Electric Insights
EXPLANATORY POLLING · JFK CASE STUDY

Polls tell you what people think.
This shows you why.

A framework for turning survey data into explanations — not just numbers. Built on Harris’s 1963 Newsweek survey, open to anyone.

57%

Original JFK Approval

30pt

Maximum Swing Potential

6

Key Drivers Identified

7

Interactive Analysis Tools

What’s Wrong with How Polls Are Reported

Most polls tell you the headline number but not what’s driving it. That’s the gap this work addresses.

The Standard Model

  • Headline percentages only
  • No explanation of what drives the number
  • Static — no way to ask “what if?”

The Electric Insights Approach

  • Model what drives the outcome
  • Show effects in plain percentage points
  • Let anyone test “what if?” interactively

The Goal

  • Open methods, open data
  • Any poll can ship with a browser-based model
  • Public can verify the numbers themselves

How It Works in Practice

Four steps from raw survey data to something anyone can use

1

Ask the Right Questions

Design surveys to capture not just what people think, but the issue-level evaluations that explain why they approve or disapprove.

2

Speak in Plain Numbers

Report effects as percentage-point shifts — not regression coefficients. If Khrushchev handling moves approval by 30 points, say that.

3

Let People Test “What If?”

Ship a browser-based simulator alongside the topline numbers. Anyone — journalist, researcher, or curious reader — can shift the dials and see what changes.

4

Show Your Work

Publish the model, the data, and the code. Independent verification shouldn’t require a data request — it should be one click away.

Case Study: JFK Approval

Our reanalysis of Louis Harris's survey for Newsweek just prior to John F. Kennedy's assassination demonstrates how explanatory modeling methods identify the drivers behind critical outcomes like approval.

Key Findings:

  • Economic perceptions: 24-point impact on approval
  • Khrushchev handling: 30-point approval swing
  • Civil rights opposition: 11-point reduction

Simulation Preview

What if all likely voters (rather than 12%) believed President Kennedy excelled at keeping the economy healthy?

57%
Original Approval
68%
Simulated Approval
Launch All-or-Nothing Simulator

Five Ways to Explore the Data

Each tool asks a different question about the November 1963 Harris–Newsweek survey.

All-or-Nothing Simulator

What if every respondent felt the same way about an issue? Move one variable to an extreme and see how much it would have shifted Kennedy’s approval.

Try it

Fine-Tuning Simulator

Instead of extremes, shift the mix gradually — fewer people rating Vietnam as “Poor,” more rating the economy as “Excellent.” Watch how small distributional changes move the approval needle.

Try it

Non-Response Test

Not everyone answers a survey. Could the people who didn’t respond have changed the headline result? Set how many non-respondents to assume and what they would have said.

Try it

Survey Explorer

Browse the raw responses. See how Americans answered each question, which questions moved together, and how one group compared to another.

Try it

Model Builder

Pick survey questions, build a logistic regression model, and see which variables actually predicted approval. Test whether the published model holds up — or whether a different set of questions fits better.

Try it
All tools use the original 1963 Harris/Newsweek survey of 1,283 likely voters

See the Framework in Action

The JFK case study is live and free. Explore the data, run the models, and see what explanatory polling looks like in practice.

Get In Touch

Interested in applying this framework to a current survey? Want to discuss the methodology? Reach out directly.

Location

Las Vegas, NV