Participants will use Rendever's newly designed VR-CARES support platform, developed with insights from the first focus group, for 8 weeks. Participants engage with other direct care workers in a virtual community support setting intended to provide supportive resources and content for group social activities and improve wellbeing and job satisfaction for direct care workers.

Caregiver empowerment is assessed with the 13-item Personal Self-Efficacy (PSE) and the 7-item Caregiver Self-Efficacy (CSE) subsections of the Caregiver Empowerment Scale. For more accurate detection of significance and measure of change over time, items are rated using the GHIS bipolar 100-point sliding response scale (-100 labeled as the most unpleasant personal experience, 0 neutral, and +100 representing the most pleasurable personal experience (Bartoshuk et al., 2004; Lim et al., 2009)) in place of a 5-point categorical scale of low to high. Item's scores are averaged for a total score. To score it traditionally for direct comparison in the literature, items' scores are divided back into the 5-point categorical scores, totaled, and multiplied by 2.5 (range: 0-100) (Lewis & Sauro, 2020). Higher scores indicate greater empowerment and confidence as a caregiver. Caregiver empowerment is a key outcome valued in social engagement interventions.

Three study-specific items assess caregivers' experiential and emotional dimensions of job satisfaction, complementing the structural job satisfaction domains captured by the CMS Employee Satisfaction Survey (primary outcome), providing a 2-dimensional view of job satisfaction. Items are rated on the Global Hedonic Intensity Scale (GHIS) bipolar 100-point sliding scale (-100 = most unpleasant personal experience, 0 = neutral, +100 = most pleasurable personal experience). The three items are: (1) Overall, how do you feel about the work you do? (2) How would you rate the level of care you provide to clients? (3) How positive, neutral, or negative do you feel about continuing to work at your current home care provider office? Scores are averaged across the three items. Higher scores indicate more positive feelings about caregiving work.

Technology acceptance is indicated with the 6-item Perceived Usefulness subset of the TAM, one item from the TAM Perceived Ease of Use subset (How pleasant or unpleasant was learning how to use the technology?) and one item from the TAM Attitude of Use subset (All things considered, how do you feel about using the technology?). The numerical GHIS, -100 (most unpleasant experience of your life) to 0 (neutral) to +100 (most pleasant experience of your life) is provided as the response scale (Lim et al., 2009), in place of a 7-point Likert scale. Scores of each item are averaged for a total score. To score it traditionally for direct comparison in the literature, scores of each item are divided back into the 7-point Likert scale scores and averaged. Higher scores indicate better acceptance, perceived usefulness, and adoption likelihood by the target population. TAM provides the adoption and sustainability narrative that complements usability data from the SUS.

The UEQ-S measures the overall appeal of the technology in two domains and 6 dimensions of user experience: pragmatic quality and hedonic quality. Pragmatic quality refers to the usability, efficiency, and clarity of the technology, while hedonic quality refers to the stimulation, interest, and novelty of it. Each of the 8 items consists of a pair of contrasting attributes (e.g., boring vs. exciting) rated on a 7-point semantic differential scale from -3 to +3. Two additional study-specific items were added: at T2 only, 'confusing for my clients vs. reassuring for my clients'; and at T2 and T3, 'creating more work for me vs. providing relief for me.' These study-specific items capture caregiver-relevant dimensions absent from the standard UEQ-S. Scores for each dimension are averaged. Values between -0.8 and 0.8 represent a neutral evaluation, while values > 0.8 indicate a positive user experience.

The UEQ+ Type 1 for Voice Assistants measures acceptance and satisfaction of voice interaction with a virtual assistant/chatbot/embodied, AI-powered companion and guide (Nova) within the Rendever social engagement platform that allows conversational control of the technology. 12 items make up 3 subsections of the UEQ+VA: Response Behavior, Response Quality, and Comprehensibility. 12 additional, study-specific items assess the voice assistant's communication and companionship capabilities (e.g., impatient with my clients with dementia vs. patient with my clients with dementia; a poor companion vs. a good companion). Each item consists of a pair of contrasting attributes (e.g., artificial vs. natural) rated on a 7-point semantic differential scale (-3 to +3). Scores for each dimension are averaged. Values between -0.8 and 0.8 represent a neutral evaluation. Values > 0.8 indicate a positive user experience. This measure captures the voice interaction experience unique to intervention 1.

Three of the four surveys of the UEQ+ Social Scales for interactive products are used to measure the social dimension of the user experience of the VR-CARES platform. The 3 sub-surveys measure social interaction, social stimulation, and social acceptance enabled by the VR-CARES program for the caregivers. Each of the 13 items consists of a pair of contrasting attributes (e.g., left out by others vs. welcomed by others) rated on a 7-point semantic differential scale from -3 to +3. Scores for each dimension are averaged. Values between -0.8 and 0.8 represent a neutral evaluation, while values > 0.8 indicate a positive social experience by the user. This measure is time-matched exclusively to Intervention 2 (the VR-CARES caregiver support program) and captures the social UX that is unique to that phase.

Three study-specific items assess experiential dimensions of the Rendever social engagement platform that are not captured by the standardized UX instruments. Items are rated on the Global Hedonic Intensity Scale (GHIS) bipolar 100-point sliding scale (-100 = most unpleasant personal experience, 0 = neutral, +100 = most pleasurable personal experience). The three items are: (1) What level of enjoyment did you experience when using Rendever by yourself? (2) What level of enjoyment did you experience when using Rendever with your clients? (3) How positive or negative was the impact of using Rendever with clients on your job satisfaction? Item 2 captures the shared-experience dimension that is central to the Rendever platform's value proposition for caregivers and clients. Item 3 directly links the VR experience to the job satisfaction outcome domain. Scores are reported individually and as a mean across all three items.

