Electric Insights: Simulation, Prediction, Forecasting, Optimization
Overview & Tips
    Crux reported in November 2021 that 17% of US adults (45 of 260m) trust pollsters/polling organizations. (That means a randomly selected adult's probability was .17.) Using Crux survey data, we built a model to predict trust then packaged it in this simulator to help interested parties figure out how to increase the 17% (.17) and 45m numbers:
  • Tip 1. If you select hold constant, you retain the response choices of every Crux interviewee (n=1,198) for that question. So if you select hold constant for every question, the predicted probability of trusting pollsters/polling organizations still will be .17.
  • Tip 2. If you select different values, the simulator assumes every interviewee chose them, too.
  • Tip 3. Be careful with socio-demographic variables (in purple). See advice on applying probabilities.

 Trust in Pollsters "All or Nothing" Simulator
  Use the fine-tuner to test more-nuanced scenarios.

Crux p: 0.17, Pop: 45 of 260m

Applying Probabilities

Potentially, you can apply the same predicted probability to three groups:

  1. All Adults. If you see a way for your selections to apply to all 260m adults, you can regard those values as targets to hit to increase (or decrease) the 17% (.17) and 45m numbers. To ensure they apply, hold constant every socio-demographic variable.
  2. The Target. If you select values that can't apply to all adults (e.g., everyone can't be 18-34), no worries. The target group is limited to those with your socio-demographic selections. So if you choose Age ("18-34"), Urbanicity ("Suburban"), and Education ("BA/BS+"), the result will apply to 7m adults.
  3. Specific Individuals. If the values you choose are your own, or those of someone you had in mind, the resulting estimate will be your, or their, predicted probability.

Simulation Possibilities

You can simulate thousands of possibilities to answer questions like this:

  • If all 260m adults, rather than the 3% (8m) Crux reported, felt "extremely confident" that pollsters will correctly predict the next president (and nothing else changed), what impact would that make on the percentage (17%) and number (145m) of adults who trust pollsters/polling organizations? [The figures would increase to 26% and 67m.]
  • What's my probability of trusting pollsters/polling organizations? [Enter your values to see for yourself.]
See our socio-demographics FAQ See our research design FAQ See our crosstabulation FAQ