Abstract
People with substance use disorders often differ in their decision-making styles. The present study addressed the impact of two decision-making styles (rational and dependent) on outcomes from a StaySafe tablet computer app intervention designed to improve decision-making around health risk behaviors and previously found to be effective for justice-involved people receiving treatment for a substance use disorder and under community supervision.
Participants were justice-involved residents in residential treatment. After completing a baseline survey, participants were randomly assigned to either complete the StaySafe app or to a standard procedure condition; and then asked to complete a post-intervention survey three months after baseline (this protocol has been registered with clinicaltrials.gov NCT02777086): 348 participants completed a baseline survey and 238 completed the post-test survey. Outcomes included measures of confidence and motivation around HIV knowledge and risks and getting tested. Multilevel analyses addressed the hypothesis that outcomes were related to decision-making style. Multiple imputation (MI) was used to address the effects of missing data.
StaySafe was more effective for those in the lower half of the decision-making dependent scale for HIV risks (HIV-Knowledge, Hepatitis testing, HIV Services testing, and Sex Risk, as well as motivation for treatment. The decision-making rational scale was less consistently related to HIV risk.
The present study showed individuals with substance use disorders who differed in their decision-making styles reacted differently to the StaySafe intervention. Two scales, rational decision making, and dependent decision making are relevant to consider with respect to interventions targeting improving decision making among drug users.
Keywords: Decision making, Technology, Health risk, Community Supervision, Substance use
Introduction
Assessing the efficacy of an intervention has traditionally focused on statistically significant differences between intervention and control conditions in randomized controlled trials. In addition to controlling for baseline measures of outcomes and/or use of covariates that might affect the statistical test being used, cofactors also need to be considered that are pertinent for understanding for whom the intervention is relevant.
Decision-making styles are an important cofactor to consider in determining the relevancy of a given intervention for specific individuals. Decision-making styles have been described as “the habitual decision-making pattern” (Driver, 1979) used by individuals. Harren (1979) referred to decision-making styles as the characteristic mode of perceiving and responding to decision-making tasks. As noted by Scott & Bruce (1995), many researchers consider it as the approach the individual might use in making sense of the information gathered when making a decision. For example, they note that McKenney & Keen (1974) mentioned “habit and strategic modes of thinking” in tasks of organizing and processing information.
Harren (1979) proposed a 3-factor model of career decision-making styles: dependent, rational, and intuitive. Dependent decision-makers project responsibility for decisions upon others. Rational decision-makers take personal responsibility for decisions and use a deliberate and logical approach. Intuitive decision-makers use internal hunches that decisions are basically correct and thus make their decisions quickly. Scott & Bruce (1995) developed scales based on Harren’s three factors measuring: (1) rational decision-making style, (2) intuitive decision-making style, (3) dependent decision-making style, and also added a fourth scale: (4) avoidant decision-making style, which is characterized by attempts to get around having to make decisions. Scott & Bruce (1995) developed their instrument on a large sample of military officers but validated the scale structures on samples of MBA students, upper-level undergraduate business students, and engineers and technicians from a research and development facility. Others have used the scales with students (Gambetti et al., 2008), adolescents (Baiocco et al., 2009), and administrative officers and investigators (Allwood & Salo, 2012).
The relationship between decision-making styles and treatment responsivity has received considerable investigation (e.g., Bavolar & Bacikova-Sleskova, 2020; Curşeu & Schruijer, 2012; Douma et al., 2020; Naveed Riaz et al., 2012). For example, Bavolar & Bacikova-Sleskova (2020) investigated DM styles and mental health, showing that dependent and rational helped define different clusters of individuals and that the rational cluster had the lowest level of psychological distress. Curşeu & Schruijer (2012) found that the rational style predicted rationality in decision-making. Douma et al. (2020) showed that individuals who scored higher in dependent or spontaneous decision-making styles were more likely to have participated in colorectal cancer screening, while those who scored higher on an avoidant decision-making style were more likely not to have participated in the screening. Naveed Riaz et al. (2012), in their study of personality types, examined five decision-making styles. Of particular interest was their finding that of the five personality types examined, “agreeableness” positively predicted dependent decision-making style, and “conscientiousness” positively predicted rational decision-making style.
It has been noted in the literature that people with substance use disorders often differ in their decision-making styles from people who do not have substance use disorders, and these differences can impact how people react to interventions. Shaghaghy et al. (2011) compared men with and without substance use disorders on decision-making and attributional styles. Findings showed that men with substance use disorders were lower in rational decision making and, among men with substance use disorders, higher levels of rational decision making were associated with lower levels of optimistic attributional style, internal attribution and stability of positive events, and higher levels of learned helplessness. Chen et al. (2020) conducted a meta-analysis of risky decision making among individuals with substance use disorders and found that substance use was associated with impaired risky decision making.
While several decision-making styles have been identified (see Scott & Bruce, 1995), two styles that have received perhaps the most attention are the rational and dependent decision-making styles. In their scale construction and construct validity, Scott and Bruce (1995) found support for their rational decision making scale with respect to control orientation and also support that internally controlled individuals were less likely to be dependent on others in making their decisions. As shown by Bayolar and Bacikova-Sleskova (2020), these scales helped define different clusters of individuals in their research on mental health and psychological distress. In their evaluation of the StaySafe app intervention, Lehman and his colleagues (2021) also included both Scott and Bruce’s (1995) rational and dependent decision-making scales in their study.
Due to the differences found for these two decision-making styles in the literature, the current study examines the impact of decision-making styles as a critical -moderator when assessing the receptivity of a tablet-based, self-administered intervention designed for justice-involved people on probation who were identified as having a substance use disorder. The intervention, StaySafe, utilizes an analytically-created schema (ACS) to train people to make better decisions around health risk behaviors, especially HIV risks. Rational and dependent decision-making styles (Rational DM Style and Dependent DM Style), may be of particular relevance for the StaySafe intervention. Rational DM Style involves a “thorough search for and logical evaluation of alternatives,” while Dependent DM Style involves “a search for advice and direction from others.” That is, with regard to interventions aimed at improving decision-making on risky behaviors related to infectious diseases such as HIV, STIs, and hepatitis, decision-making styles should be regarded as a potentially important moderator.
For the present study, we re-examine the StaySafe intervention, which has been shown to be helpful in making decisions by improving critical thinking. StaySafe uses an analytical model of decision-making called WORK-IT: emphasizing what the problem is, who is affected and who can help with the problem, what are your options, rating the options, knowing what to do, imagining carrying out the steps, and then testing those steps. StaySafe has been shown to be effective (Lehman et al., 2021) in helping clients achieve greater knowledge, confidence, and motivation around proximal outcomes of HIV knowledge, avoiding sex risks, seeking HIV services, and risk reduction skills when compared with a control group.
