The Cellphone Effect

When all votes are counted, one side is dancing with euphoria and the other screaming in the streets (or drinking heavily in a dark room back home), my bet is one real winner out of this election will be Nate Silver and the crew at FiveThirtyEight.com. They have taken the science of statistics and provided a fresh and (arguably) even look at the true state of the presidential race by aggregating, weighting, and analyzing state and national polling data with remarkable detail. Others do something similar (Real Clear Politics and Pollster.com) but 538’s plain English walk-throughs of the data win hands down. I can’t look at any single poll reported in the media without immediately seeing how it fares under 538’s analysis.

It’s become a must read for me, and I’m very curious what the post-election period will bring for the site.

Some of my favorite posts involve the “let’s tackle a ‘conventional wisdom’” angle, two big ones being the possible impact of any Bradley Effect and the so-called Cellphone Effect. Nate has a fascinating look at the cellphone effect up today (with an update) looking at the Obama leads in national tracking polls, and how much more he leads among polls that bother to call cell numbers:

(linked image found at FiveThirtyEight.com)

This has got to become a much more visible issue in future national elections. The venerable land line is fast becoming irrelevant for many, especially younger, people. I have one, but I use it almost exclusively the way I use my @yahoo.com email address – as a SPAM catcher. It’s the # I list when replying to some marketing promotion, buying something online, etc., as I know that’s the # that will (and does in fact) get flooded by telemarketers, and I assume lately, pollsters.

Thanks to caller ID, I just don’t answer any unknown # coming into that line, though often I will pick up unknowns coming into my cell (they have that #, I must know them somehow…right?). In effect, I’m one of this crowd in the cellphone effect Nate mentions – I could only be polled if someone called my cell. And for many who ONLY have a cell, the case is more slam dunk.

I’m curious how dramatic this effect will be four, eight, or twelve years from now. Will pollsters have adjusted their models enough to capture responses from the cell-only (or cell-preferred, in my case) crowd?

I suspect a lot will depend on where the actual election results fall in that chart above.