Recently, one of the MSM outlets had the posted results of African-American voters in selected states - it showed a range of zero percent to 10 percent support for Donald Trump in various battleground states. Zero in one state with an adjacent state at 10 percent was one situation. Something in my brain went off multiple times and so I decided to share it with Steemit's intelligent readers.
The first thing to say is having Trump in the race, with his rhetoric, and Hillary's ads of Trumps own words over and over, has fueled all eyes to racial identification this election cycle. The MSM has made it quite obvious that for Trump to win he has to have a huge Caucasian turnout (whites) to offset his lack of appeal to voters of color. The opposite is also true of course - the Democrats to win need minorities to respond as they did for Obama's run in 08 ad 12. And, all that focus on race, most likely means that some folks are very sensitive about saying that they support Trump at all. Almost obviously, right?
So - let's take you inside one of those moments in the polling - you're an African American - the phone rings and you pick it up - it's one of those important Presidential polls (and suddenly you feel important) and the nice researcher on the phone, who sounds African American to you, asks you the big question about who you are supporting this time around in 2016. Or, the same situation - an African-American phone researcher is asking you - a Caucasian - about who you support this time in 2016 - do you think that there might be an impact on the answer? As a researcher I can tell you that the answer is - for sure.
The opposite of the above is true too - a Caucasian researcher - asking the same questions - will get different %'s than the black researcher. Does it all balance out you ask? To a degree, but only to a degree - and one reason is generally the phone banks of callers (not a highly paid job and almost always done by part time workers) doing the research are indeed over-represented by minorities. People tend to answer in the way that they expect the researcher asking the questions would like them to answer. Just the way it is. The end result IMO - is going to be an under-representation of Trump answers in the polling.
But, there's more to it than that when one state has 10% black support for Trump and another similar demographic state has zero percent - and that is that the range of the confidence of answers within minority segments of the population are generally HUGE compared to the overall total sample. The plus or minus range on any answer is much larger within smaller sub segments of the population (non white) and can easily be 6% or so with a sub-group like African-Americans. And yes, that means the 0% answer is technically the SAME as the 10% answer. As you can see, the MSM is placing lots of emphasis being on answers that one simply can't pin down in randomize research.
And, I won't even get into the difficulties of sampling in this day and age to produce a valid random sample. Indeed, most samples end up getting `weighted' to balance out the mis-representation of the final collected sample - so that it fits the real demographic sample outcome. Therefore, it's easily possible, that an under-represented sample to begin (at the end of sampling) will be boosted so that only a very few 'votes/opinions' may be projected into something very unrepresentative. Additionally, the demographic differences in folks with land lines and cellphones is significant (as will be the time of day that the contact is initiated) - which all makes for some very squishy answers for a poll done at one moment of time.
Finally, as you can hear by the talking heads who read the results on the MSM to you, - often these polls DON'T EVEN INCLUDE WHO WILL BE ON THE BALLOT AT THE TIME OF VOTING - meaning, yes, that some of these polls are leaving out the 3rd party candidates who will be on the ballot.
Bottom line - don't go to far down the rabbit hole on polls in a race that has such racial overtones.
Over all, it mostly doesn't matter. There aren't enough white voters in enough states in the electoral college to get Trump elected. He’d have to win every state that Romney won and then some—and that's unlikely to happen.
Yes, Hillary's lead is shrinking, but she just needs to win the traditional blue states, which gives her over 200 electoral votes and win a couple of battleground states and she gets the 270 electoral votes she needs to win.
According to Five ThirtyEight, which correctly predicted all 50 state outcomes in 2012 and 49 out of 50 in 2008, Clinton has a 68.2% chance of winning the presidency.
All of which misses my entire point.