There is no one who is more skeptical of a political poll than a pollster. Pollsters are well aware of the fact that slight variations in question wording or sequencing, events in the news, and sampling can affect the results of a survey.
Yesterday, I attended a political landscape panel. One of the panelists shared data from recent political polls about the upcoming presidential election and the voters perception of the country. Many of the slides were definitely attention grabbers- the majority of respondents felt that the country was on the wrong track, ratings for Congress and the President were in the toilet, and so on and so on.
Anytime that I see results of a political poll, whether it be in a presentation or a news article, I always examine the poll by asking the following questions:
Yesterday, I attended a political landscape panel. One of the panelists shared data from recent political polls about the upcoming presidential election and the voters perception of the country. Many of the slides were definitely attention grabbers- the majority of respondents felt that the country was on the wrong track, ratings for Congress and the President were in the toilet, and so on and so on.
Anytime that I see results of a political poll, whether it be in a presentation or a news article, I always examine the poll by asking the following questions:
- How many people were included in the sample? What was the precision rate?
- Who was being sampled? Registered voters, likely voters? Was the final sample representative of the population or was the data significantly weighted?
- What about early voters, were they included in the projections?
- Were cell phones included in the sampling frame, or was it strictly landlines?
Most legitimate pollsters who release their data also release their research methodology (or make it available upon request) that answers the above questions. In the absence of this information, my final question is "How accurate are these results, really?".
No comments:
Post a Comment