As the 2023 campaign cycle ramps up, relevant scores will be created and made available to candidate committees. In the interim, the following scores will be accessible for campaigns to start their efforts. As new scores are available, campaign users will be alerted and provided information on these scores (including documentation).


Along with models that are developed nationally, the KDP Data Team will be working on creating and validating models in house for campaign usage. Typically, the fair market value for custom state models is at least $30k-$50k. 


2022 DNC Dem Support (v1 & v2) --- Released 4/1/2022

Modeled by the DNC, this score predicted the likelihood that a voter supports a Democratic candidate in a head-to-head matchup against a Republican in the 2022 General Election. Scores closer to 100 indicates confidence in Democratic support, and scores closer to 0 denote confidence in GOP support. This score does not predict intensity of support, persuadability, or likelihood of identify as an "Independent".


2023 KYDP Gubernatorial Primary Turnout Model (Labeled as 2023: KDP Gubernatorial Primary Turnout in VAN)

Built in house based and trained upon Kentucky specific data, this model seeks to predict the likelihood of a registered voter     turning out in the 2023 Primary Election.

Documentation: https://kydemocratshelp.freshdesk.com/en/support/solutions/articles/44002390744-documentation-2023-kdp-primary-turnout


2023 KYDP Gubernatorial General Turnout Model (Labeled as 2023: KDP  General Turnout [v1 & v2] in VAN)

Built in house based and trained upon Kentucky specific data, this model seeks to predict the likelihood of a registered voter turning out in the 2023 General Election.

Version 1: Trained upon 2019 General vote history, this is the initial version of the model built in 2021 and updated through the end of 2022.

Version 2: A rebuild of the model - and still trained and validated on 2019 General vote history - this version incorporates 2022 General vote history to further inform the model. 

Documentation: https://kydemocratshelp.freshdesk.com/en/support/solutions/articles/44002390745-documentation-2023-kdp-general-turnout


2023 KYDP AVEV Likelihood  Model (Labeled as 2023: KDP  General Turnout [v1 & v2] in VAN)

Built in house based and trained upon Kentucky specific data, this model seeks to predict the likelihood of a registered voter voting absentee (AV) or early (EV) in the 2023 General Election. This model should be used in conjunction with other models such as candidate/party support scores and turnout scores. 

Documentation: https://kydemocratshelp.freshdesk.com/en/support/solutions/articles/44002424040-documentation-2023-avev-likelihood


KYDP Precinct Density Ranks (County and Statewide Versions)

Built in house to combine data from the Kentucky Voter File, voting boundaries and geographical measurements/calculations to: 1) manipulate precinct size and VR total to calculate people per square mile  and 2) standardize precinct density on a 0-1 rank scale.

Documentation: https://kydemocratshelp.freshdesk.com/support/solutions/folders/44001222303


2022 Civis College Graduate

Modeled by Civis, this score predicts the likelihood that a voter has a four-year degree. 


2020 DNC Volunteer Propensity Overall

Modeled by the DNC, this score is designed to identify the most promising targets for volunteer recruitment by Democratic campaigns and party committees. 


2020 TargetSmart Trump Support

Modeled by TargetSmart, this score is designed to target and identify supporter's of Trump's 2020 re-election campaign.   


2020 Civis Kids in Household

2020 Civis Marriage

2022 Civis Spanish Language Preference

Modeled by Civis, this predicts the probability of an individuals probability of preferring to communicate verbally in Spanish on a scale of 0-100. This does not capture ones fluency in Spanish. Use as high of a score as possible, and for planning purposes start with 80+ score.


2018 Civis Race

Modeled by Civis, this predicts an individual’s likely race with the ability to select on LOW,MEDIUM, and HIGH likelihoods. This score is useful for development of targeting, different outreach programs, and message development.


2018 TargetSmart Ethnicity

Modeled by TargetSmart, this predicts a voters’ ethnicity with what they would most likely identify as.  This score is useful for development of targeting, different outreach programs, and message development.


TargetSmart Union Support