Assessing Attitudes Towards Defendants with Mental Illness
R. Alan Thompson, Ph.D., Associate Director,
Mississippi Statistical Analysis Center
During the budget crises of the 1980s and 1990s, public institutions that cared for the mentally ill began closing, leaving these individuals with no real alternatives for effective treatment. Many found themselves unable to find jobs and adequate shelter, and in short order they ran afoul of the law and became chronic offenders. This gradual shift of responsibility for handling the mentally ill - from public mental hospitals to the criminal justice system - may best be described as the "criminalization of mental illness." More simply stated, the criminal justice system now bears considerable responsibility for responding to both the immediate and long-term needs of a unique population and an exceedingly complex social problem.
It is estimated that today in excess of 1.5 million individuals are incarcerated in state and federal prisons. Studies suggest conservatively that as many as a quarter of a million inmates suffer from mental illness, and that there may be more mentally ill individuals in jails and prisons than there are in mental hospitals.
In light of this situation, it is imperative that we better understand how the contemporary criminal justice system responds to its broadened public welfare mandate. The Mississippi Statistical Analysis Center recently undertook an exploratory research initiative that surveyed judges, prosecutors, and public defenders within the state to assess their beliefs, perceptions, and attitudes toward defendants with mental illness. This group was chosen because of its significant role not only in the adjudication process, but also in determining public risk as well as appropriate methods of treatment and/or confinement. This article highlights some of the study results and identifies their implications for policymaking.
The instrument developed for the study was based on three existing scales with a history of use for assessing public attitudes regarding mental illness. These included adapted versions of the Attitudes Toward Mentally Ill Offenders (ATMIO) scale, the Community Attitudes Toward the Mentally Ill (CAMI) scale, and the Self Stigma Mental Illness Scale (SSMIS). Also included were originally conceived and experiential items, demographic measures, and one open-ended question. Thus, the final form of the instrument consisted of 79 items total. Sixty were attitudinal in nature and were based upon a traditional five-point Likert continuum ranging from "strongly disagree" to "strongly agree."
The self-administered survey instrument was distributed by U.S. mail to members of the target population during the summer of 2013. Of the 539 surveys distributed, 169 were returned and used for analysis, for an overall response rate of 31%.
The results presented here are excerpted from the complete study report, which can be viewed at: http://mssac.org/wp-content/uploads/2012/06/Perceptions-of-Defendants-with-Mental-Illness.pdf. In the full report, results are presented in six sections that contain: 1) demographic information, which provides a general "profile" of respondent characteristics; 2) descriptive results associated with each survey item; 3) results of the reliability analysis including the extent to which study participants were consistent in their expressed beliefs, perceptions, and attitudes; 4) summated scores on each of the adapted scales and subscales; 5) results of bivariate analyses between various demographic variables and the survey items; and 6) verbatim comments by participants in response to an open-ended solicitation for qualitative input.
In the interest of brevity, the demographic profile and results of the frequency and bivariate analyses are excluded from this article. What follows, then, are the results of the reliability analysis and summated scale scores, along with a brief discussion and interpretation of each.
Cronbach's alpha coefficient was computed for the full instrument and for each of the adapted scales and subscales. This coefficient represents a measure of internal consistency, which may be defined as the extent to which a set of items (such as the survey items used in this study) are closely related as a group representing some underlying dimension or latent construct. Generally speaking, a coefficient of .70 or higher is considered acceptable for this type of research.
The table below presents the reliability coefficients for all participants on all 60 Likert-type survey items, as well as across all adapted scales and subscales. The reliability coefficients for each group - judges, prosecutors, and public defenders - are also shown.
|Full Instrument (60 items)||.947||.923||.933||.939|
|Adapted ATMIO scale (20 items)||.886||.811||.857||.870|
|Positive Attitudes subscale (8 items)||.851||.755||.827||.832|
|Community Risk subscale (4 items)||.592||.436||.556||.567|
| Rehabilitation/Compassion subscale
| Diminished Responsibility subscale
|Adapted CAMI scale (22 items)||.857||.829||.810||.845|
| Anti-Authoritarianism subscale
|Benevolence subscale (9 items)||.663||.597||.584||.642|
| Community MH Ideology subscale
| Anti-Social Restrictiveness subscale
|Adapted SSMIS Agreement subscale
|Originally Conceived Items (9 items)||.693||.598||.765||.613|
Note. ATMIO = Attitudes Toward Mentally Ill Offenders; CAMI = Community Attitudes Toward the Mentally Ill;
MH = Mental Health; SSMIS = Self Stigma Mental Illness Scale.
As seen in the table, the full instrument achieved high reliability (.947). Not only does this indicate that study participants were consistent in their pattern of responses, but also that the instrument has strong potential for future research application and replication. Also of note is that each of the three adapted scales (ATMIO = 20 items; CAMI = 22 items; SSMIS Agreement = 9 items) achieved reliability scores ranging from .857 to .886, again indicating highly consistent patterns of response.
The subscale reliability measures of the ATMIO and CAMI scales, however, fell below .70. The same was true for the nine originally conceived items (except the coefficient for prosecutors). Obtained coefficients less than .70 suggest a need for further multivariate exploratory factor analysis to assess whether the various survey items are in fact measuring the intended constructs (e.g., diminished responsibility, benevolence, etc.) or instead represent other attitudinal dimensions.