As this app used a schema that promotes critical thinking about health risks associated with substance use and unprotected sex, it was hypothesized that decision-making style would be a factor in the effectiveness of the StaySafe app. In this regard, the present study examines the effectiveness of StaySafe with respect to those who measured high and low on the Rational Decision Making and Dependent Decision-Making scales developed by Scott and Bruce (1995). The present study examines these styles as potential moderators for those who used the StaySafe tablet computer app. Because of the emphasis on critical thinking, it might be expected that StaySafe would be less effective for those who are more dependent upon others when making decisions (i.e., Dependent DM Style).
Materials and Methods
StaySafe intervention
StaySafe is a self-administered tablet app consisting of 12 brief (each less than 10 min) weekly sessions (Lehman et al., 2021). Nine of the 12 sessions involve the WORK-IT ACS (analytically-created schemas) and three Participant Choice sessions include several information-based activities around HIV and health risks. The ACS is an approach for organizing information through a series of steps and exercises for developing analytic thinking. WORK-IT is a specific ACS that employs a simplified structure for analyzing problems, weighing and rating options for making a decision, followed by planning how to carry out that decision—that is, What is the problem; Who is affected by the problem; Who can help with the problem; Options for dealing with the problem; Rating the options; Knowing what to do based on the ratings; Imagining steps to carry out the decision; and then Testing those steps. Several studies using WORK-IT have shown improvement in decision-making, self-awareness, problem recognition, and indicators of treatment engagement (Becan et al., 2015; Knight et al., 2015, 2016).
In using StaySafe, participants view a brief video showing actors working through a risky situation (vicarious learning) and then proceed to the WORK-IT steps. The WORK-IT ACS starts with choosing from a list of 11 problem themes relevant to the target population based on input from probation officers, substance use treatment counselors, clients, and recommendations on reducing health risks from the National Institutes of Health and the Center for Disease Control and Prevention (e.g., “Asking a partner about his or her HIV testing,” “Practicing safe sex”). In using the app, the individual works through the rest of the WORK-IT steps, sometimes making explicit choices including options for dealing with the chosen theme, and other times thinking about choices. This includes considering a list of possible choices to questions such as “Who is affected by the decision?” or “Who can help with the problem?” For the step, “Options for dealing with the problem,” the participant is presented with a set of four options related to the chosen topic to consider. For example, for the problem theme options might be, “It’s hard to ask a partner about his or her HIV testing,” “Don’t ask, but always use condoms,” and “Have unprotected sex just this one time.” Health facts related to each option “pop up” on the screen, thus associating relevant information within the decision-making schema. Participants in some WORK-IT steps are asked to think about choices without making an explicit response. The WORK-IT sessions end with a “maze” game in which an animated character moves around a maze as the participant answers quiz questions designed to reinforce information from the session.
This mental practice is intended to help the participant learn to use WORK-IT steps in the “real world” and therefore can be an effective learning approach (Cooper et al., 2001). The goal of WORK-IT is to have participants internalize the schema through repeated practice using relevant examples across multiple StaySafe sessions. That is, the goal of the StaySafe intervention is to assist the participant in making their own decisions in reducing health-risky behaviors. Further information about StaySafe and the study protocol is described in Lehman et al. (2018).
Procedures
Study participants were required to be at least 18 years old, have six months of probation supervision remaining so they could complete the study, not be a sex offender or convicted of a violent crime, not have a mental illness that could prevent study participation, not have charges pending that could lead to incarceration in the next six months, and be able to read English at the fifth-grade level or higher. Sex offenders or those convicted of a violent crime were not included as they were not eligible for the programs utilized for recruiting. Participants were recruited at three residential community supervision and corrections (probation) facilities in two large counties in Texas. Participants were also recruited from community-based facilities in two counties. Analyses were run separately for the two samples although only the results for the residential samples are discussed here as the community-based sample was small with high attrition rates. We note below how the results of the two samples were similar. All study procedures were approved by the TCU IRB. Participant recruiting conducted by TCU researchers included brief presentations to groups of new admissions during their first week at the residential facility.
Interested participants were asked to complete TCU IRB-approved informed consent and brief demographic forms and baseline surveys the following week. Randomization into study conditions (i.e., StaySafe and standard practice [SP] arms) followed the completion of the baseline survey. Participants in both arms were asked to complete a post-intervention survey three months after the baseline.
Compensation for study participation was provided by payments toward probation fees -- $20 was paid toward probation fees for the baseline survey and $20 for the post-intervention. Participants in the StaySafe arm were also compensated $10 for each completed StaySafe session.
Sample
The current study examined the residential sample used by Lehman and associates (2021); 370 were consented with 348 providing baseline surveys. Of these, 179 were in StaySafe and 169 in Standard Practice. Post interventions surveys were obtained for 125 in StaySafe and 113 in Standard Practice, giving a total of 238 with follow-up surveys.
Specifically, there were no post-intervention surveys for 31.6% of the residential sample, with 30.2% (n = 54) missing in the StaySafe sample and 33.1% (n = 56) in the Standard Practice sample. Lehman et al. (2021) provides a consort diagram and additional detail regarding the sample. Because the outcome study of the effectiveness of the StaySafe intervention reported by Lehman et (2021) did not address the relationship between decision-making styles and the outcomes, the present study is new information based on the analyses of data that was previously collected.
Data Collection
Data was obtained from self-report surveys and output from the StaySafe app. Scheduling for surveys and StaySafe sessions was coordinated with program staff and typically conducted during times of the day when program activities were not taking place.
StaySafe sessions
The 12 StaySafe sessions were typically administered weekly for 12 weeks, although two sessions were sometimes administered when a session was missed. A research assistant provided an Android tablet and headphones for the participants to complete the sessions. Participant data were linked by a study identification number only; identifiable information was not recorded on the data sources. Other than completing the brief, weekly StaySafe sessions, participants followed the normal program schedules.
Standard practice
Participants randomly assigned to the SP arm went about their normal daily activities such as attending probation meetings or groups. They were requested to complete only the baseline and post-intervention study surveys and did not participate in the StaySafe sessions.
Data collection and measures
Study measures were collected via paper and pencil surveys at baseline before random assignment and 3 months after baseline (post-intervention). Outcomes included four TCU Knowledge, Confidence & Motivation (KCM) scales (Lehman et al., 2015), three scales assessing decision-making styles (Rational Decision Making, Dependent Decision Making; Scott & Bruce, 1995), and the TCU decision-making scale (Institute of Behavioral Research, 2007). Other outcomes included scales from the TCU CEST and the frequency of infectious disease tests. Demographic and background measures included age, gender, race/ethnicity, education, marital status, and in the last 6 months before entering their current program – employment, public assistance, arrest status, on parole or probation, in jail or prison, and treatment for mental health, alcohol use, substance use, or treatment in an emergency room (from the TCU Adult Risk Assessment [TCU A-RSK] form; Institute of Behavioral Research, 2008).
