Quote:
Originally Posted by Aery11
If people MUST poll during elections, preferably just ask me who I will vote for. Nothing more. You don't even have to give me a list of candidates. Don't ask me questions about the economy (unless that is all you want to know), don't ask me my age or my income or my sex or my 'ethnicity'. If you need to understand the demographics I will answer some of those questions but I will NEVER tell you my race/ethnicity (nor will I comment about that of others either). I will tell you every time that that question is divisive and unfair to everyone. I won't tell you my income or income source either (even the range) because that introduces bias in my opinion. I will probably always tell you I am female and well into my 60s. My marital status also has nothing to do with my personal opinion - never has, never will. I am my own person. I am also a major skeptic - and I think that is a good thing. I will let you ask me why I am that (skeptic).
|
Thanks for the response Aery11. The reason why the pollsters are asking you that is to identify certain data points used in data science/analytics. As annoying as that might seem those areas are all traditional analytics parameters as they apply to political elections. Ideally, those data points are integrated with the poll results from millions of people, fed into a very powerful computer, and used in effecting a prediction. But like I said, that's with everything being equal.
What I find that perplexing is that this far into the digital age that their is no answer which has emerged from experts other than to say "we just didn't know how the people where actually going to vote". I don't fret at all about who won or lost, what boggles me is that the field of data science is something applied to a vast landscape of areas including sports, climate, healthcare, law enforcement, etc. So if data science "misses the road" in one area (in this case politics), and the fix is not found, chances are it will miss the road again in another area when presented with an identical situation. So now are you seeing my point?
So in the history of political elections this is the first time this type of bias in polling has manifested? Have the experts in that field not identified situations in which there will be bias and designed effective checks against them?
These are question marks that need to be filled or else then the people we watch in suits and ties in CNN and other major outlets who throw numbers in our faces are just "putting on a show"
Speaking from the solutions side of the discussion I will say though, that I had the opportunity to discuss this very subject with a university professor with significant credentials that happens to teach the subject. He said that the root cause of the statistical anomaly was that the wrong type of data gathering (polling people by phone, etc.) was used in obtaining the data. In order to effectively identify the much talked about "silent majority" the best mode of data gathering would have been focus groups instead of the traditional polling.
Hopefully other experts in the field would read this discussion and also chime in because data science is as important of a field as transportation, law, or biology. You can imagine what would happen if a certain medication was applied to a patient to treat a specific ailment without taking into account other peripheral elements such as temperature, past medical history, location, etc. The stakes then would be much higher because it would be a matter of the patient living or "taking the 'ole dirt nap" as some people put it.