HaystaqDNA/L2 U.S. Voters Predictive Analytics models (Sample 10k/209M)

The Haystaq/L2 models provide users with over 120+ issue, behavioral, and candidate support scores (from 0-100) and flags (top 10%-15 of the model) from which users can select. These scores and flags include some of the following categories:

Behavior: activism, ballot drop off, political/charitable giving, engagement with postal mail, text messaging, political calls, ticket splitters, work from home, and church attendance/religiosity.

Candidates/Officials: Biden, Harris, Trump, Bernie Sanders, representative Alexandria Ocasio-Cortez (AOC), Mike Pence, and others.

Consumer Focus: Price versus environment, movie viewers, Rideshare users, TV viewership (cord-cutters versus cable users)

COVID: Vaccines, China involvement, continuing restrictions, and telemedicine

Economy: Housing, worker treatment, unions, build back better, capitalism, crypto, economic anxiety, gentrification, gig worker, income inequality, inflation, infrastructure, jobs, minimum wage $15, monopolies, Social Security, student debt, taxes, tobacco taxes, CEO trust, Universal basic income

Education: Charter Schools, Common Core, Community College for free, school boards, teachers' unions

Environment: climate change, gas tax, Green New Deal, pipeline

Ideology: General ideology, fiscal ideology, social ideology, third party moderate, partisanship, republican/democrat moderates versus ideologues

Media: Social media trust versus Free Speech, social media usage, TV cable news sources (Fox, MSNBC, CNN)

Military/International: Biden foreign policy, defense spending, immigration, Ukraine
Hot Topics: Black Lives Matter, critical race theory, drug prices, January 6th, QAnon, right-wing malicious, gay marriage, trans athletes, government Trust, violent crime

Role of Government: Casinos, civil liberties, death penalty, gun control, marijuana, online gaming /gambling, public transit funding

Trump: Presidential primary support, Republican and Democrat potential supporters, Trump V Biden matchup

Turn Out: Early / Absentee voter, midterm voter, presidential voter

Voting & Courts: DC statehood, early vote, Electoral College, felon voting, Puerto Rico statehood, Rank Choice Voting, redistricting, Supreme Court, voter fraud

These Haystaq/L2 models are updated one to two times per calendar year and are built off of L2’s best-in-class national voter file. Utilizing tens of thousands of survey responses through mixed modal research, including text messages, calls, emails, as well as panel-based information HaystaqDNA matches these response data back to the L2 National voter file with over 750 fields of voter, demographic, consumer, contribution, and other information.

The indicator data used by these models is a combination of individual-level voter registration information and past voting behavior individual and household demographic and consumer characteristics Census and Geographic data validation.

For all models, Haystaq excludes a random 1/3 sample of the survey data from the modeling process as a test set to use for validation.

Tables:
-Flags
-Scores

搜集汇总

背景与挑战

背景概述

该数据集基于L2全国选民档案，提供超过120项选民行为和候选支持评分及标志，涵盖政治、经济、社会等多个主题。模型每年更新一至两次，通过混合模式调研数据构建，并采用随机抽样进行验证。