More on cell-phone skew

I posted in July about the Pew Research Center’s findings that non-inclusion of cell phone respondents “modestly affects” poll estimates of voting support in the US presidential race

They’ve done another poll with cell respondents included, and updated their findings. Their key finding is unchanged. Including cell respondents makes a little difference–some 3% in the September poll (see table below).

But they are now able to combine the data from the July and September polls to look more closely at the differences between young people responding on land-lines and cells. This is important because one way that the pollsters overcome the problem of excluding cellphone only respondents–as is the case in NZ–is to weight the data. But if they use the young respondents on landlines to weight the data, then they are skewing the estimates to the extent that there is a difference between young land-line and cell phone respondents.

And, yes, there is a considerable difference:

“Among landline respondents under age 30, there is an 18-point gap in party identification – 54% identify or lean Democratic while 36% are Republican. Among the cell-only respondents under age 30, there is a 34-point gap – 62% are Democrats, 28% Republican.”

In the US context this is important to predicting the final result only if the cell and landline respondents have similar turnout. This is unknowable in advance, but the unusual youth of one candidate, his high energy campaigning, unprecedented organisation on the ground, etc, indicate that turnout for the young should be historically high.

In NZ, we don’t know whether there is a similar “youth with cellphone effect”. but if there is, it’s probably not large after the data has been weighted.

Update: The “cell phone effect” was probably part of the reason that the GWU/Battleground tracking poll showed the race tied at the end of September, when other tracking polls showed Obama 5-6% in front.

Update:  Professor of political science at the University of Massachusetts, Brian Schaffner, analysed 2006 Cooperative Congressional Election Study data and says that weighting by party is not likely to fully account for the differences between cell phone only respondents and those with landlines.

Advertisement

Tags: polls