Summated Scale Scores
Given the reliability coefficients reported in the previous section, it is reasonable to use the various adapted scales and subscales to calculate summated scores for all participants and each of the three distinct groups. First, however, an important methodological issue must be clarified. Some of the 60 attitudinal survey items were "positively" worded (meaning the phrase had a positive connotation regarding attitudes toward mental illness), while other items were "negatively" worded (meaning the phrase had a negative connotation). In order to calculate summated scale scores, responses for all "negatively" worded items were reverse coded so that they instead represented "positive" statements. This procedure was applied to 29 survey items.
As a consequence, one subscale from the ATMIO scale and two subscales from the CAMI scale needed to be "relabeled/reconsidered": the ATMIO "Negative Stereotypes" subscale and the CAMI "Authoritarianism" and "Social Restrictiveness" subscales. The new labels suggested for each are "Positive Attitudes," "Anti-Authoritarianism," and "Anti-Social Restrictiveness," respectively. Thus, when summed, the labels and scores for each survey respondent on all 60 items will represent the magnitude of positive attitudes regarding mental illness. The higher a participant's summated score on the overall instrument, the adapted scales, and the various subscales, the more positive her/his attitudes regarding mental illness are. Conversely, the lower a participant's scores, the more negative her/his attitudes regarding mental illness are.
Individual summated scores are not presented in order to maintain confidentiality. Rather, the table below presents results in aggregate form for all participants and for each of the three groups.
|Full Instrument (60 items)||3.83 on a scale of 1-5||3.78||3.54||4.02|
|Adapted ATMIO scale (20 items)||3.82 on a scale of 1-5||3.68||3.51||4.04|
| Positive Attitudes subscale
|4.10 on a scale of 1-5||3.88||3.80||4.33|
| Community Risk subscale
|3.44 on a scale of 1-5||3.30||3.14||3.64|
subscale (5 items)
|3.98 on a scale of 1-5||3.94||3.73||4.11|
| Diminished Responsibility
subscale (3 items)
|3.36 on a scale of 1-5||3.27||2.79||3.66|
|Adapted CAMI scale (22 items)||3.85 on a scale of 1-5||3.83||3.57||3.99|
| Anti-Authoritarianism subscale
|3.67 on a scale of 1-5||3.63||3.46||3.79|
|Benevolence subscale (9 items)||3.95 on a scale of 1-5||3.94||3.62||4.10|
| Community MH Ideology
subscale (2 items)
|3.91 on a scale of 1-5||3.94||3.64||4.02|
| Anti-Social Restrictiveness
subscale (4 items)
|3.85 on a scale of 1-5||3.82||3.55||4.01|
|Adapted SSMIS Agreement
subscale (9 items)
|3.81 on a scale of 1-5||3.82||3.53||3.93|
|Originally Conceived Items
|3.87 on a scale of 1-5||3.83||3.60||4.02|
Note. Note. ATMIO = Attitudes Toward Mentally Ill Offenders; CAMI = Community Attitudes Toward the
Mentally Ill; MH = Mental Health; SSMIS = Self Stigma Mental Illness Scale.
The value in each table cell represents the average score for participants using a standardized continuum of 1-5, where 1 represents negative views toward defendants with mental illness and 5 represents positive views. Thus, the standardized values within each group/column can be directly compared to others to determine which group is more relatively positive or negative on all scales and subscales.
In every instance, public defenders manifested the most positive average scale scores, followed by judges. Prosecutors consistently manifested the least positive average scale scores of the three groups across all scales and subscales.
While public defenders as a group manifest the most positive attitudes and prosecutors the least positive, this finding should not be construed to mean that members of the former group "coddle" or those in the latter group "vilify" the mentally ill. In fact, on the Likert scale of 1-5, the highest summated subscale score for public defenders was 4.33 on the dimension of "positive stereotypes." For prosecutors, the average summated score on that same subscale was 3.8, which is still relatively positive overall. The lowest summated subscale score for prosecutors was 2.79 on the dimension of "diminished responsibility." For public defenders, the average summated score on the same subscale was 3.66.
On balance, then, the overall assessment to be drawn from the scores is that judges, prosecutors, and public defenders manifest neutral to positive (but not overly or exceedingly positive) beliefs, perceptions, and attitudes toward defendants suffering from mental illness. This is indeed encouraging, particularly in light of the sometimes common perception that the South (generally) and Mississippi (specifically) may perhaps not be as progressive or therapeutic as other regions and states.
One related issue is whether the observed differences in scores between the three groups of participants presented in the table above are statistically significant (i.e., "real") or, instead, due to chance. Our analysis found that in all instances the observed differences are statistically significant.
Discussion and Conclusion
The results presented in this article were selected from our larger study results for several reasons. First, the reliability of the adapted scales and subscales suggests that the study may be replicated by applying the instrument to another sample of judges, prosecutors, and public defenders. This may help us better understand the beliefs, perceptions, and attitudes of these individuals who play a critical role in adjudicating cases involving mentally ill defendants. Second, the summated scale scores show that although statistically significant differences exist across the three groups of participants, the participants' beliefs, perceptions, and attitudes are not as negative as may be expected. Thus, future multivariate research (e.g., discriminant analysis) might examine and identify independent variables useful in accounting for variance between groups.
While a more complete discussion and conclusion are found in the full report, it should be noted here that this exploratory study provides important information that may be used as a basis for future studies. Specifically, future studies should examine how mental illness as both a social and medical issue is being addressed by a criminal justice system that was neither designed nor adequately equipped for such purposes. With a better understanding of these issues, policymakers and practitioners can perhaps work more effectively within the constraints of an increasing problem that shows no signs of amelioration in the future.