This site was developed as the final project for the Data Visualization class at UC Berkeley’s Masters in Data Science (MIDS) program. Driven by a shared passion for mental health, we drew on the National Survey on Drug Use and Health (NSDUH) to explore trends in mental well‑being in the United States. Our goal is to transform complex public health data into a narrative that is intuitive and compelling, but grounded in concrete figures—and ultimately to raise awareness through insight.

Alongside the storytelling views, you’ll find an interactive explorer that lets curious visitors dive deep into the same NSDUH dataset and examine trends at will.

We leveraged publicly available NSDUH data, the national source of nationally representative statistics on substance use, mental health, and service utilization in the U.S. Here are a few key aspects about the data relevant to our website and analysis.

Population Sampling & Estimation Method

• NSDUH samples the civilian, non‑institutionalized U.S. population aged 12 +, using multistage, stratified area probability sampling. It surveys households and non‑institutional group quarters (e.g., shelters, dorms), while excluding active‐duty military and unsheltered homeless individuals. For our study, we have excluded participants below 18 due to survey data scarcity.
• Data is collected via face‑to‑face and web‑based interviews, depending on location and timing: in 2021–22, multimode collection was used, with over 70% of interviews completed online as COVID protocols shifted the methodology.
• Because NSDUH uses complex stratified sampling, not every person has an equal chance of selection (e.g., some states or demographics are oversampled). Therefore NSDUH provided a weight column (ANALWC_2 or variant) to each participant’s data point to correct for these unequal probabilities, nonresponse, and post-stratification adjustments so that estimates align with U.S. Census population controls.

Mental Health Severity

• In our analysis, mental illness severity (mild, moderate, or serious) is based on MICATPY field, which is based on NSDUH’s model-based predictions rather than direct clinical diagnoses. NSDUH applies a validated statistical model built from a subsample of respondents who completed detailed clinical interviews. For any of the recent surveys, the statistical model from 2013 has been used.
• This model uses responses to key survey questions—including measures of psychological distress, daily functioning, and other mental health indicators—to estimate each respondent’s likelihood of having mild, moderate, or serious mental illness in the past year.

Alcohol Use Categories

• Never Used Alcohol: Never consumed alcohol.
• Did Not Use Alcohol in Past Year: No alcohol use in the 12 months prior to the survey.
• Occasional Use: 1–12 days in the past year (less than once a month).
• Light to Moderate Use: 13–52 days in the past year (less than once a week).
• Moderate Use: 53–156 days in the past year (less than three times a week).
• Moderate to Heavy Use: 157–250 days in the past year.
• Heavy Use: 251–365 days in the past year.