I don’t know about you but I’m already sick of the 2020 election and it’s still almost a year and a half away. Granted, the coverage of the Democratic primaries is probably slightly more relevant ... but the first voting doesn’t even take place for at least six months. And how accurate are these polls this far out?
It may seem odd that a market researcher is telling you to discount ... well, market research. But political polling is somewhat different than market research. Additionally, political polling is meant to capture a decision a voter would make at a future point in time whereas market research is really meant to understand trends and attitudes over a longer period of time.
Here is what you need to take into consideration as you listen to the polling numbers you hear via the mainstream media. First, the weighting of the data has a significant influence on the poll (and may have been the cause of the Hillary Clinton polling disaster) and secondly, market research and polling are currently in transition as polls move from telephone calls to mobile technology.
What is weighting? Weighting is used to adjust the results of a study to bring them more in line with what is known about a population. For example, according to MarketResearch.org, if a sample contains 40% males and the population contains 49% males, weighting is used to correct the discrepancy. Traditionally in political polling it has been presumed that the U.S. population has more Democrats than Republicans and even fewer Independents. Therefore, when you look at a political poll, in addition to the Margin of Error (which can range from 2 to 4 percent) you should also look for the weighting. The weighting is generally not shared by the media and you need to look at the source document or other materials provided by the organization completing the poll. In 2016, and even now, you would find that polls are weighted to favor the Democrats. You might see a +4 Democrat or I have even seen some polls with a +9 Democrat. As your population gets larger, your weighting might get trickier. For example, weighting at a state or county level may be more accurate than at a national level.
For example, as I mentioned, national polls consistently weight towards Democrats. But Spectrem’s unweighted demographics show that the theory that there are more Democrats than Republicans nationally simply isn’t true. As you can see below, there are more Republicans and Independents than there are Democrats, both today and in 2015.
Clearly the Independents are a critical group who will determine the outcome. Additionally, our data is not necessarily reflective of the general population because it only includes those with a net worth of $500,000 or higher ... arguably this is Mainstream America and higher. Spectrem does not do political polling. We simply ask their political affiliation to determine if it makes a difference in their investment philosophy (it doesn’t.)
Again, I also want to mention the transition of research as mobile polling becomes more popular it is even harder to determine how to properly weight information or even to ensure demographics of the responder. I should also mention that there is always a bias in polls because some individuals simply will not respond to polls. Other personality types do respond. There was a belief that Trump voters do not respond to polls ... but who knows?
Clearly it is really difficult to rely on the polling that is occurring today. While some of it makes absolute sense, always look a little deeper – what is the weighting? Is it a national or state poll? Keep in mind that, especially now, polls are probably weighted heavily Democratic because the goal of most of the polls is to identify the Democratic primary leaders. As the primaries get closer, the polling will become more reliable, but always think about the weighting, the timing, and the size of the poll.
Hope this helps to make sense of what you are hearing. Wouldn’t it be great if politicians could only start campaigning two months before a primary? Not probable but wishful thinking.
© 2019 Spectrem Group