and then turning to religion) as more plausible. This possible explanation is also supported in the "coping" literature, which links the onset of physical illness and mental stress to a greater tendency to turn to religion or spirituality (Idler and Kasl 1992; Koenig, Cohen and Blazer 1992; O'Brien 1982). Inmates in serious trouble with institutional authorities may be more likely to attend a religious function if they believe it may benefit them in some way. Religion offers the opportunity to gain forgiveness , to start over, to receive another chance, and even to attempt to cover over (or cover up) their prison "sins." Religion can be comforting, or even a tool for manipulation. In an environment where status degradation is common, spirituality can be empowering, especially to those who find themselves in trouble and alone (Clear and Myhre 1995). Therefore those who have committed a serious infraction may be more likely to seek out religious programs.

Recidivism

Table 2 also reports the percentages of inmates in both PF and non-PF groups who were arrested after release from prison. A chi-square test shows no significant difference between the two groups: 37 percent of PF inmates, versus 36 percent of non-PF inmates, were arrested during the one-year follow-up period. As seen in Table 3, however, after controlling for level of participation in PF-sponsored Bible studies, PF inmates in the high participation category (10 or more Bible studies) were significantly less likely than their non-PF counterparts (alpha = .05) to be arrested during the follow-up period (14 percent versus 41 percent). A chi-square analysis also shows that high PF participants were significantly less likely to be arrested than PF inmates in either the medium or the low participation categories.

In the next phase of the analysis, we examined recidivism and the influence of risk level. As expected, low-risk offenders were the least likely to be arrested after release (8 percent). Medium-risk offenders were more than twice as likely as low-risk offenders to be arrested (19 percent). Offenders in the high-risk category were most likely to be arrested: 75 percent of all high-risk offenders for the total PF and non-PF sample were arrested during the follow-up period. Table 4 shows the percentages of PF and non-PF inmates arrested after release from prison, with controls for PF participation level and level of risk. For all three levels of risk, PF inmates in the highest participation category were least likely to be arrested.{9}

We also computed a likelihood-ratio chi-square for a contingency table between arrest after release and interaction effects between group (PF versus non-PF), risk (low, medium, and high), and Bible study participation (low, medium, and high). This chi-square can be partitioned into subgroups of interactions (Blalock 1979:297-99); this procedure allows us to identify the portion of the table that contributes most to the association with arrests. A partitioning of this table, separating out arrest after release versus level of risk, shows a chi-square of 177 (df=2, p<0.001). The remainder of this chi-square value (192.725 - 177.102 = 15.623) represents the association between arrests after release and group (PF inmates and their non-PF matches) by level of Bible study participation, summed over risk. Partitioning chi-square in this way shows (1) that the main effects between PF participants and the matched non-PF inmates are not significant with regard to rearrest (X² = .097, p = .756); (2) that level of risk is the variable most highly associated with postrelease arrest (p < .001); and (3) that Bible study participation among PF participants is related significantly to a lower likelihood of arrest during the follow-up period compared to the matched non-PF group (X² = 24.082, df = 5, p < .001).

Using a stepwise inclusion technique, we regressed recidivism on a group of explanatory variables with logistical regression.{10} The results in Table 5 show the logit model that provides the best fit; this model includes the variables Bible study, prior adult convictions, second felony offender, record of violence, Hispanic, and young. All six variables are correlated significantly with recidivism. Thus, being Hispanic and attending 10 or more Bible studies decreases the likelihood of being arrested during the follow-up period. Similarly, being a second felony offender, being age 31 or younger,{11} having a previous record of violence, or having prior adult convictions increases the likelihood of being arrested during the one-year study period following release from prison.{12}

DISCUSSION

This study adds to the limited literature on the role of religion in the lives of prisoners and former prisoners. Although religion is frequently practiced in prisons, either privately or programmatically, criminal justice research suffers in both quality and quantity in examining the impact of religion on prison process and outcome factors. We recognize the limits on the value of initial studies such as this, particularly because the study suffers from a number of shortcomings. First, the group of Prison Fellowship inmates chosen for inclusion in the study were not selected in any representative or systematic fashion. These 201 men simply were the population of inmates who were identified as participants in PF programs at four different prisons in New York State. The PF programs at these four sites provided a sample of convenience. We selected these four prisons because they had PF staff members and prison volunteers who were more strongly committed to collecting site data than were staff members at the other prisons in the state. As an example of sampling deficiency, this limitation accounts for the overrepresentation of Hispanics and the underrepresentation of whites in the sample (40 percent and 12 percent respectively). Second, this oversampling is problematic because Hispanics were the least likely to recidivate. Consequently our findings should be interpreted cautiously because they are not generalizable to New York or other states.

