Examining Nevada's poverty of reliable polling: What about cellphones?

One of the perplexities of the ongoing campaign season is why a state that is likely to play such a significant role in 2008 has such a poverty of reliable polling data. Putting aside the obvious fact that Nevada has a poverty of a lot of things, a constant refrain of many trying to get up to speed about the state's politics is why the dearth of polls in the Silver State?

Except for the occasional willingness of campaigns to release their polls (which coincidentally enough tend to favor their side; see Titus, Dina for recent examples), statewide polling for public consumption comes from two sources: Intermittent releases by the Las Vegas Review Journal and the Reno Gazette Journal and national operations (some of them internet based) that occasionally dabble in the state. As consumers of these polls know, they tend not to be all that good.

Two of the primary issues associated with polling in Nevada are the sporadic nature by which the polls are conducted and virtually all of the polling is done on the cheap, yielding small sample sizes and large margins of error. Moreover, the lack of an in-state, ongoing operation such as the California Field Poll or the Quinnipiac University Polling Institute makes comparisons of polls over time risky, given the differences in methodologies employed by differing polling outfits.

Equally problematic are changes to the counters of the state's political landscape. Commercial polling operations work off of sample frames that to be representative requires constant updating in changing political environments. Given that the firms polling in Nevada are profit driven with lots of clients, their ability to keep up with what has happened in the state since January might not be what it should. The end result is that the folks who end up in these samples tend to lean towards the old and reliable. In this case of Nevada, this means that those more likely to be sampled are those who regularly vote, hold stable addresses, and rely on landlines (e.g., rural voters who overwhelmingly favor the GOP). 

More importantly, the dynamics that have allowed the Democrats to turn a statewide registration difference from minus 4,000 to plus 76,000 are not going to be fully captured in these sample frames.  In Clark County alone, the Democrats have more than doubled their forty thousand registration advantage. Some in this crowd are young, first time, cell phone only voters; a group that is likely to significantly increase its turnout from four years ago and go heavily for Obama. And while polls typically use weighting to adjust for demographic differences between landline and cell phone respondents, doing so assumes that those who are reached over landlines are similar to cellphone only voters; a questionable assumption at best. Thus, it is not surprising that many of the most recent Nevada polls show John McCain with a small, albeit statistically insignificant lead over Barack Obama.

A recent study released by the Pew provides some additional insight into these considerations.  Since the end of the primary season, the Pew Research Center for the People & the Press has conducted three national surveys (the end of June, the end of July through the start of August, and the middle of September) that include both cell phone only and landline samples. In the last two surveys, McCain runs about even in the landline only samples, closing a 5- to 6-point Obama advantage from June. As expected, across all three time periods Obama dominates the cellphone only samples to the tune of 19 percent in the latest round. When the landline and cell phone only samples are combined, Obama's support increases and McCain's support decreases by a few points. These difference, however, are not statistically significant from the landline only samples.

When the data is parsed more finely to assess age differences between cellphone only and landline respondents, the expected differences between Obama and McCain's support emerge more clearly. Among the youngest voters (those 18 to 29), Obama's support increases and McCain's support decreases by six percent when the cell phone only respondents are pooled with the landline respondents. Across the other three age groupings that the Pew examined (30 to 49, 50 to 64, and 65 and above), the differences between the cell phone only samples and the pooled samples are negligible.  

What does all of this mean for Election Day in Nevada?  Obviously, one cannot make inferences from national samples to individual states. Thus, while the differences between landline and cell phone only voters are instructive, what the Pew surveys tell us about the preferences of the national electorate may not hold in Nevada. Moreover, given that many of the cell phone only voters are new to the process; these voters do not have the long voting histories of their landline only counterparts -- a significant consideration that both pollsters and campaigns consider when gauging support.

At the same time, the Pew found that both groups of voters are equally engaged in the 2008 campaign and both groups express the same intent to vote in November. