Expert on voter fraud finds evidence of serious fraud in Pennsylvania presidential election in 2024

Walter Mebane is a professor of political science at the University of Michigan. He is a leading expert on detecting voter fraud, having dedicated much of his career to developing statistical techniques for doing so.

He was asked to do a close analysis of the voting in the Pennsylvania presidential election, and his conclusions are to say the least disturbing:

Dr. Walter R. Mebane Jr. (University of Michigan)

Dr. Walter R. Mebane, Jr. is a leading U.S. expert in election forensics and detecting election fraud. He is a professor of political science and statistics at the University of Michigan.  ‘Election forensics’ means statistical methods used to determine whether the results of an election accurately reflect citizens’ voting behavior. 

• In April 2025, the Election Truth Alliance (ETA) met with with Dr. Mebane and requested his expert opinion and insights into the results of the 2024 U.S. Presidential election.

• The ETA sought Dr. Mebane’s expert opinion on election results in Pennsylvania (PA) in particular, as the ETA had recently undertaken and shared their own analysis of Three Counties in Pennsylvania (Philadelphia, Allegheny, and Erie Counties).

•  Dr. Mebane’s analysis of Pennsylvania was initially limited to the same three counties as the ETA (Philadelphia, Allegheny, and Erie). Subsequently, Dr. Mebane initiated eforensics analysis of all votes cast across all of PA’s 67 counties. 

The full report is available at https://websites.umich.edu/~wmebane/PA2024.pdf  and is also embedded as a PDF below.

This page contains three key sections:

1. High-Level Summary of Dr. Mebane’s Pennsylvania working report, which was drafted by the ETA team. It is informed by the working report, as well as conversations and written exchanges with Dr. Mebane. It is intended to help communicate his high-level findings to a broader audience. 
2. Expert Overview – “Walter R. Mebane, Jr & eforensics Model” (May 7, 2025) by the ETA Team. Provides a brief overview of Dr. Mebane’s key writings on election forensics, as well as a plain-language characterization of his eforensics model. 
3. Mebane’s Working Paper on the 2024 Presidential Election in Pennsylvania – “eforensics Analysis of the 2024 President Election in Pennsylvania” (June 2, 2025) by Dr. Walter R. Mebane Jr., a working report analyzing 2024 Pennsylvania results from the 2024 U.S. Presidential election.

Also on  this page:

• Previous Mebane ‘Three Counties’ Analysis (Summary + PDF) – “eforensics Analysis of Three Pennsylvania Counties in the 2024 Presidential Election“  (May 6, 2025) by Dr. Walter R. Mebane Jr., a working report analyzing Philadelphia, Allegheny, and Erie county results from the 2024 U.S. Presidential election. 

High-Level Summary

High-Level Summary of “eforensics Analysis of the 2024 President Election in Pennsylvania” by Dr. Walter R. Mebane, Jr. 

• The eforensics finite mixture model defines latent categories of fraud (no fraud, incremental fraud, and extreme fraud) based on votes and turnout, as well as relevant covariates (e.g., fixed county effects).
• Data from 7,040,360 votes (3,543,308 for Trump, 3,423,042 for Harris) across 67 PA counties (9,157 wards/precincts). 
• The eforensics model estimated that 225,440 votes in the Pennsylvania presidential race were possibly fraudulent. This would exceed the 120,266 vote margin of victory between Trump and Harris. 

More fine-grained analysis attempted to distinguish between strategic voting behaviors from “malevolent manipulation of votes”, i.e. how many votes may have been misdirected or misallocated due to malevolent distortions of voters’ intentions.

• In this analysis, 111,088 of the 225,440 possibly fraudulent votes[2] were estimated with high confidence to be malevolent manipulations of votes while the remainder were estimated to be a mix of manipulated votes and strategic voting behaviors.

A more conservative eforensics model including additional fixed county level effects estimated that 210,392 votes in the race were possibly fraudulent. This would exceed the 120,266 vote margin of victory in the race.

• Fine-grained analysis of the more conservative model attempted to distinguish between strategic voting behaviors from malevolent manipulation of votes. 
• In this analysis, 88,600 of the 210,392 possibly fraudulent votes were estimated to be malevolent manipulations of votes while the remainder were estimated to be a mix of manipulated votes and strategic voting behaviors.

The most conservative of the eforensics analyses estimated that 25,374 votes were due to malevolent manipulation of votes.

In summary: 

There is very high probability that a meaningful number of votes in the PA presidential election were subject to malevolent manipulation — and it was “a close call” whether “the election was decided or nearly decided by malevolent distortions of electors’ intentions.” (Mebane, Page 6)

 

Notes:

• Statistics, no matter how accurate, cannot provide definitive proof of voting fraud or election manipulation. Actual proof can only be found by comparing paper ballot audits to electronic voting records. 
• However, statistical information from eforensics and other data analysis approaches can be used to identify precincts and counties where voting fraud is most likely to have occurred.

A link for downloading the whole paper is here.

I haven’t read the paper yet, and I don’t know anything about election forensics, but given that EAIAC is the single most powerful explanatory principle of the entire Trump era . . .