Decision-Making (DM) scales
Scott and Bruce (1995) in their review of the decision-making styles identified four basic types (rational decision making which emphasized logical evaluation; dependent decision making in which the advice and direction of others are relied upon heavily; intuitive decision making where hunches and feelings are relied upon; avoidant decision making whereby attempts are made to avoid making decisions). Of these four styles, two were more relevant to the StaySafe intervention, which emphasized the development of analytic thinking in the making of decisions to reduce risky health choices. The first is rational decision making (which would be related to analytic thinking) and the second is dependent decision making (which would be a style almost opposite to analytic thinking). The intuitive decision-making scale and the avoidant decision-making scales were not used in the original StaySafe evaluation as they were deemed less relevant to the WORK-IT ACS process that was emphasized. Two decision-making style scales were adapted from Scott & Bruce (1995). Dependent Decision-Making style (α = .79) consisted of 5 items and emphasized the importance of using input from others in making decisions. This scale is relevant for the “W” component of WORK-IT (who will be affected and who can help with a problem). The Dependent Decision-Making style items included:
“I often need the assistance of other people when making important decisions.”
“I rarely make important decisions without consulting other people.”
“If I have the support of others, it is easier for me to make important decisions.”
“I use the advice of other people in making important decisions.”
“I like to have someone to steer me in the right direction when faced with important decisions.”
The Rational Decision-Making style (α = .84) consisted of 5 items emphasizing logical and systematic decision making and exploring all options before making a decision. This scale is relevant for the “O” in the WORK-IT schema and involves exploring options for a problem. Its items included:
“I double-check my information sources to be sure I have the right facts before making a decision.”
“I make decisions in a logical and systematic way.”
“My decision-making requires careful thought.”
“When making a decision, I consider various options in terms of a specific goal.”
“I explore all of my options before making a decision.”
Other scales used in the evaluation include the TCU Knowledge, Confidence, and Motivation scales (KCM). Four scales included HIV Knowledge Confidence, Avoiding Risky Sex, HIV Services & Testing, and Risk Reduction Skills. Each of the scales, except for HIV Services & Testing, included items that assessed how knowledgeable (K) the participant felt about the topic, how confident (C) they were about their knowledge, and how motivated (M) they were to act on the knowledge. HIV Services & Testing only included knowledge and motivation items (Lehman et al., 2015).
The TCU HIV Knowledge Confidence scale has 13 items (α = .93), such as, “You know enough to teach others what they should do if they think they have been exposed to HIV” (K). The TCU Avoiding Risky Sex scale has 13 items (α = .93) and includes items such as, “During the past month, you have learned about what situations might lead you to make a poor decision about risky sex” (K).
The TCU HIV Services & Testing scale consists of 7 items (α = .81), such as, “During the past month, you have become more knowledgeable about how to get HIV services in the real world” (K).
The TCU Risk Reduction Skills scale is comprised of 14 items (α = .91) and includes items such as, “During the past month, you have a better understanding of how your shoulds and wants can conflict in the real world” (K).
TCU Client Evaluation of Self-in Treatment Scales (CEST)
The TCU CEST (Joe et al., 2002; Simpson et al., 2012) is a self-administered instrument that includes scales organized into three domains – Treatment Motivation, Psychological Functioning, and Social Functioning. Four scales from the Treatment Motivation domain included Problem Recognition (9 items, α = 94), Desire for Help (6 items, α = .84), Treatment Participation (7 items, α = .78), and Treatment Needs (5 items, α = .80). Five scales from the Psychological Functioning domain included Self-Esteem (6 items, α = .77), Depression (6 items, α = .83), Anxiety (7 items, α = .87), Decision Making (9 items, α = .79) and Expectancy (4 items, α = .78). The Decision-Making scale from the CEST is of particular interest as it also emphasized rationality in making decisions. Three scales from the Social Functioning domain included Hostility (8 items, α = .86), Risk Taking (7 items, α = .78), and Social Support (9 items, α = .84). All items used a 5-point Likert-type Disagree Strongly to Agree Strongly scale. Scale scores were the average of item responses multiplied by 10 for a range of 10 to 50.
Analytical approach
Three approaches were employed in evaluating the relationship of decision-making styles in the re-evaluation of the StaySafe app. These were using decision making as a moderator in the evaluation, using multiple imputation in the re-evaluation of StaySafe where decision-making style is a moderator, and partial correlation where decision-making style is used as a continuous variable in re-evaluating its effect on StaySafe outcomes.
Multilevel modeling
The initial analysis addressing the hypothesis that follow-up outcome was related to decision-making style was done by a multilevel model in which facility, intervention (StaySafe vs. SP), and decision-making style dichotomy (either DM-rational or DM-dependent) were used as classification variables. Each follow-up outcome measure was modeled in terms of intervention, the dichotomy of one of the decision-making style variables, the interaction of intervention with the decision-making style dichotomy, and the corresponding baseline measure of the outcome measure. The multilevel model specified a random intercept and an unstructured covariance matrix for the facility. The analyses were performed using SAS Proc Mixed.
Multiple imputation
A large amount of missing data represented a validity concern for the results of the study. To address this, a series of 100 multiple imputations using SAS Proc MI were done for each of the outcomes found to be significant at the p < .10 level for intervention or intervention within a decision-making style dichotomy. In the estimation process, the fully conditional specification (FCS) method was designated with the discriminant function method used for the imputation method. Additionally, the pattern-mixture model approach, assuming the missing data are missing not at random (MNAR), was specified for imputing the missing values. There was support for the data being missing not at random as stepwise logistic regressions predicting whether a follow-up occurred differed for StaySafe and SP. For StaySafe, those more likely to have a follow-up were married (OR=3.0) and had treatment for an alcohol problem (OR=10.5); whereas, those in the SP condition were more likely to have more education (OR=1.5), be on probation/parole (OR=2.8), and less likely to have been in jail or prison (.34).
Possible site differences were examined through the use of a nested model (SAS Proc Mixed). The summaries of the multiple imputations by outcome were performed by using SAS Proc Mianalyze.
Partial correlations
Partial correlations were computed between the intervention (StaySafe vs. SP) and each follow-up measure with the decision-making styles partialed out to assess the magnitude of the effect of decision-making style on the correlation. A series of 100 multiple imputations were performed for computing the partial correlations. For each partial correlation, the corresponding mean of the Fisher Z transformation over 100 MI was calculated and a ratio of the mean Fisher Z to its standard deviation (SD) was computed to compare with the mean Fisher Z in which the decision-making scale was not used as a covariate. The larger change in the magnitude of the Fisher Z/SD was used as an indicator of which decision-making scale (DM-rational or DM-dependent) had the larger effect when partialed out.
