Ground Truth Strata

Dig into the data.

About

Methodology & attribution

Strata is a static explorer for the precomputed outputs of the UAS social-media panel survey (UAS514–UAS519). Every chart you see is a slice of a JSON file generated by a documented R pipeline. Nothing is computed on the fly.

Survey waves

Six waves of the Understanding America Study were collected between 2023 and 2025. Each row below shows the field dates and the unweighted sample size in the cleaned panel that Strata serves.

WaveUAS #Field datesTargeted nn in cleaned
Wave 1UAS514March 2 - May 7, 20233,0661,966
Wave 2UAS515August 7 - September 17, 20231,9661,619
Wave 3UAS516September 4 - October 15, 20231,9661,600
Wave 4UAS517November 6, 2023 - February 18, 20244,5202,984
Wave 5UAS518February 5, 2024 - May 19, 20244,4372,992
Wave 6UAS519October 14, 2024 - January 4, 20254,0902,742

Variable coverage

The Strata data dictionary catalogs 182 variables across 12 survey domains. 154 of those have analyzable rows in the precomputed JSONs — the remainder are reserved for free- response text (which lives outside the public repo) or items that require additional runtime expansion not implemented for v0.1.0.

DomainVarsAnalyzableText-onlyOther
AI_ATTITUDES644717
PLATFORM_USE433310
DEMOGRAPHICS1515
WELLBEING1212
SOCIAL_MEDIA_BELIEFS1010
POLITICAL1010
INSTITUTIONAL_TRUST88
DEPRESSION_ANXIETY66
TECH_IDENTITY55
LONELINESS44
TECH_REGULATION33
SURVEY_QUALITY211
Total182154280

Suppression

Cells with fewer than 30 respondents are suppressed by design. The underlying JSON rows are emitted with all statistic fields set to null and a suppressed: trueflag, rather than being omitted. Strata renders an “insufficient n” placeholder when it encounters one. The exact policy documented in meta.json:

Cells with n < cell_floor are emitted with NA values and suppressed:true (not omitted).

Weighting

Every precomputed row carries weighted estimates (weighted_value, with matching standard errors, confidence intervals, and effective sample sizes). UAS provides probability weights to adjust for panel design and non-response; weighted estimates are generalizable to U.S. adults at the time of each wave.

Correlations

All pairwise correlations in Strata are Spearman’s ρ, computed per wave. Spearman handles ordinal Likert and skewed count variables better than Pearson and does not assume a linear relationship. Where the underlying scale is reverse-coded, the flip is applied at the cleaning stage; the UI does not need to re-flip. Correlation is not causation, and small samples or small ρ values should be interpreted with caution.

Cite this tool

For aggregate citation:

Motyl, M. (2026). Ground Truth Strata (v0.1.0)
  [Data explorer for the UAS social-media panel survey].
  https://strata.mattmotyl.com

Individual findings have a per-chart citation block that records variable, wave, weighting status, and the precomputed-JSON generated timestamp.

Required UAS attribution

These data are from survey(s) administered by the Understanding America Study, which is maintained by the Center for Economic and Social Research (CESR) at the University of Southern California. The outputs of Strata are the responsibility of its users and do not necessarily represent the official views of USC or UAS. The tool should be considered an early beta-level app and results should be confirmed with the raw data. Consistent with open science practices, any individual is able to access these data for purpose of replication or research after free registration with the Understanding America Study (uasdata.usc.edu/page/Registration+Form), and provision of a signed data use agreement. You may also communicate with this tool’s creator via email or GitHub requests (github.com/mattmotyl/ground-truth-strata).

Precomputed JSON generated: May 27, 2026 at 3:22 PM

Pipeline: r/precompute/build_meta.R

Source data: r/output/cleaned/all_waves_long.rds

Source dictionary: docs/data-dictionary.json

← Back to Strata