Credible polling absolutely matters - you just have to know how to use it
Bottom Line: Recently I received a thoughtful note from a listener that raised an important point and topic I haven't addressed in a while. Here's an excerpt:
I am a big fan. One of the things that has bothered me for some time has been the amount of time and resourcing you use citing current polling information. This is not a criticism, just a fan wondering if you might address at some point the validity of polling that occurs in this day and age.
Frankly I just do not trust polling anymore. I have been suspicious for some time, but no I flat out don’t believe them now. This has nothing to do with my political point of view. I frankly don’t think the pollsters can get an accurate amount of valid respondents.
I get it. And by the way I don't simply trust any one poll independent of other information either. The problem with most accredited polling is how the information is used. First, it's important to discern between accredited and non-accredited information. Using non-accredited polling for any meaningful topic is almost always a bad idea. But accredited polling with a lot of additional context can provide a road map along with a lot of other information. As an analyst, I use polls in context with historical information and current conditions to provide story lines and information you won't obtain by looking at polls alone and often that are highly predictive of political outcomes.
Regular listeners during the 2016 election cycle will remember that I was able to use polling information as the base information to be able to accurately project Hillary Clinton to win the popular vote, Donald Trump the Electoral College along with Republican control in Congress. Additionally, I accurately projected every race in Florida I tracked. Every circumstance is different but here's an idea of how I apply this information during election cycles but also a peak into what I do when I'm sharing information with you generally. There are three levels of analysis I apply.
Base polling
History
Current conditions
So, the polls are only a third of the picture and independent of additional information aren't terrible useful to your point. And that's where the hard work also begins. Between cycles I use polling information along with history and current conditions to paint pictures about what's really going on in context - rather than the pictures painted politically in the media. For example, with the weekly midterm elections update I'm been able to provide the analysis that there's a 92% chance Democrats will gain seats and that generic ballot polling has been predictive of what party will fare best in very midterm cycle that's existed.
In the case of some the weekly Tracking Trump updates I've been able to reveal that among likely voters President Trump, as recently as two weeks ago, had the exact same approval rating as President Obama at the same stage of his Presidency and that our country is far more optimistic about our country than under the prior administration. These are all story lines and pieces of information that hopefully are useful and that you wouldn't get elsewhere.