Results
Sample
Among the 348 participants who completed a baseline survey, 49% were male and 51% female; 61% were African American, 29% white, and 33% Hispanic. Just over two-thirds were between the ages of 21 and 39 (69%), 62% had a high school diploma (24% with post-H.S. education), and 56% were single. Prior to entering their present program, 49% had full-time employment, 37% were not employed nor looking for employment, 26% received public assistance, and 64% or more had been arrested, on parole or probation, and/or in jail or prison. Treatment for mental health, illegal substance use, or in an emergency room was reported by at least 28%, and 12% for alcohol use.
Comparisons at baseline between StaySafe and SP participants showed that the StaySafe arm had a higher percentage of participants who had spent time in jail or prison prior to entering the current program (71%) than did the SP arm (59%), however, this background difference did not correlate significantly with whether a post-intervention survey was completed. A further inspection of each of the three residential sites found that in two of the sites, there was no significant difference between the percentage having spent time in jail or prison before entering the current program. In the third site (the largest site), however, there was a significant difference between having spent time in jail or prison, with 74% for the StaySafe arm and 57% for the SP arm in this facility. In terms of attrition (comparing participants who completed and did not complete a post-survey), those not completing a post-survey had lower percentages of full-time employment in the previous 6 months (40% vs 53%) or had been on probation or parole (52% vs 69%). They also scored lower at baseline on many of the outcome measures or subscales (HIV Knowledge Confidence, the Knowledge subscale for Avoiding Risky Sex, HIV Services & Testing, Risk Reduction Skills, as well as DM-rational and the TCU Decision Making scale).
Background relationships with Decision-Making Style
The means and standard deviations for background measures are presented in Table 1. Also, in this table are the correlations of each measure with dichotomized (based on median splits) DM-rational and DM-dependent scales.
Table 1.
Correlations of Background with DM-rational and DM-dependent dichotomies (N=348)
| Correlation | ||||
|---|---|---|---|---|
| Mean | SD | DM-rational | DM-dependent | |
|
| ||||
| Decision-making Styles | ||||
| DM-rational | 1.51 | 0.50 | 0.087 | |
| DM-dependent | 1.46 | 0.50 | 0.087 | |
| Demographic/Background | ||||
| Age | 33.18 | 9.88 | −0.039 | −0.069 |
| Gender (male) | 1.52 | 0.50 | 0.023 | −0.027 |
| Hispanic | 0.32 | 0.47 | −0.033 | −0.032 |
| Race | ||||
| Black | 0.28 | 0.45 | 0.109 * | −0.043 |
| White | 0.62 | 0.49 | −0.135 * | 0.026 |
| Other | 0.10 | 0.30 | 0.055 | 0.024 |
| Education | 3.69 | 1.04 | 0.113 * | −0.046 |
| Marital status | ||||
| Single | 0.56 | 0.50 | 0.098 | 0.009 |
| Married | 0.21 | 0.40 | −0.013 | −0.055 |
| Widowed/Sep/Divorced | 0.24 | 0.43 | −0.102 | 0.042 |
| Children (any) | 1.61 | 1.43 | −0.023 | 0.034 |
| Last six months prior to incarceration | ||||
| Employed full-time | 0.49 | 0.50 | 0.050 | −0.005 |
| Unemployed | 0.37 | 0.48 | −0.031 | 0.013 |
| Public Aid | 0.26 | 0.44 | −0.002 | −0.020 |
| Locked-up | 1.61 | 1.43 | −0.023 | 0.034 |
| Probation/parole | 0.64 | 0.48 | −0.025 | 0.069 |
| ER visit | 0.31 | 0.47 | 0.042 | −0.033 |
| Mental Health Problem | 0.28 | 0.45 | 0.034 | 0.031 |
| Alcohol treatment | 0.12 | 0.32 | −0.023 | −0.006 |
| Drug Treatment | 0.35 | 0.48 | −0.037 | 0.079 |
| Arrested | 0.70 | 0.46 | −0.098 | 0.051 |
| Jail/prison | 0.66 | 0.48 | −0.032 | −0.018 |
| TCU Scales | ||||
| Treatment Motivation | ||||
| Problem Recognition | 3.57 | 1.15 | −0.006 | 0.141 ** |
| Desire for Help | 3.92 | 0.91 | 0.054 | 0.107 * |
| Trt Participation | 3.43 | 0.83 | 0.063 | 0.158 * |
| Treatment Needs | 3.43 | 0.97 | −0.020 | 0.130 * |
| Psychological Functioning | ||||
| Self Esteem | 3.46 | 0.84 | 0.289 *** | −0.103 |
| Depression | 2.60 | 0.93 | −0.239 *** | 0.053 |
| Anxiety | 3.11 | 1.01 | −0.090 | 0.115 * |
| Decision Making | 3.71 | 0.63 | 0.471 *** | −0.014 |
| Expectation of no drug use | 4.06 | 0.88 | 0.222 *** | −0.139 * |
| Social Support | ||||
| Hostility | 2.55 | 0.90 | −0.092 | 0.072 |
| Risk Taking | 3.04 | 0.77 | −0.097 | 0.004 |
| Social Support | 4.11 | 0.64 | 0.338 *** | 0.057 |
p < .05
p < .01
p < .001
The table shows that few demographic and background measures were significantly correlated with the dichotomized DM-rational and DM-dependent scales. Blacks and those with more education were more likely to have higher DM-rational scores.
The TCU measures of Motivation for Treatment, Psychological Functioning, and Social Functioning were found to be more strongly correlated with the decision-making style scales as shown in the lower part of Table 1. Reporting higher motivation was related to being in the upper half of the DM-dependent scale; generally, motivation tended to be unrelated to DM-rational. With respect to the psychological functioning scales, overall, these scales tended to be related to the DM-rational scale. Generally, there was higher decision-making as measured by the TCU Decision Making scale, higher self-esteem, and expectation for no substance use for those who scored in the upper half of the DM-rational scale. There was a tendency for this group to report lower depression. It was also seen that the DM-dependent scale was related positively to the anxiety scale and negatively to an expectation for no drug use; that is, for the latter there would be an expectation for substance use for those who scored higher on DM-dependent.
