Analyst: Gender Deviation Analysis Points to 300 000 Fraudulent Votes in Elections

Roman Udot, an election data analyst, forensic expert and former political prisoner in Russia, has shared his statistical analysis of the gender composition of voters in the October 26 elections, pointing to serious discrepancies according to the data provided by the Central Election Commission (CEC).

Udot, co-chair of Golos, Russia’s leading election monitoring organization, identifies “gender deviation” (disparity between male and female voter behavior) that appears to correlate with GD’s official results. Using a method similar to the Kiesling-Shpilkin technique – a tool commonly employed to detect and correct anomalies in election data, he estimates that GD benefited from 300,000 fraudulent votes at October 26 elections. Without this distortion, GD’s vote share drops to 45%, while further redistribution of stolen votes to other parties reduces it to 39%, costing GD its parliamentary majority.

Udot notes that when the distortion caused by gender deviation of the GD is corrected then:

• The suspicious distributions detected on October 27, 2024, disappear;
• All parties’ results become proportional to each other;
• GD’s results conform with the exit polls;
• Sexual anomalies in the data disappear;
• Marneuli, the zone of maximum violations is confirmed by the model as the most deviate;
• Georgian politics returns to a gender-neutral state.

Udot highlights the unique nature of Georgia’s election data, where traditional methods for detecting fraud are largely inapplicable because the Central Election Commission (CEC) only publishes voting results. However, the CEC also shares additional data, such as voter turnout by gender, with local observers. This data drew criticism from ISFED, prompting the CEC to revise it. Despite these changes, the data remains perplexing, prompting further scrutiny.

A visualization of registered male and female voters, as well as those who voted, reveals notable patterns, according to the elections analyst. Areas such as Tbilisi, Rustavi, and Kutaisi show gender balance and align with regions where Georgian Dream (GD) performed poorly. In contrast, significant gender imbalances are evident elsewhere. Aggregating the data by District Election Commission (DEC) highlights a “traffic light” effect, emphasizing these variations and raising further questions about the election results.

Udot’s analysis further examines unusual gender-based voting patterns, focusing on areas where civil election monitoring was limited. In regions with proper monitoring, deviations in turnout between men and women are minimal. To estimate the extent of these anomalies, Udot calculates the number of “Extra Men”—men who appear to have voted beyond what would be expected if male turnout matched female turnout. The formula used is:

• Men expected = (Registered men × Turned-out women) ÷ Registered women
• Extra men = Turned-out men – Men expected

Applying this method, Udot finds a total deviation of 94,000 “Extra Men” across the data, though it remains unclear whether these represent real individuals or irregularities in the voting process. Udot notes that “it is particularly striking how well the Marneuli district, notorious for its violations, anomalous data, and observer complaints, is identified by this parameter.”

“The unexpected hyperactivity of the male segment of the electorate could have been dismissed as an exotic feature of Georgian elections if it didn’t have a significant impact on voting results. And the results are strongly influenced — though only for one party. The influence of the “Extra Men” on voter turnout is negligible. This is a critical factor for restoring the results, notes the author. When plotting all values together, it becomes evident that there is a drastic difference in how various parties depend on the presence of these mysterious “Extra Men,” stresses Roman Udot.

Udot’s analysis compares two metrics—Men’s Hyperactivity and Gender Deviation—used to study irregularities in voter turnout. Men’s Hyperactivity measures the excess of male voting activity compared to female voting activity and is calculated as the difference between the turnout rates of men and women. In contrast, Gender Deviation highlights disruptions in the gender balance among voters and is calculated as the normalized difference in turnout rates between men and women. Both metrics are relative, making them unaffected by differences in the gender composition of the electorate.

Assuming that parties should generally reflect the gender composition of their voters, Udot breaks GD’s results into two components: normal votes (proportional to other parties’ patterns) and anomalous votes. Using a method similar to the Kiesling-Shpilkin approach for detecting turnout distortions, he estimates the number of anomalous votes at approximately 300,000.

This decomposition suggests two possible outcomes for GD’s vote share, depending on the nature of the manipulation. If fraudulent votes were added (e.g., through ballot stuffing or multiple voting), GD’s share adjusts to 45%. However, if votes were stolen from other parties (e.g., via intimidation, vote-buying, or misrecording), their share drops to 39%.

Udot’s analysis revisits previously identified voting anomalies, characterized by slanted clusters in election data favoring the ruling Georgian Dream (GD) party. These “tilted clouds,” located in the classic “comet tail” formation seen in manipulated elections, also display a distinct “same-sex voting” pattern—where turnout deviates from the gender-neutral voter composition seen in opposition-supporting regions. This phenomenon aligns with the Sobyanin-Sukhovolsky (S-S) Hypothesis, a framework for detecting electoral fraud through statistical patterns in voting data. The hypothesis suggests that fair voting results should align proportionally with turnout, but anomalies such as ballot stuffing or misrecording skew this relationship, creating steeper or even negative slopes in turnout-result correlations.

By applying this method to Georgia’s electoral data, Udot divides districts into two groups. In Group 1 (urban centers like Tbilisi, Rustavi, and Batumi), results align with the fair voting slope expected under the S-S Hypothesis. In Group 2 (other districts), however, slopes are distorted, suggesting ballot-stuffing (slopes approaching 45°) and vote misrecording (slopes steeper than 45°). These irregularities overwhelmingly favor the ruling party, confirming earlier observations of urban-rural voting discrepancies and pointing to manipulations outside major urban areas.

“It is important to emphasize once again that the identified manipulations did not affect turnout. This means that the predominant form of election result distortion was “reallocation.” As a result, when restoring the true results, the formula (Official results – fraudulent votes)/(Official turnout) should be used, which yields approximately 39% for Georgian Dream (GD),” the author notes.

“In other words, this party does not have a majority in Parliament,” concludes Udot.

Also Read:

• 25/11/2024 – “Georgian Dream” Usurps Power

This post is also available in: ქართული (Georgian)