Mixed Method Evaulation of Jockey Club Project ReBond

To capture the two-dimensional nature of act desistance-specifically the frequency and severity of reoffending, this construct is measured using the self-constructed item: "I have committed fewer crimes than before." Both items are rated on a 6-point scale ranging from 0 (Not committing any crime) to 5 (Committing more/more serious crimes than before). Higher scores indicate worse reoffending outcomes.

The PROLI-ex measures post-release daily routines and behaviors linked to health and desistance outcomes, which predict self-reported reoffending. Upon request, the original authors provided an unpublished, localized short Chinese version of the scale. This 17-item scale comprises four subscales: institutional routine, active living, online leisure, and support seeking (α=0.71,0.69,0.81,0.69). Participants indicate how regularly they performed specific routines over the past two weeks on an 11-point Likert scale ranging from 0 (Not regular at all) to 10 (Very regular). Scores are calculated by averaging the items within each subscale, as well as averaging all items for a total score.

To measure engagement with the ReBond project, participants will complete a self-constructed 4-item measure of continuance intention. Items are rated on a 7-point Likert scale ranging from 1 (Strongly disagree) to 7 (Strongly agree). Higher scores indicate a stronger intention to continue. The internal consistency of this measure will be calculated and reported in the final analysis.

Pro-social identity, conceptualized as identity desistance, is measured using the 5-item Identification with Desistance scale, which assesses an individual's subjective law-abiding identity. Participants rate their responses on a 5-point Likert scale ranging from 1 (Rarely) to 5 (Always). The final score is calculated by summing the items, with higher scores indicating a stronger pro-social identity. The reported internal consistency for this scale is α=0.62.

Relational desistance is operationalized as the level of concern regarding various reentry barriers across several domains: basic needs, work and career, health and well-being, digital literacy, and interpersonal relationships. Adapted from the Barriers to Reentry Success Inventory and integrated with clinical insights, this scale asks participants to rate their concerns on a 4-point Likert scale ranging from 1 (No Concern) to 4 (Great Concern), with an additional "Not Applicable" (N/A) option. Scores are calculated by summing the items for each subscale and for the total score, with higher scores indicating a higher need for reintegration support. Internal consistency will be calculated and reported after data collection.

Stigma consciousness refers to the anticipated stigma an individual expects to experience. A modified 10-item Stigma Consciousness Questionnaire is used, with the subject adapted to "ex-offenders." Participants rate items on a 7-point Likert scale ranging from 1 (Strongly disagree) to 7 (Strongly agree). Higher scores indicate higher anticipated stigma. Internal consistency will be calculated and reported after data collection.

The MSPSS measures an individual's perception of social support across three sources: significant others, family, and friends. This 12-item questionnaire utilizes a 7-point Likert scale ranging from 1 (Strongly disagree) to 7 (Strongly agree), with higher scores indicating higher perceived social support. Previous research indicates good to excellent reliability for the subscales: significant other (α=0.91), family (α=0.87), and friends (α=0.85).

To capture objective markers of social integration, the following self-constructed items are utilized:

• Education: Participants indicate whether they have received any education or training post-release (Yes/No). If responded "Yes," they specify the type.
• Housing: Participants indicate whether they have a stable shelter for the next 12 months (Yes/No), followed by categorical questions regarding ownership and housing type.
• Employment: Participants indicate their current employment status (Yes/No), where unpaid labor (e.g., caregiving) is classified as work. Data regarding occupation type and income level are also collected.
• Health: Participants indicate whether they have been diagnosed with any new health conditions post-release (Yes/No). Subjective health is assessed using a single item: "On a scale from 0 to 10 (0 = worst, 10 = best), how would you rate your own health over the past month?"

The ADQ assesses offenders' intentions to change and their perceived ability to stay crime-free. The original 10-item scale has acceptable internal consistency (α=0.77). Following communication with the original authors, 6 additional items were added to more accurately measure the construct. The internal consistency of the final 16-item scale will be calculated and reported in the final analysis.

The FoM-D assesses motivational factors (attitude, perceived social norm, self-efficacy, and positive affect) expected to predict the ADQ. The first three constructs are measured using a modified scale. Participants rate their responses on a 5-point Likert scale ranging from 1 (Rarely) to 5 (Always), some items with an added "Not Applicable" option. Scores are summed, with higher scores indicating stronger motivation. Reported internal consistencies for these subscales are α=0.62, α=0.65, and α=0.61. An option 'Not Applicable' is added so some questions where it asks the frequency of an experience. Since an option that is close to 'never' is not provided in the original scale. A 'Not Applicable' is added to those options with instructions to allow respondents to answer that the particular question does not apply to them.

Measured using the Positive Feelings subscale of the Comprehensive Inventory of Thriving (CIT). This is a 3-item subscale rated on a 5-point Likert scale from 1 (Strongly disagree) to 5 (Strongly agree). Scores are summed. While the full CIT has shown strong reliability (α=0.77 to 0.96), internal consistency for this specific subscale alone is not previously available and will be calculated.

The PEC measures individual differences in the identification, understanding, expression, regulation, and use of emotions. To align with workshop objectives, only three subscales are used: 'Identification of own emotions,' 'Understanding of own emotions,' and 'Expression of own emotions.' Items are rated on a 5-point Likert scale ranging from 1 (Does not describe me at all) to 5 (Describes me very well). The full 50-item scale has excellent internal consistency (α=0.93); the reliability of the specific subscales used will be calculated.

The BRS measures the ability to bounce back or recover from stress. This 6-item scale utilizes a 5-point Likert scale ranging from 1 (Strongly disagree) to 5 (Strongly agree). Previous research demonstrates good to excellent reliability across samples (α=0.80 to 0.91).

The AIL is a combined 12-item measure of subscales from ChatGPT literacy and AI self-efficacy scale. It sources 8 items from the intrinsic motivation and behavioral commitment subscale of the AILQ ( a = 0.88) and 4 items from the creative application subscale of the ChatGPT literacy scale (a = 0.79). Both scales use a 5-point Likert format ranging from 1 (Strongly disagree) to 5 (Strongly agree).

The PSI measures problem-solving confidence, approach-avoidance style, and personal control. To align with workshop objectives, only the problem-solving confidence subscale is used. Items are rated on a 6-point Likert scale ranging from 1 (Strongly disagree) to 6 (Strongly agree). The full inventory has excellent reliability (α=0.90), and the selected subscale is also highly reliable (α=0.85).

General well-being is assessed using a 6-item scale recommended by the literature as an optimal short measure. Items are rated on a 10-point Likert scale ranging from 1 (Not at all) to 10 (Completely). Internal consistency will be calculated and reported in the final analysis.

The CIT measures a broad range of psychological well-being constructs, representing a holistic view of positive functioning. This 54-item scale covers 18 domains, including support, learning, mastery experience, and autonomy. Participants rate items on a 5-point Likert scale ranging from 1 (Strongly disagree) to 5 (Strongly agree). Across five distinct samples in previous research, the CIT demonstrated acceptable to excellent internal consistency (α=0.77 - 0.96).