Hypothesis of DM-styles as moderators for StaySafe: follow-up outcomes
As the first stage in addressing DM-rational and DM-dependent as possible moderators in the interpretation of the effectiveness of the StaySafe intervention, each of the two factors was considered separately in terms of their possibility as cofactors of StaySafe. The results of the two sets of the multilevel analysis of variance in which each outcome is modeled in terms of the corresponding baseline measure, the intervention, and one of the baseline decision-making style scales (DM-rational or DM-dependent) dichotomized close to their medians are presented in Table 2. Overall, the two decision-making scales overlapped: 55% of those in the high DM-dependent group also reported high DM-rational versus 47% of those in the low DM dependency group (χ2 = 2.65, p = .103). Approximately 53% of those in the low DM-dependent group reported high DM-rational. The sample available for the follow-up analysis had 238 individuals.
Table 2.
Summary of Significant Post-Intervention Outcomes by DM-rational and DM-dependency Styles and Corresponding Multiple Imputations: Residential sites
| DM-rational | DM-dependent | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
|
|
|
|||||||||||
| Lower (n = 110) | Higher (n = 128) | Total (N = 238) | Lower (n = 124) | Higher (n = 114) | Total (N = 238) | |||||||
| Follow-up Outcomes | F-test | p | F-test | p | F-test | p | F-test | p | F-test | p | F-test | p |
|
| ||||||||||||
| Original Analysis | ||||||||||||
|
|
||||||||||||
| KCM Scales | ||||||||||||
| HIV Knowledge Confidence | 6.26 | 0.013 | 7.47 | 0.007 | 13.68 | 0.001 | 12.99 | 0.001 | 2.78 | 0.097 | 13.76 | 0.000 |
| Avoiding Risky Sex | 6.75 | 0.010 | 0.72 | 0.397 | 6.15 | 0.014 | 5.11 | 0.025 | 1.22 | 0.271 | 5.63 | 0.018 |
| Risk Reduction Skills | 3.84 | 0.051 | 3.65 | 0.057 | 7.50 | 0.007 | 7.74 | 0.006 | 1.20 | 0.274 | 7.45 | 0.007 |
| HIV Services & Testing | 6.17 | 0.014 | 9.13 | 0.003 | 15.05 | 0.000 | 14.64 | 0.000 | 3.02 | 0.084 | 15.30 | 0.000 |
| Treatment Motivation | ||||||||||||
| Desire for Help | 0.10 | 0.758 | 1.25 | 0.265 | 0.97 | 0.325 | 4.77 | 0.030 | 0.53 | 0.466 | 0.99 | 0.321 |
| Treatment Needs | 1.49 | 0.224 | 0.74 | 0.391 | 2.18 | 0.141 | 4.02 | 0.046 | 0.00 | 0.967 | 2.08 | 0.150 |
| Psychological Functioning | ||||||||||||
| Depression | 1.09 | 0.297 | 1.80 | 0.181 | 2.82 | 0.094 | 1.39 | 0.240 | 1.84 | 0.176 | 3.25 | 0.073 |
| Anxiety | 1.23 | 0.268 | 8.06 | 0.005 | 1.26 | 0.263 | 1.86 | 0.174 | 0.31 | 0.578 | 1.87 | 0.173 |
| Social Functioning | ||||||||||||
| Risk Taking | 4.82 | 0.029 | 0.52 | 0.473 | 4.40 | 0.037 | 2.41 | 0.122 | 1.77 | 0.185 | 4.15 | 0.043 |
| Social Support | 0.03 | 0.864 | 2.30 | 0.131 | 1.35 | 0.247 | 3.34 | 0.069 | 0.00 | 0.963 | 1.69 | 0.194 |
| KCM Scales | ||||||||||||
| HIV Knowledge Confidence | 6.45 | 0.011 | 7.34 | 0.007 | 12.11 | 0.000 | 10.63 | 0.001 | 3.46 | 0.063 | 11.90 | 0.001 |
| Avoiding Risky Sex | 5.38 | 0.021 | 0.96 | 0.326 | 5.48 | 0.019 | 4.08 | 0.044 | 1.46 | 0.228 | 4.84 | 0.028 |
| Risk Reduction Skills | 4.41 | 0.036 | 3.65 | 0.056 | 6.10 | 0.014 | 6.10 | 0.014 | 1.66 | 0.198 | 5.91 | 0.015 |
| HIV Services & Testing | 7.18 | 0.008 | 7.84 | 0.005 | 13.62 | 0.000 | 11.63 | 0.001 | 4.08 | 0.043 | 13.69 | 0.000 |
| Treatment Motivation | ||||||||||||
| Desire for Help | 0.20 | 0.651 | 0.85 | 0.360 | 0.88 | 0.346 | 2.89 | 0.090 | 0.10 | 0.756 | 0.88 | 0.348 |
| Treatment Needs | 1.59 | 0.210 | 0.61 | 0.436 | 1.96 | 0.163 | 2.72 | 0.100 | 0.07 | 0.784 | 1.85 | 0.174 |
| Psychological Functioning | ||||||||||||
| Depression | 1.46 | 0.228 | 2.19 | 0.138 | 2.99 | 0.084 | 2.02 | 0.155 | 1.64 | 0.201 | 3.53 | 0.061 |
| Anxiety | 0.14 | 0.701 | 5.95 | 0.015 | 1.44 | 0.229 | 1.99 | 0.160 | 0.67 | 0.390 | 2.16 | 0.141 |
| Social Functioning | ||||||||||||
| Risk Taking | 4.16 | 0.042 | 0.94 | 0.332 | 4.58 | 0.033 | 2.99 | 0.084 | 1.61 | 0.204 | 4.41 | 0.036 |
| Social Support | 0.17 | 0.685 | 2.10 | 0.147 | 1.14 | 0.287 | 2.19 | 0.140 | 0.12 | 0.733 | 1.39 | 0.239 |
Note: Bolded entries p < .05
The overall intervention tests identified significant intervention differences in HIV Knowledge Confidence, Avoiding Risky Sex, Risk Reduction Skills, and HIV Services & Testing scales and one scale in the Social Functioning domain (Risk Taking). With regard to the hypothesis of DM styles as cofactors, the cofactor results showed some significant differences in every outcome domain. It was seen that those in the lower halves of both of the decision-making styles (particularly the DM-dependent style) had significant intervention effects for the KCM scales and the motivation scales.
For the HIV Knowledge Confidence scale, the pattern suggested that those in the lower half of the DM-dependent group had significant intervention effects when interpreting the overall significance of the outcomes targeted by the StaySafe intervention (i.e., HIV Knowledge Confidence, Avoiding Risky Sex, Risk Reduction Skills, and HIV Services & Testing). In contrast, for DM-rational, the results were similar, but not as pronounced when compared with DM-dependent. In addition, intervention effects were found for those in the lower half of the DM-rational scale for the Risk Taking outcome.
