Skip to main content

Research methods

Methodology and analytical framework

The research methodology appendix provides a transparent account of survey design, data preparation, variable construction, vulnerable-group definitions, statistical modeling, and study limitations underlying Economic and Human Cost of the USCIS Adjudication Pause.

Transparent sample construction

The study collected 189 completed survey forms between February and March 2026. The core econometric and demographic analyses used 149 valid responses after excluding records missing core demographic variables.

Mixed-method, seven-domain instrument

The survey combined structured measures across demographics, education, employment, immigration, financial, legal and family, health, and migration-intention topics with voluntary open-ended testimony.

Multiple statistical checks

Analysis included descriptive statistics, chi-square tests, Welch’s t-tests, ANOVA, Pearson correlations, OLS models with HC3 robust standard errors, and binary logistic regression reported as odds ratios.

Explicit limits on interpretation

The appendix identifies purposive sampling, self-reporting, subgroup size, temporal specificity, and cross-sectional design as limitations. Regression coefficients are treated as conditional associations, not isolated causal effects.

Study design

The primary data source was a self-administered online community survey distributed through immigrant advocacy networks and community organizations between February and March 2026. The instrument moved from stable factual questions to more sensitive financial, health, and future-intention questions, and allowed respondents to report multiple simultaneous harms.

The study collected 189 completed survey forms. For core econometric and demographic analysis, records missing core demographic variables were excluded, producing an analytic sample of 149 valid respondents. Individual findings may use smaller denominators where respondents did not answer a particular question.

Voluntary open-ended fields supplied the qualitative evidence used alongside the quantitative analysis. This convergent approach preserves both the measured incidence of hardship and respondents' accounts of how those harms affected daily life.

Analytical methods

Impact indices

Additive financial, legal, and family impact counts built from binary survey items.

Bivariate tests

Chi-square tests, Welch’s t-tests, one-way ANOVA, and Pearson correlations.

OLS models

Continuous and count outcomes modeled with HC3 heteroskedasticity-consistent standard errors.

Logit models

Binary outcomes estimated by maximum likelihood and reported as odds ratios.

Vulnerability analysis

Four non-exclusive subgroups and a cumulative vulnerability score tested compounding harm.

Diagnostic discipline

Pairwise correlations checked before modeling, with standard significance thresholds reported.

Interpretation and limitations

The sample was purposively recruited rather than randomly drawn, all measures are self-reported, some vulnerable subgroups are small, and the survey captures a single period during the pause. The analysis therefore documents experiences and conditional associations within the reached population; it does not establish population-wide prevalence or isolated causal effects. These boundaries are stated directly in the appendix and should accompany use of the findings.

Download the Research Methodology Appendix

The seven-page appendix contains the full methodological account, including formal model specifications and vulnerable-subgroup definitions.

Live data on this site

Figures cited on this page come from the published July 2026 report. Interactive dashboards on this site show approved Round 1 survey data from the live database and may differ slightly if additional responses are approved after publication.

Take the Round 2 survey · Explore Round 1 interactive data