Federal Sentencing and January 6 Defendants

This study examines whether sentencing outcomes involving January 6 defendants differ from sentencing outcomes observed among other federal defendants convicted of the same or meaningfully comparable offenses.

The project is being developed as a structured, provenance-tracked research dataset rather than as a simple collection of sentence totals. The analysis is intended to account for legally and factually relevant differences among defendants and cases before evaluating whether sentencing disparities remain.

Data collection, coding, and validation are ongoing. Findings are intentionally withheld until the underlying case, conviction, sentencing, guideline, conduct, and comparison data have been sufficiently reviewed.

The study is designed to examine both unadjusted sentencing differences and differences that remain after accounting for sentencing-relevant characteristics.

This project is an observational study rather than an experiment. Defendants were not randomly assigned to prosecutorial, adjudicative, or sentencing conditions, so the analysis must distinguish descriptive differences from causal claims.

The early phase of the project is exploratory and inductive. The dataset is being constructed, validated, and characterized before conclusions are drawn about the direction or magnitude of any sentencing differences.

Later phases may use formally specified hypotheses, matched comparison groups, regression models, or other statistical methods to determine whether observed differences remain after accounting for measured case and defendant characteristics.

A future confirmatory phase may evaluate a null hypothesis that sentencing outcomes for January 6 defendants do not systematically differ from outcomes among comparable federal defendants after adjustment for relevant characteristics.

The corresponding alternative hypothesis would be that such differences remain after adjustment.

The study does not assume in advance that January 6 defendants were treated either more harshly or more leniently.

The January 6 research cohort is being reconstructed from public source material and federal court records.

Raw source files and API responses are preserved independently of normalized research tables so that analytical observations can be traced back to their underlying evidence.

The project distinguishes among defendants, cases, source records, documents, conviction counts, sentencing events, and post-judgment events.

A federal docket may contain multiple defendants. A defendant may have multiple counts of conviction. A case may also contain multiple sentencing or post-judgment events.

As a result, the unit of analysis will depend on the specific question being evaluated. Some analyses may operate at the defendant-case level, while others may require count-level or sentencing-event-level observations.

The primary comparison is not intended to contrast January 6 defendants with the entire federal defendant population.

Instead, comparison groups will be constructed from federal defendants convicted of the same or meaningfully comparable offenses.

Potential comparison characteristics include:

• Statute or offense of conviction
• Guideline range
• Total offense level
• Criminal history category
• Plea versus trial disposition
• Acceptance of responsibility
• Violence or threats
• Weapon involvement
• Obstruction-related conduct
• Number and type of conviction counts
• Departures or variances
• Sentencing year
• District or judge where analytically appropriate

Final comparison variables will be specified after data completeness and measurement quality have been evaluated.

The research database is being built as a reproducible data pipeline rather than as a manually maintained spreadsheet.

The pipeline uses deterministic identifiers, source checksums, raw data caching, reconciliation reports, manual review states, relational constraints, and automated tests.

analytic observation
        ↓
coded fact
        ↓
source document
        ↓
docket entry
        ↓
original source

Automated extraction may be used to identify candidate facts from docket entries and court documents, but legally significant fields are not assumed to be correct solely because a text pattern or source indicator is present.

Ambiguous or conflicting information may be flagged for manual review. The project is designed to preserve the distinction between source-provided information, machine-extracted candidates, and independently reviewed research variables.

The study will distinguish among descriptive disparities, adjusted statistical associations, and causal explanations.

An observed difference in sentence length would not by itself establish political discrimination, prosecutorial intent, or another causal mechanism.

Likewise, the absence of a detectable aggregate difference would not establish that every individual case was treated equivalently.

This study depends on the completeness and quality of public court records and available comparison data.

Some docket documents may be unavailable, sealed, missing from RECAP, or lack usable text extraction. Legal events may also be described differently across courts and cases.

Measurement decisions, comparison-group construction, missing data, and unobserved case characteristics may affect interpretation of the final results.