Those in the lower part of DM-rational tended to have significant intervention effects for HIV Knowledge Confidence and Avoiding Risky Sex. Those in the lower half of DM-dependent fared better in StaySafe than in SP, and those in the upper half of DM-rational did better in StaySafe than in SP on HIV Knowledge Confidence. A similar result was found for the outcome of HIV Services & Testing, though in each the lower half of DM-rational also had a significant intervention effect, though not as strong as those in the upper half.
For motivation outcomes, those in the lower half of DM-dependent had significant intervention effects on Desire for Help and Treatment Needs, even though the test on the total sample was not significant. For Psychological Functioning, decision-making style proved to be a useful consideration. There was a significant intervention effect for those in the upper half of the DM-rational scale for the Anxiety scale, even though the test on the overall sample was not significant. For Social Functioning, the test on the overall sample for Risk Taking was significant, and the sub-analysis by DM-rational showed that the significance lay largely in those in the lower half of that scale’s distribution. The tests for the overall sample were not significant for the Infectious Disease tests.
Multiple imputations
Because of the relatively large percentage of missing follow-up data, multiple imputation analyses were conducted on the outcomes that were found to be significant in the analyses presented in Table 2. A series of 100 multiple imputations were done for each of the 12 outcomes for which either the overall test or a subtest based on the DM-median split analyses was significant at p < .10. For these multiple imputations, a nested model was used to address possible site differences. The summaries of the multiple imputations by outcome are presented in the lower half of Table 2.
There was statistical support in the MI analyses for the significant outcomes found in the original analyses. All of the intervention effects that were significant in the original analyses were also found to be significant in the MI analyses. These included HIV Knowledge Confidence, Avoiding Risky Sex, Risk Reduction Skills, HIV Services & Testing, and Risk Taking. Also, there was statistical support for the findings found for the partitions by DM-rational and DM-dependent.
Partial correlations of DM-rational and DM-dependent for outcomes with intervention
The previous analyses dealt with the dichotomizations of DM-rational and DM-dependent as moderators in the analyses of the outcomes and the StaySafe intervention. In addition to those analyses, it was deemed useful to address the effects of the DM-rational and DM-dependent variables as continuous variables in addition to their being dichotomized on the relationship between the outcomes and the intervention. Reported here are the results of 100 multiple imputations for these correlations. The outcomes addressed are only those previously addressed in the multiple imputations reported in Table 2. In the MI partial correlations, the average correlation was obtained, as well as the range of the correlations, the average of the Fisher Z transformations of each of the 100 correlations, and the standardized statistic of average Fisher Z divided by the average standard deviation of the Fisher transformations (Fisher Z/SD). It is this statistic that is useful for statistical comparison.
It was found that partialing out DM-rational when compared to DM-dependent had a slightly more consistent effect on reducing the correlations between the outcomes and StaySafe; that is, as shown in Table 3, it was found that 6 of the 10 outcome results favored DM-rational when partialing the original correlations between the outcomes and intervention. This conclusion was based on comparing the Fisher Z/SD for the original correlations with the corresponding Fisher Z/SD in which the DM-rational variable was partialed out to that in which the DM-dependent variable was partialed out.
Table 3.
Correlations from Selecteda Multiple Imputation Analysesb of Intervention with DM-rational and DM-dependent
| 100 MI | DM-rational as Covariate | DM-dependent as Covariate | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
|
|
|
|
||||||||||
| Follow-up Outcomes | Average Corr | Fisher Z /SD | Correlation Range | Average Corr | Fisher Z | Fisher Z/SD | Correlation Range | Average Corr | Fisher Z | Fisher Z/SD | ||
|
| ||||||||||||
| KCM Scales | ||||||||||||
| HIV Knowledge Confidence | −0.227 | −6.49 * | −0.313 | −0.149 | −0.225 | −0.23 | −6.13 * | −0.318 | −0.159 | −0.230 | −0.24 | −6.35 * |
| Avoiding Risky Sex | −0.146 | −4.05 * | −0.248 | −0.069 | −0.149 | −0.15 | −3.98 * | −0.248 | −0.073 | −0.148 | −0.15 | −3.99 * |
| Risk Reduction Skills | −0.159 | −4.04 * | −0.266 | −0.062 | −0.169 | −0.17 | −4.43 * | −0.261 | −0.061 | −0.161 | −0.16 | −4.23 * |
| HIV Services & Testing | −0.242 | −6.37 * | −0.333 | −0.129 | −0.246 | −0.25 | −5.84 * | −0.333 | −0.139 | −0.248 | −0.25 | −6.02 * |
| Treatment Motivation | ||||||||||||
| Desire for Help | −0.069 | −1.91 | −0.170 | 0.020 | −0.065 | −0.07 | −1.71 | −0.178 | 0.007 | −0.074 | −0.07 | −1.98 * |
| Treatment Needs | −0.094 | −2.61 * | −0.202 | −0.013 | −0.099 | −0.10 | −2.64 * | −0.199 | −0.015 | −0.096 | −0.10 | −2.59 * |
| Psychological Functioning | ||||||||||||
| Depression | 0.113 | 3.12 * | −0.009 | 0.195 | 0.105 | 0.11 | 2.84 * | −0.004 | 0.194 | 0.107 | 0.11 | 2.86 * |
| Anxiety | 0.092 | 2.54 * | −0.033 | 0.170 | 0.080 | 0.08 | 2.14 * | −0.022 | 0.176 | 0.090 | 0.09 | 2.41 * |
| Social Support | ||||||||||||
| Risk Taking | 0.131 | 3.59 * | 0.009 | 0.213 | 0.125 | 0.13 | 3.36 * | 0.012 | 0.211 | 0.125 | 0.13 | 3.36 * |
| Social Support | −0.081 | −2.23 * | −0.179 | 0.009 | −0.077 | −0.08 | −2.04 * | −0.186 | −0.001 | −0.082 | −0.08 | −2.22 * |
Selection based on significance from initial analyses of original data
number of imputations=100; StaySafe=1, control=2 (negative r=StaySafe is higher)
p < .05 (bolded)
Residential vs Community samples
Results for the community samples (not reported here were similar to those for the residential samples). For the community sample, StaySafe was more effective for those who scored in the lower half of decision-making dependent. For the residential sample, StaySafe was more effective for those in the lower half of the decision-making dependent scale for HIV-Knowledge, Hepatitis testing, HIV Services testing, and Sex Risk, and was also more effective for those in the lower half of decision-making rational scale for the Sex Risk and for the Risk-Taking outcomes. For HIV Knowledge and HIV Testing Services Testing, StaySafe was more effective than SP regardless of decision-making style.
