Cross-Cultural Comparison Dashboard
Aggregate data only — no individual records
This portal provides an interactive comparison of harmonized survey data from two independent studies on AI use for mental health support. All data shown is aggregate only — no individual participant data is stored or transmitted through this portal.
| Israel | United States | |
|---|---|---|
| Sample | General population adults, ages 18-70 | Adults with ongoing MH conditions who used LLMs |
| N | 584 | 499 |
| AI-MH Users | 171 (29.3%) | 243 (48.7%) |
| Recruitment | iPanel (online panel) | Prolific |
| Investigators | Refoua, Hadar Shoval, Elyoseph, Levkovich, Haber (Bar-Ilan University) | Rousmaniere (Sentio University) |
| DASS-21 | Collected (0-126) | Not collected |
| Key Difference | The US sample is pre-filtered (all have MH conditions + all used LLMs). The Israeli sample is general population. Direct prevalence comparisons require careful interpretation. | |
The two surveys measured overlapping constructs using different instruments and languages. Harmonization was performed in R, with every recoding decision documented in the project codebook.
Israel: Numeric codes (1=Male, 2=Female, 3=Other). US: Full text labels including "Man (incl Trans Male/Trans Man)", "Woman (incl Trans Female/Trans Woman)", "Non-binary/Genderqueer", etc. Mapped to three categories: Male, Female, Other.
Israel: Years of schooling (numeric). US: Categorical text labels. Both mapped to a common 0-5 ordinal scale.
| Ordinal | Label | Israel (years) | US (text) |
|---|---|---|---|
| 0 | No formal | — | "No formal qualifications" |
| 1 | High School | ≤12 | "Secondary" / "High school diploma" |
| 2 | Some College | 13-15 | "Technical/community college" |
| 3 | Bachelor's | 16 | "Undergraduate degree" |
| 4 | Master's | 17-18 | "Graduate degree" |
| 5 | Doctorate | ≥19 | "Doctorate degree" |
The two surveys used different frequency anchors. A best-effort harmonized ordinal (0-6) was created, but the mapping is imperfect.
| Harmonized | Israel Label | US Label |
|---|---|---|
| 0 | — (not in data) | Never |
| 1 | Once a year | — |
| 2 | Few times a year | 1x month |
| 3 | About monthly | 2-3x month |
| 4 | About weekly | Weekly |
| 5 | Several times a week | Several times a week |
| 6 | Daily | Daily |
Israel: Wide-format binary columns (Tool_1 through Tool_9). Each column is 1 if selected, NA if not. US: Single comma-separated text field listing all tools used. Parsed with regex pattern matching.
Israel: 1-5 Likert numeric scales. US: Text labels ("Very Unhelpful"=1, "Unhelpful"=2, "Neutral"=3, "Helpful"=4, "Very Helpful"=5). Both available only for AI-MH users.
Israel: Direct binary question (MHAI_severe: 0=No, 1=Yes). US: Derived from crisis types question — if participant selected "I have not used" or left blank, coded as 0; otherwise coded as 1.
The DASS-21 (Depression Anxiety Stress Scales) was administered only in the Israeli sample. This portal allows filtering the Israeli subsample by DASS severity levels to create more comparable subgroups.
| Subscale | Normal | Mild | Moderate | Severe | Extreme |
|---|---|---|---|---|---|
| Total (0-126) | 0-29 | 30-41 | 42-59 | 60-83 | 84-126 |
| Depression (0-42) | 0-9 | 10-13 | 14-20 | 21-27 | 28-42 |
| Anxiety (0-42) | 0-7 | 8-9 | 10-14 | 15-19 | 20-42 |
| Stress (0-42) | 0-14 | 15-18 | 19-25 | 26-33 | 34-42 |
Since the US sample was pre-selected for MH conditions, filtering Israelis to higher DASS levels creates a more comparable distress profile.
Click the info button on any chart card in the Dashboard tab to see exact question wordings, source columns, recoding details, and caveats.