(Community Matters) Interesting to read about Gallup’s review, learning how they got the presidential race so wrong. They’re publicly releasing their report next month but some of the preview includes:
- how they ask about race makes a difference – appears yes or no doesn’t work as well as allowing respondents to choose more than one answer (& since they extrapolate and model sample results in part based on race, this can significantly impact modeled results)
- In the future traditional polling could be supplemented by non-probability data, administrative data, social media data.
Community Matters 