Discussion
The literature on decision making has suggested that decision-making styles should be considered when addressing the effectiveness of interventions. The present study re-examined the data used in the assessment of the TCU StaySafe app intervention with respect to two of the decision-making styles identified by Scott and Bruce (1995) and which were relevant for the WORK-IT schema used in StaySafe: DM-rational and DM-dependent. The analytical approach considered these two decision-making styles in several residential-based, probation substance use treatment samples used by Lehman and his associates (2021) in the evaluation of StaySafe. The approach for addressing the decision-making styles was first to dichotomize the sample on each of the two decision-making style scales (DM-rational and DM-dependent). For each outcome used in the StaySafe evaluation study (Lehman et al., 2021b), a multilevel analysis was performed in which the factors of the model were intervention (StaySafe vs SP), the dichotomy of one of the decision-making style scales, the interaction of the intervention with the decision-making style scale dichotomy, and the baseline measure of the outcome (used as a covariate). Furthermore, a series of 100 multiple imputation analyses were performed for each of the outcomes using the analytic model for confirmation of the results while addressing the issue of missing data.
Analyses demonstrated that decision-making style had a significant effect on the receptivity of the StaySafe intervention. However, the results were somewhat complex with regard to the two decision-making style scales. The results showed that five outcomes were significantly affected by both types of decision-making styles (HIV Knowledge Confidence, Avoiding Risky Sex, Risk Reduction Skills, HIV Services & Testing, and Risk Taking). The StaySafe intervention was more effective for those in the lower half of the DM-dependent scale for HIV-Knowledge Confidence, Tested for Hepatitis B or C, and HIV Services & Testing. It was also significant for the Avoiding Risky Sex outcome. Some interesting differences occurred in the DM-rational dichotomy. StaySafe was found to be more effective for those in the lower half of the DM-rational scale for Avoiding Risky Sex and Risk-Taking outcomes. However, for the outcomes on HIV Knowledge Confidence and HIV Services & Testing outcomes, StaySafe was more effective than SP regardless of whether they scored in the upper or lower half of the DM-rational scale.
To further address the relationship of decision-making style scales to the outcomes, the two decision-making style scales were also considered as continuous variables and were used in partial correlation analyses. In these partial correlation multiple imputation analyses, it appeared that partialing the DM-rational scale had a more consistent effect in reducing the relationship between intervention and the outcomes.
In the creation of interventions, it is important to realize that individuals vary in their willingness to take personal responsibility for their actions and that not everyone will emphasize rationality in that process and use a deliberate and logical approach. As such, some interventions that target an individual’s decision makingmay need to include multiple paths to improving decision making and consider other factors that impact decisions. For example, emotions have been viewed as a potentially strong, pervasive, and predictable influence on decision making and have been addressed in regard to rational decision making (Lerner, Li, Valdesolo, & Kassam, 2015). The influence of emotion on those who rely on others to help make their decisions may be equally important or more so. For individuals who rely on others to help make decisions, it may be important to assess how that may influence their emotions with regard to that process. Individuals who are more dependent on others in making decisions might be more likely to have complementary, reciprocal, or shared emotions with others (e.g., Keltner & Haidt 1999). As shown in the current study, participants who were on the higher end of DM-dependent did not seem to benefit as well from the StaySafe training as did those low on DM-dependent. Harren (1979) defined dependent decision makers as those who “project responsibility for decisions upon others.” Thus, the tendency to not take responsibility for decision making may interfere with learning new methods for making decisions. These results imply that some initial training modules on decision-making styles given prior to StaySafe might be appropriate to prime participants for receptivity to an intervention such as StaySafe. That is, including some early training modules on taking responsibility for one’s own decisions might be helpful. Including decision-making scenarios involving dependent decision making may also be a useful alternative.
Although the relationships with DM-rational were in the same directions generally as with DM-dependent (that is, people high on rational decision making were less likely to benefit from the StaySafe intervention), the underlying process may be different. People high in DM-rational assume “personal responsibility when making decisions and use a deliberate and logical approach” (Harren, 1979). Thus, it is possible that these individuals do not see a strong need to change their approach and may be more resistant to learning a new method. Taking responsibility for their own decisions as well as being logical and deliberate does not necessarily imply that they are making good decisions. Adding a training module to address such resistance and to build on the personal responsibility to be open to new techniques might be helpful.
Study limitations
Several study limitations should be noted. The study sample includes clients from probation-run substance use treatment centers and may not be representative of others with substance use disorders who are on probation. Health risks and decision making in a controlled setting, such as residential sites, may have a different meaning than in community settings. Correspondingly, outcome measures on motivation for HIV testing or risk reduction strategies may have different implications for people in residential settings versus those in the community. The study sample also was limited in that there was considerable attrition in the follow-up due to individuals not returning for their follow-up interviews. We have conducted research into testing the hypotheses comparing StaySafe vs. the individuals receiving the standard procedure (SP) using a paradigm for replacing missing data for those who did not return for their follow-up interview and also the use of multiple imputation for testing the hypotheses comparing StaySafe and SP. The results supported the original findings concerning StaySafe vs. SP.
Furthermore, the current research is limited by the reliance on self-report measures of attitudes toward HIV needle and sex-risky behaviors, motivation for treatment, and psychological and social functioning as outcome measures rather than actual behaviors. These attitudes are more indicators of intent and therefore are more intermediary outcomes.
Future studies would benefit by incorporating behavioral tests and measures to validate self-reports (e.g., HIV testing data when interpreting participant motivation and its contribution as a risk reduction strategy).
Another limitation of the current research is that it did not include all of the decision-making styles identified by Scott and Bruce were able to be included in this research due to the StaySafe study did not include scales to measure the “Intuitive” decision-making style nor did the “Avoidant” decision-making style.
Conclusions
The present study augments current literature through findings that show individuals with substance use disorders often differ in their decision-making styles and that these differences can impact intervention receptivity (Shaghaghy et al., 2011). Among the many decision-making styles for which scales have been developed, those by Scott and Bruce (1995) have been well-studied and validated. Two of their scales, decision making rational and decision making dependent, are relevant to consider with respect to interventions targeting improving decision making among drug users.
Acknowledgments
The authors of this paper wish to thank CSCD officials in the three Texas counties in which this study took place for allowing us to conduct the research at their facilities, for staff at the participating facilities for their assistance in providing space and helping to schedule data sessions, and to the clients who willingly participated in the research.
Funding details:
This work was supported by the National Institute on Drug Abuse, National Institutes of Health (NIDA/NIH) under Grant R01DA025885. Interpretations and conclusions in this paper are entirely those of the authors and do not necessarily reflect the position of NIDA/NIH or the Department of Health and Human Services.
Footnotes
Disclosure of interest statement
The authors report that there are no competing interests to declare.
Data availability statement
Data supporting the results and analyses presented in this paper reside with the corresponding author.
