Now That the Primaries Are Over...
When it became apparent that the presidential primary in Indiana was going to matter, it dawned on me that many voters might think this is the norm. For those who have looked at the past or who have lived through it, they know that 2008 and 2016 are not the norm. The presidential primary in Indiana usually includes very little excitement.
The Mike Downs Center for Indiana Politics conducted three polls this season – one statewide Republican primary poll, one statewide Democratic primary poll, and one Republican primary poll in the 3rd district. Polls are not a prediction of the outcome of an election. They are a snapshot in time. Things can change after the poll is taken and a different picture can emerge. Also, the methodologies used by social scientists mean that the results can be wrong one time out of twenty. Finally, primaries in Indiana can be difficult to poll because of the large fluctuations in who actually shows up to vote.
One of our polls showed Jim Banks in the lead and he won. Another showed Todd Young in the lead and he won. Our statewide Democratic poll showed Hillary Clinton in the lead. This is what most of the polls showed. The reality was that Clinton did not need to win Indiana and Bernie Sanders did. He campaigned throughout the state. She made only a few appearances in the state and left the rest of the campaigning to surrogates. In other words, events after the poll probably caused a different picture to emerge.
Our statewide Republican poll showed Ted Cruz in the lead. This was the only publicly released poll that showed Cruz in the lead. We were hesitant to release this result because it differed from the others by quite a bit. We looked through the data and found that many other things about the poll appeared to be aligned with other polls. The percentage of people supporting John Kasich was similar to what other polls found. The reasons people were giving for supporting their favored candidate and the likelihood they would support a different candidate from their party for president were similar to what other polls found. We decided to release the data and try to figure out after the election why our poll was both aligned and not aligned with other polls.
Just about every time we have conducted a poll, we have gotten comments from people who disagreed with our methodology and others who expressed displeasure with the findings. This year, the questions about methodology and expressions of displeasure reached new highs, or lows depending on your perspective.
One of the people criticizing the methodology said that we needed to have a sample size of 3,000 to be legitimate. Another said that our sample needed to be 0.5% of the population of Indiana (approximately 33,000). Both people said they knew about polling. The largest sample of the Indiana polls that were posted on www.realclearpolitics.org was 645. The largest sample in the national polls of Trump versus Clinton this year on www.realclearpolitics.org is 1,574. The sample sizes for the polls on www.pollingreport.com are similar. Our sample size was on the small end, but it was not the smallest one. We stated clearly that our polls had a margin of error of 4.9 points. Moving up to 3,000 or 33,000 respondents would have lowered the margin of error (1.79 and 0.54 respectively). It would have been nice to have the smaller margin of error, but the sizes the people were suggesting are cost prohibitive.
Another caller questioned the wording of the questions. This is a serious conversation to have. Pew Research Center did an experiment in 2000 and found markedly different responses about what to do with the federal budget surplus at the time (see below). We did a series of polls in 2008 that demonstrated how the wording of the horserace question influences the snapshot of who is leading. Our experiment showed that the gubernatorial race that year either was looking like a very close one or a comfortable victory for incumbent Governor Mitch Daniels. We know the wording of questions matters. The wording we used this year for the horserace questions was included in the press releases and is a relatively standard way to ask this question.
Yet another caller suggested that we needed to deliver a campaign message about one of the candidates before asking for whom the respondent was voting. This is something that campaigns do as a way to test a message, but not something that pollsters do when trying to figure out who is ahead when the poll is conducted.
The people who were dissatisfied with the results of the survey were more violent in their expressions. Brian Francisco from the Journal Gazette wrote about the comments he received including a person who hoped someone would “burn in Hell” because of the poll results. Someone sent an e-mail to me hoping that I was, “murdered by an illegal spic or muslim!!” Still another wrote that, “In Indiana you can be shot for declaring falsehoods!”
Finally, there was one voicemail that included this quote, “We’re voting for Donald Trump as an alternative to going to your house and killing you and your family.” That was the longest part of the message without cussing. The next sentence was, “Just thank God that you’re still alive and we haven’t (expletive) killed you.”
I was not the first person to hear things like this during the nominating process and I probably won’t be the last to hear them this year. I applaud passionate interest in campaigns and wish more people would get involved. Also, I personally don’t mind colorful language. The problem comes when disagreements and criticisms move so quickly to threats of violence.
This primary season has not been the norm. Here is to a vibrant, informative, and civil general election.
Andrew Downs is Director of the Mike Downs Center for Indiana Politics at IPFW.
Opinions expressed in this column are those of the individual writer and do not necessarily reflect the opinions of the staff, management or board of Northeast Indiana Public Radio. If you want to join the conversation, head over to our Facebook page and comment on the post featuring this column.