In this study, we sought to determine whether religiously involved inmates, as measured by participation in volunteer-sponsored religious programs, would be less likely (1) to violate prison rules, and especially to commit more serious infractions, and (2) to break the law or commit an act that could lead to an arrest during community supervision in the first year after release from prison.

We found no overall difference between PF inmates and non-PF inmates on measures of institutional adjustment or recidivism. Certain differences emerge however, after controlling for level of participation in PF programs. Participation in two of the PF programs, the in-prison seminars and the life plan seminars, failed to influence either institutional adjustment or recidivism. However, the inmates most actively involved in PF Bible studies (high participants), operationalized as attending as few as 10 or more Bible studies over the course of one year, differed not only from the non-PF matched group but also from medium and low PF participants. High PF participants (1) were as likely to commit infractions as either low or medium PF participants, but were less likely than their non-PF counterparts to commit any institutional infractions; (2) were slightly less likely than their non-PF matches to commit serious institutional infractions, but were significantly more likely than low or medium PF participants to do so; and (3) were significantly less likely than low or medium PF participants as well as their non-PF matches to be arrested during the one-year follow-up period.

High PF participants, as measured by 10 or more Bible studies, were less likely to be arrested during the year-long follow-up period, but only 22 of 201 inmates (11 percent) were in the high participation category. In addition to the small cell size, we do not necessarily recognize that the measure we used--attending only 10 or more Bible studies in a one-year period--is a religiously acceptable definition of high religious participation. Most studies in the clinical research literature, for example, use measures such as weekly (or higher) church attendance (Gartner et al. 1991; Levin and Vanderpool 1987). With these issues in mind, we are impressed by the significant reduction in recidivism among PF inmates who participated in as few as 10 Bible studies during the one-year study period. Another possible interpretation is that because high participants were more likely to be "already committed in their faith," we may be observing a selection effect. Consequently one could argue that participation in Bible study is most frequent among inmates most strongly committed to succeeding after release. Even such an interpretation, however, would suggest that religious commitment may be related to other factors that enhance postrelease success.

Because Bible studies meet weekly, perhaps a more adequate single-item criterion would have been attendance at half or more of the Bible studies: that is, every other week, or 25 times, not 10 times, in a given year. This higher cutoff would have left only 10 inmates in the high participation group (none of whom, incidentally, were arrested during the follow-up period), rendering the results intriguing but making the category too small for comparative analysis. This research, however, like other published research on this subject, needs to improve its measurement not only of what is religious, but also of what is spiritual, with or without religion. In addition, researchers must include measures that more adequately address issues of religious commitment, including religious practices, attitudes, and beliefs (Bergin 1983; Larson et al. 1986; Larson et al. 1994; Gartner et al. 1991).

A final limitation of this study is that the one-year follow-up period is not optimal for determining the influence of religious programs on recidivism over time. Though one-year recidivism studies can provide meaningful data (MacKenzie et al. 1995), research that allows for several more years of observation is more likely to capture the great majority of those releasees who will fail during the observation period (Schmidt and Witte 1988, 1989).

Finally, the literature on recidivism contains few empirical accounts in which correctional programs or treatment interventions have lowered recidivism rates substantially. Consequently, most criminal justice researchers would agree that it is very difficult to influence recidivism rates. Though this study has several acknowledged limitations, our findings at least suggest that religious programs have the potential to affect former inmates' behavior after release. Because religious programs are prominent in most prisons and because such programs are used widely, it would seem logical to coordinate the two themes more fully in future research--first, by assessing the influence of religion or religious programs on adjustment to prison and secondly, by examining the long-term recidivism patterns of inmates who are released. In addition, research must measure assessments of religion or religious practices more accurately and track them more reliably to determine how such assessments affect prison adjustment and postrelease behavior over longer follow-up periods. For example, our study included no postrelease information on offenders (other than their later arrest); future research must explore how Bible studies or similar religious practices or participation may facilitate an offender's integration into a more prosocial, more caring, even more moral community upon release.

