Phone App Predicts Trump Win

I stumbled upon this article and was intrigued.  I had never heard of the Zip app and thought a voluntarily-installed app created a self-selection that should invalidate the poll, more so when you consider that it claims answers are anonymous.  Still, I was curious and decided to download the app and swipe through the questions.  Quite a few of them did determine valid factors about who was answering.  There was a question about black or white (92% of respondents were white), male or female (56% male), over 50 or under (48% were over), Conservative or Liberal (85% conservative).  Yikes, that last one is a real killer.  Of course, I understand that one can weight the numbers and get interesting data but I'm still skeptical.

Having played with the app and gone through 40 or 50 questions, the answers of which leaned heavily conservative, it was time to check Google.  A search offered links to several stories that date back months.  There was one that claimed to predict the victor for each party of the New Hampshire primary and another predicting Super Tuesday results.  Still, not impressed.  When it proves right when a standard poll proves wrong, then I'll give it some credit.

 

The big selling point is the anonymity aspect, which should minimize or eliminate the Bradley Effect in the responses.  Bradley, a long-serving mayor of Los Angeles, ran for governor twice.  Though the polls showed him leading, he lost.  The theory goes that voters didn't want to admit they were going to vote against the black candidate and thereby be labeled a racist.  It has been thirty years since Tom Bradley lost the race in California and I suspect that pollsters have mostly determined how to overcome the effect.  Obama polled like he was going to win and then won.  However, if Zip app proves correct, will there then be a Clinton Effect regarding voters saying they will vote for a woman for president but then choosing the male candidate in the privacy of the voting booth?