References
- Allwood CM, & Salo I. (2012). Decision-making styles and stress. International Journal of Stress Management, 19(1), 34–47. 10.1037/A0027420 [DOI] [Google Scholar]
- Baiocco R, Laghi F, & D’Alessio M. (2009). Decision-making style among adolescents: relationship with sensation seeking and locus of control. Journal of Adolescence, 32(4), 963–976. 10.1016/J.ADOLESCENCE.2008.08.003 [DOI] [PubMed] [Google Scholar]
- Bavolar J, & Bacikova-Sleskova M. (2020). Decision-making styles and mental health—A person-oriented approach through clustering. Journal of Behavioral Decision Making, 33(5), 629–642. 10.1002/BDM.2183 [DOI] [Google Scholar]
- Becan JE, Knight DK, Crawley RD, Joe GW, & Flynn PM (2015). Effectiveness of the Treatment Readiness and Induction Program for Increasing Adolescent Motivation for Change. Journal of Substance Abuse Treatment, 50, 38. 10.1016/J.JSAT.2014.10.002 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Chen S, Yang P, Chen T, Su H, Jiang H, & Zhao M. (2020). Risky decision-making in individuals with substance use disorder: A meta-analysis and meta-regression review. Psychopharmacology, 237(7), 1893–1908. 10.1007/S00213-020-05506-Y [DOI] [PubMed] [Google Scholar]
- Cooper G, Tindall-Ford S, Chandler P, & Sweller J. (2001). Learning by imagining. Journal of Experimental Psychology. Applied, 7(1), 68–82. 10.1037//1076-898X.7.1.68 [DOI] [PubMed] [Google Scholar]
- Curşeu PL, & Schruijer SGL (2012). Decision Styles and Rationality: An Analysis of the Predictive Validity of the General Decision-Making Style Inventory. Educational and Psychological Measurement, 72(6), 1053–1062. 10.1177/0013164412448066/ASSET/IMAGES/LARGE/10.1177_0013164412448066-FIG1.JPEG [DOI] [Google Scholar]
- Douma LN, Uiters E, & Timmermans DRM (2020). Decision-making styles in the context of colorectal cancer screening. BMC Psychology, 8(1). 10.1186/S40359-020-0381-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Driver M. (1979). Individual Decision Making and Creativity. In Organizational Behavior. Grid Publishing. [Google Scholar]
- Gambetti E, Fabbri M, Bensi L, & Tonetti L. (2008). A contribution to the Italian validation of the General Decision-making Style Inventory. Personality and Individual Differences, 44(4), 842–852. 10.1016/J.PAID.2007.10.017 [DOI] [Google Scholar]
- Harren VA (1979). A model of career decision making for college students. Journal of Vocational Behavior, 14(2), 119–133. 10.1016/0001-8791(79)90065-4 [DOI] [Google Scholar]
- Institute of Behavioral Research. (2007). TCU Psychological Functioning (TCU PSYForm). TCU Institute of Behavioral Research. https://ibr.tcu.edu/wp-content/uploads/2016/10/TCU-PSYFORM-updated-footer.pdf [Google Scholar]
- Institute of Behavioral Research. (2008). TCU Global Risk Assessment. TCU Institute of Behavioral Research. https://ibr.tcu.edu/wp-content/uploads/2017/05/A-RSKForm-sg-v12.pdf [Google Scholar]
- Joe GW, Broome KM, Rowan-Szal GA, & Simpson DD (2002). Measuring patient attributes and engagement in treatment. Journal of Substance Abuse Treatment, 22(4), 183–196. 10.1016/S0740-5472(02)00232-5 [DOI] [PubMed] [Google Scholar]
- Knight DK, Donald, Dansereau F, Becan JE, Rowan GA, & Flynn PM (2015). Effectiveness of a Theoretically-Based Judgment and Decision Making Intervention for Adolescents. Journal of Youth and Adolescence, 44(5), 1024–1038. 10.1007/s10964-014-0127-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Knight DK, Joe GW, Crawley RD, Becan JE, Dansereau DF, & Flynn PM (2016). The Effectiveness of the Treatment Readiness and Induction Program (TRIP) for Improving During-Treatment Outcomes. Journal of Substance Abuse Treatment, 62, 20–27. 10.1016/J.JSAT.2015.11.007 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Lehman WEK, Pankow J, Muiruri R, Joe GW, & Knight K. (2021). An evaluation of StaySafe, a tablet app to improve health risk decision-making among people under community supervision. Journal of Substance Abuse Treatment, 130. 10.1016/J.JSAT.2021.108480 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Lehman WEK, Pankow J, Rowan GA, Gray J, Blue TR, Muiruri R, & Knight K. (2018). StaySafe: A self-administered android tablet application for helping individuals on probation make better decisions pertaining to health risk behaviors. Contemporary Clinical Trials Communications, 10, 86–93. 10.1016/J.CONCTC.2018.03.010 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Lehman WEK, Rowan GA, Greener JM, Joe GW, Yang Y, & Knight K. (2015). Evaluation of WaySafe: A Disease-Risk Reduction Curriculum for Substance-Abusing Offenders. Journal of Substance Abuse Treatment, 58, 25–32. 10.1016/J.JSAT.2015.05.007 [DOI] [PMC free article] [PubMed] [Google Scholar]
- McKenney J, & Keen P. (1974). How Managers’ Minds Work. Harvard Business Review. https://hbr.org/1974/05/how-managers-minds-work [Google Scholar]
- Naveed Riaz M, Riaz MA, & Batool N. (2012). Personality Types as Predictors of Decision Making Styles. Journal of Behavioural Sciences, 22(2). [Google Scholar]
- Scott SG, & Bruce RA (1995a). Decision-Making Style: The Development and Assessment of a New Measure: Educational and Psychological Measurement, 55(5), 818–831. 10.1177/0013164495055005017 [DOI] [Google Scholar]
- Scott SG, & Bruce RA (1995b). Decision-Making Style: The Development and Assessment of a New Measure. Educational and Psychological Measurement, 55(5), 818–831. 10.1177/0013164495055005017 [DOI] [Google Scholar]
- Shaghaghy F, Saffarinia M, Mohadeseh I, & Ali S. (2011). The Relationship of Decision-Making Styles and Attributional Styles in Addicted and Non-addicted Men - PubMed. Addiction Health, 3–4, 99–104. https://pubmed.ncbi.nlm.nih.gov/24494123/ [PMC free article] [PubMed] [Google Scholar]
- Simpson DD, Joe GW, Knight K, Rowan-Szal GA, & Gray JS (2012). Texas Christian University (TCU) Short Forms for Assessing Client Needs and Functioning in Addiction Treatment. Journal of Offender Rehabilitation, 51(1–2), 34–56. 10.1080/10509674.2012.633024 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
Data supporting the results and analyses presented in this paper reside with the corresponding author.