What specific datasets and methods did Election Truth Alliance use in their Florida county scatterplot analyses?

Executive summary (2–3 sentences)

The Election Truth Alliance (ETA) constructed Florida county scatterplots using precinct‑level vote results as individual data points, separating vote types where possible and applying turnout‑vs‑vote‑share plotting, binning, correlation and linear‑slope measures to look for systematic dependence of a candidate’s share on turnout (their “turnout analysis”) ^{[1] [2]}. ETA’s public writeups cite specific county statistics (for example St. Lucie’s r = +0.750 and slope +1.2969) and compare Florida patterns to anomalies they previously highlighted in Pennsylvania ^{[3] [4]}.

1. Data inputs ETA says it used

ETA’s documentation states that each plotted point typically represents a candidate’s result in a single precinct, meaning their primary dataset is precinct‑level election returns (vote counts and calculated share) and precinct turnout figures when available ^[1]. The group also emphasizes examining different “vote types” (Mail‑In, Early, Election‑Day) in isolation whenever jurisdictional data allow, and notes that turnout analysis requires precinct voter registration data where jurisdictions publish it ^[2]. ETA hosts an interactive dashboard for its datasets, indicating they aggregate and display compiled precinct returns on data.electiontruthalliance.org ^[5].

2. Scatterplot construction and visual framing

ETA’s method frames scatterplots with each precinct as a colored dot to show the relationship between turnout and candidate vote share, a visualization they call useful because it “allows us to look at all of the data points at once” ^{[2] [1]}. They describe using binning — grouping precincts by turnout ranges — to clarify trends rather than relying on a raw‑point cloud alone, and explicitly distinguish that their turnout charts are not time‑series of ballots counted over time but cross‑section correlations between turnout and vote share ^[2]. Their materials show linear fits (slope) and Pearson correlation coefficients (r) to quantify the strength and direction of any turnout‑vote‑share association ^[3].

3. Additional analytical techniques ETA reports using

Beyond turnout scatterplots, ETA lists other approaches such as “Down‑Ballot Difference” or “Drop‑Off Analysis” — comparing vote shares across races to spot anomalies where top‑of‑ticket and down‑ballot patterns diverge — and they indicate more techniques are under vetting before broader release ^{[2] [1]}. They state a practice of isolating vote types and comparing jurisdictions that use particular voting systems (ETA cites St. Lucie County’s use of Dominion systems in its county discussion) as part of contextual interpretation ^[3].

4. What ETA claims they found in Florida counties

In their Florida county writeups ETA reports seeing a “strong correlation” in counties like St. Lucie (r = +0.750; slope +1.2969) and patterns in Miami‑Dade where higher‑turnout precincts “heavily favor Donald Trump” in their plots, and they liken these visuals to anomalies observed in Pennsylvania ^[3]. Those statements are presented as preliminary and framed as potentially impactful to margins if the pattern were statewide ^[3].

5. Strengths, transparency claims and limits in ETA’s published materials

ETA publicly documents their conceptual workflow (precinct points, binning, separate vote types) and provides an interactive dashboard, which are transparent practices that allow independent inspection in principle ^{[2] [5]}. However, the sources here do not enumerate every raw source file, data‑pull date, precinct mapping decisions, software/code notebooks, or the statistical tests and robustness checks applied beyond reporting r and slope; ETA acknowledges that some analyses require further vetting and peer review before wider release ^[1]. The provided materials therefore make clear methods at a high level but leave open verification of replication details and potential confounders ^{[2] [1]}.

6. Bottom line and what remains to be checked

ETA’s Florida county scatterplots are built from precinct‑level returns and turnout figures, use turnout‑vs‑vote‑share scatterplots with binning, and report linear slopes and correlations to flag systematic relationships; they supplement this with down‑ballot comparisons and vote‑type disaggregation and publish findings on a dashboard and in county reports ^{[1] [2] [5] [3]}. What remains necessary for independent adjudication are the raw precinct datasets, precise preprocessing steps, code and sensitivity analyses that ETA has not fully published in the material provided here—gaps ETA itself signals will be closed after vetting and peer review ^[1].