12 study-specific items assess intervention-relevant dimensions of the VR-CARES platform that aren't captured by standardized UX instruments, such as peer support, training quality, burden and stress relief. Items are rated on the GHIS bipolar 100-point sliding scale (-100 = most unpleasant personal experience, 0 = neutral, +100 = most pleasurable personal experience). Examples of the 12 items: In general, what level of enjoyment did you experience when using the VR-CARES platform? How was VR-CARES at creating new friendships and strengthening relationships between caregivers? How would you rate the group sessions in VR-CARES on relieving job stress? How was the VR-CARES training on caring for clients with dementia? How was VR-CARES at helping you overcome the social isolation that can occur with being a caregiver? Scores are reported individually and as a total mean. Higher scores indicate greater impact and satisfaction with the VR-CARES program.

The PHQ-4 is a 4-item ultra-brief clinical screening tool for depression and anxiety combining the PHQ-2 (depression) and GAD-2 (anxiety) subscales. Participants rate how intensely they have felt symptoms over the past 2 weeks using the 100-point Global Sensory Intensity Scale (GSIS): 0 = Not at All, 100 = as Intensely as the Intensity of the Brightness of the Sun. The total score is the average score of all items: range 0-100, with 25-49 indicating mild symptoms, 50-74 moderate symptoms, and 75-100 severe symptoms. The GSIS is a numerical scale with ratio properties and a universal intensity anchor independent of the domain of the items, allowing accurate measurement of change in score from each intervention and determination of clinical significance free from statistical artifacts of ordinal data. To score it traditionally for direct comparison in the literature, scores are divided back into a 4-point Likert scale and totaled. Lower scores indicate better mental health.

The Mental Health Inventory 5-item Scale will be used to assess general mental health and wellbeing. Items are rated on the 100-point Global Sensory Intensity Scale (GSIS): 0 = Not at All and 100 = as Intensely as the Intensity of the Brightness of the Sun. The total score is the mean of all items (range: 0 to 100). Two items are reverse coded before averaging so that higher means indicate better mental health. The GSIS provides ratio-level measurement with increased sensitivity to detect change and determine significance compared to the original categorical response options.

The PROMIS Companionship Short Form 4a measures perceived availability of friends and companions for positive social interaction. To increase measurement sensitivity while preserving the construct and response framework of the validated instrument, items are rated on a continuous 100-point slider scale ranging from 0% to 100% of the time, with the original 5-point Likert categorical labels retained as reference anchors along the slider: Never (0%), Rarely (25%), Sometimes (50%), Usually (75%), Always (100%). Higher scores indicate greater companionship. Raw scores will also be converted to standard T-scores (mean = 50, SD = 10) referenced to the U.S. general population for normative comparison in the PROMIS Social Relationships Scoring Manual.

The PROMIS Emotional Support Short Form 6a measures perceived availability of someone to talk to about problems, confide in, and turn to for support and understanding. To increase measurement sensitivity while preserving the construct and response framework of the validated instrument, items are rated on a continuous 100-point slider scale ranging from 0% to 100% of the time, with the original 5-point Likert categorical labels retained as reference anchors along the slider: Never (0%), Rarely (25%), Sometimes (50%), Usually (75%), Always (100%). Higher scores indicate greater emotional support. Raw scores will also be converted to standard T-scores (mean = 50, SD = 10) referenced to the U.S. general population for normative comparison in the PROMIS Social Relationships Scoring Manual.

Social network size and engagement will be assessed using the 6-item Lubben Social Network Scale. Items are rated on a 5-point scale with 1 being the least amount of contacts and 5 being the highest amount of contacts. The total score is the sum of all items (range: 6 to 30), with higher totals representing larger and more engaged social networks. Scores below 12 indicate risk for social isolation. The LSNS-6 captures objective network characteristics that complement the perceived social support measures (PROMIS Companionship and Emotional Support).

The short form, 3-item version of the UCLA Loneliness Scale will be used to assess loneliness. The three items are adapted to use the 100-point Global Sensory Intensity Scale (GSIS): 0 = No Sensation and 100 = Intense as the Brightness of the Sun. The score is the sum of all items (range: 0 to 300). 0-85 = not lonely, 86-171 = moderately lonely, 172-300 = very lonely. Higher scores indicate greater feelings of loneliness. The ULS-3 provides a loneliness-specific measure not captured by the companionship, emotional support, or social network scales, and its use of the GSIS places it within the psychophysical scaling framework used across the study battery, improving methodological consistency and measurement sensitivity.

11 items of the Archangels Caregiver Intensity Index assess the intensity and attitudes of caregiving responsibilities. ACII items pertaining to family caregivers (e.g. financial impact on family) were removed. Items evaluate stress, preparedness, ability to cope, and the personal impact of caregiving. To increase measurement sensitivity and maintain consistency with the psychophysical scaling approach used across the study battery, items are rated on the 100-point Global Sensory Intensity Scale (GSIS): 0 = No Sensation and 100 = Intense as the Brightness of the Sun. The total score is the mean of all items (range: 0 to 100). Higher scores indicate greater caregiving burden and intensity. This measure captures the practical demands placed on caregivers and complements the empowerment measures by providing a burden relieving-focused perspective.

Empathy will be assessed using a 20-item Jefferson Scale of Empathy adapted for Care Professionals (Hojat et al., 2018). Each item will be scored on a bipolar 100-point scale: -100 representing the strongest level of disagreement and +100 representing the strongest level of agreement. Higher scores indicate greater empathy toward clients. This measure assesses whether the social VR interventions, particularly the shared experience and perspective-taking facilitated by VR, influence caregivers' empathic engagement with their clients.