SoCal Strategies National, Pennsylvania, and Arizona Polls

Toplines

In the final On Point/Red Eagle Politics/SoCal Strategies polls of the cycle, Donald Trump holds a 50-49 lead nationally and a 50-49 lead in Arizona, while Kamala Harris leads 50-48 in Pennsylvania.

Methodology-National

Methodology

The SoCal Strategies National Poll of 984 likely voters was conducted from October 30-31 using the Pollfish and Lucid Theorem online panels. The sampling error for this survey is +/-3.1%. It is important to note that sampling errors for subgroups are greater, and this does not account for other error such as non-coverage bias, response bias, or a non-representative sample.

Data Collection-Pollfish

The Pollfish Panel utilizes thousands of partner apps to contact respondents through random digital engagement. Respondents are filtered to include only US Registered Voters. 600 registered voters were contacted on Pollfish. 

Data Collection-Lucid Theorem

Lucid Theorem delivers a nationally representative sample of US adults based on age, gender, ethnicity, and region according to the US census. Lucid respondents are directed to the questionnaire created on Qualtrics and re-directed when finished. 600 registered voters were contacted from Lucid Theorem. 

Respondent Quality

Both Pollfish and Qualtrics utilize anti-fraud systems to ensure data quality. Pollfish uses an AI-driven algorithm that detects suspicious responses by examining response speed. Qualtrics detects fraudulent responses by looking at response speed and eliminating straight-lining. SoCal Strategies also uses an attention check question in its questionnaire. Respondents who failed this question were eliminated from the survey.

Likely Voter Model

The likely voter model consists of 3 questions:

How likely are you to vote?

How enthusiastic are you to vote in the 2024 presidential election?

How often do you vote?

Each answer to these questions is given a point value. To determine which respondents are likely voters, all points are added up, and those with under 4 points are screened out of the likely voter sample.

Weighting-Likely Voters

The SoCal Strategies National Poll was weighted by race, age, gender, education, and party identification The survey is weighted in two phases. First, SoCal Strategies’ weighting model combines Pew Research, Catalist, AP VoteCast, and Edison Research Exit Polls to create a baseline estiamte for the 2024 electorate. After conducting the survey, SoCal Strategies examines which groups represent a greater percentage of the likely voter sample than expected compared to the registered voter sample. These groups are given greater weight in the survey than the original baseline. Party ID is weighted to match the latest Pew Research estimates. Weights are applied in Qualtrics using rake weighting.

Methodology-AZ/PA

The SoCal Strategies Arizona and Pennsylvania Poll were conducted from October 30-31 using the Pollfish online panel. The sampling error in Pennsylvania is +/-3.6% and is +/-4% in Arizona . It is important to note that sampling errors for subgroups are greater, and this does not account for other error such as non-coverage bias, response bias, or a non-representative sample.

Data Collection-Pollfish

The Pollfish Panel utilizes thousands of partner apps to contact respondents through random digital engagement. 850 respondents were contacted on Pollfish in Pennsylvania and 750 were contacted in Arizona. The sample was selected to include only registered voters. A gender quota was implemented in Pennsylvania to ensure a 53-47 female sample.

Respondent Quality

Pollfish utilizes anti-fraud systems to ensure data quality. Pollfish uses an AI-driven algorithm that detects suspicious responses by examining response speed. SoCal Strategies also uses an attention check question in its questionnaire. Respondents who failed this question were eliminated from the survey.

Likely Voter Model

The likely voter model consists of 3 questions:

How likely are you to vote?

How enthusiastic are you to vote in the 2024 presidential election?

How often do you vote?

Each answer to these questions is given a point value. To determine which respondents are likely voters, all points are added up, and those with under 4 points are screened out of the likely voter sample.

Weighting

The SoCal Strategies Arizona and Pennsylvania Polls were weighted by race, age, gender, education, and statewide. It is not weighted by 2020 recalled vote or party identification. The survey is weighted in two phases. First, SoCal Strategies’ weighting model combines ACS, Bonfire Voter File, AP VoteCast, and Edison Research data in each state to generate a baseline for turnout. After conducting the survey, SoCal Strategies examines which groups represent a greater percentage of the likely voter sample than expected compared to the registered voter sample. These groups are given greater weight in the survey than the original baseline. Weights are applied in Qualtrics using rake weighting.

Pennsylvania Region Definitions:

• Philadelphia: Philadelphia

• Southeast: Bucks, Chester, Delaware, Montgomery

• Northeast: Carbon, Lackawanna, Lehigh, Luzerne, Monroe, Northampton, Pike, Schuylkill, Susquehanna, Wayne, Wyoming

• Dutch: Adams, Berks, Cumberland, Dauphin, Franklin, Lancaster, Lebanon, Perry, York

• Central: Armstrong, Bedford, Blair, Bradford, Cameron, Centre, Clarion, Clearfield, Clinton, Columbia, Elk, Forest, Fulton, Huntingdon, Indiana, Jefferson, Juniata, Lycoming, McKean, Mifflin, Montour, Northumberland, Potter, Snyder, Sullivan, Tioga, Union, Venango, Warren

• West: Beaver, Butler, Cambria, Crawford, Erie, Fayette, Greene, Lawrence, Mercer, Somerset, Washington, Westmoreland

• Allegheny: Allegheny

Click here to view the interactive crosstabs and weighting targets for all 3 surveys

Click here to view the Nationwide and Arizona survey questionnaire

Click here to view the Pennsylvania survey questionnaire