When viewing all political races in Minnesota we have a Checks & Balances Rule of Thumb, the DFL base is 43% and the Republican base is 37% with 20% Independent/Undecided. In each contest, we analyze we measure each candidacy thus. This means if a candidate falls below the baseline there are showing weak performance, or if the exceed then it is a strong performance. The ability to appeal to an amount higher than their respective party’ baseline, is often easier when there isn’t a viable third-party candidate in the race.
The KSTP/Survey USA Poll and the MPR News and Star Tribune Minnesota Poll are far different creatures. and we place more stock in one than the other. This is not our first challenge to the KSTP/Survey USA Poll. In 2016, we did the same, we questioned the validity of the poll because of its small sample size and lack of expression of its scientific methodology. Unlike the MPR News and Star Tribune Minnesota Poll, which through American Public Media Research Lab, expresses the entirety of the poll.
Our criticism resulted in KSTP’s Tom Hauser questioning our understanding of polling, and we are quite confident we understand polling. As has been often the case, critics of KSTP decry it as the “conservative” station, our equivalent to Fox News, in spite of there also being Fox affiliate present in this market. Frankly, the questions often asked in polls done by media are intended provide information for news stories and ask we can see in both the KSTP/Survey USA Poll and the MPR News and Star Tribune Minnesota Poll each has generated a number of stories from one poll by each entity.
The sharpest criticism of the KSTP/Survey USA Poll is again its small sample size 574 respondents of likely voters compared to 800 in the MPR News and Star Tribune Minnesota Poll. But we are not clear what the true numbers are in the KSTP/Survey USA Poll it was first reported as 574 respondents, but there also is a story on the same Poll saying, 587 voters. The definition of “likely” is normally determined by a screening question, like, “Do you intend on voting in the November election?” When the question elicits an affirmative response then the questioner proceeds to ask the additional questions. Now, the human effect can have influence, because while a person may make a response to an automated poll, which may not be completely true, people are less inclined to misrepresent themselves to a live caller. It is generally human nature to offer up the truth when asked rather than lie unless the person is pathological.
Everyone in politics knows a poll is only valuable when you can see the crosstabs. The MPR News and Star Tribune Minnesota Poll provides them, but the KSTP/Survey USA Poll does not. So, the method of how the questions were asked in the KSTP/Survey USA Poll has not been fully explained nor have the type of device the respondents used. In this day of cell phones, the question of how were the respondents contacted. If it was strictly through landlines, then likely the respondents were older, because it is mainly older people who still have landlines. If it is a broader distribution with a balance between cell phones and landlines, then it was a far costlier poll, because those lists are more expensive rather than just a list from the Secretary of State’s Office, which may also be why only 457 people were contacted. Unlike the KSTP/Survey USA Poll, the MPR News and Star Tribune Minnesota Poll does provide a breakdown of the different devices. There were 800 total calls with 483 to landlines and 317 to cell phones. Now we wonder what the results would have show if only the 483 landline calls were viewed.
The issue over the sample size is also important because of the margin of error. The purpose of a poll is to determine by random sample how closely a smaller group aligns with the larger population. In the KSTP/Survey USA Poll because of the small sample size, it is +/-4.5%. This means every response can have a swing of up to 9-points, this is a very large differential, is important when determining confidence in the poll result. Comparatively, the MPR News and Star Tribune Minnesota Poll margin of error is +/- 3.5% which is a significantly better confidence figure.