According to DiIulio (1991), the relationship between social science research, (on one hand) and correctional policymaking and administrative practice (on the other) is essentially nonexistent. He states, "There is no meaningful body of social science research on corrections" (1991:121). DiIulio further argues that this relationship has been inadequate because it has not brought together practitioners and researchers. If and when such a collaboration begins, it will demonstrate the type of applied, relevant research that can and should take place. Researchers must begin to examine elements that are prevalent in the prison context and show much promise, such as religious factors; in addition, they must approach these factors with improved research measurement and study designs that will provide correctional decision makers with applied, useful, and understandable social science data. Though we are not as pessimistic as DiIulio we tend to agree with his view if these research steps are not taken. Even though our study lacks random assignment and has other methodological limitations, as described above, we believe it is an important first step in initiating the type of social science sought by DiIulio.

We are also heartened to know that the study of religion in criminology is now receiving scholarly attention. In fact, our findings are consistent with those of a recent and important study of religion and crime (Evans et al. 1995), which found that participation in religious activities was a persistent and noncontingent inhibiter of adult crime. The authors of that study conclude by stating that "researchers are beginning to understand the processes that may link religion and crime, but much more work remains" (1995:212). We agree fully, and hope that this initial study is only the beginning of such needed social science research in many more correctional contexts.

ENDNOTES

{1}This research was funded through the generous support of The John Templeton Foundation, Radnor PA. Additional support was provided by Monarch Pharmaceuticals, a Division of King Pharmaceuticals, Bristol, TN. We would also like to thank Dr. Frank Cullen and anonymous reviewers for helpful comments and suggestions.

{2}The Bureau of Justice Statistics conducted this study in the summer of 199 i, conducting face c face interviews with 13,986 inmates from 277 correctional facilities in 14 states. The sampling design used a stratified two-stage selection process whereby the prisoners interviewed were selected from more than 711,000 adults held m state correctional facilities throughout the United States. Similar surveys were conducted in 1974, 1979, and 1986.

{3}Chaiken (1989), Gendreau and Ross (1987), and Palmer (1992), among others have written about principles or strategies for effective interventions

{4}During this period, Apalachee Correctional institution was an all-male, adult, minimum-security prison for first-time commitments.

{5}This study culminated in a report titled The New York Study of Prison Fellowship Programming Executive Summary and Final Report (O'Connor et al 1994).

{6}Arthur Kill, Bare Hill, and Franklin are medium-security prisons, Elmira is a maximum security facility.

{7} We included eight other variables in the multivariate matched sampling method, but dropped them because they were not predictive for those PF inmates who made up the group These variables were: offense type, drug used in crime education, most serious prior record, second felony offense, alcohol use, marital status , and maximum sentence

{8} For example, escapes or assaults would be considered serious infractions.

{9}Some cells in Table 4 have expected counts of less than 5. The chi-square test may not be valid in such cases.

{10} The stepwise inclusion technique identifies variables to be included in the model by calculating the adjusted chi-square statistic for each potential addition at each stage. The variable with the highest chi-square value (alpha = .05) then is added to the model Previously added variables then are reevaluated to determine whether they still meet the minimum requirement. The procedure is repeated until no variable meets the criterion

{11} Because a person's age in years is not a continuous variable (it is measured in discrete increments) and because there is no reason to believe (for instance) that equivalent declines in recidivism rates occur between ages 20 and 26 and between ages 50 and 55, the model, as calculated, includes the variable young, a binary variable taking a value of 1 when age is less than or equal to 31 and 0 otherwise. Though the choice of 31 as the cutoff age was a function of the age distribution of the two samples (31 is the median age for both groups), parameter estimates vary minimally with variation in that age breakpoint.

{12} Though survival analysis is a common feature in recidivism studies, we could not use such an analysis here because of (1) the small number of rearrested offenders in the high participation category and (2) the observation that those few who were rearrested happened to be arrested almost immediately after release. The small number of rearrests and the rapid rearrested of the few persons arrested cause a downward bias in the mean.

REFERENCES

© 1997 Academy of Criminal Justice Sciences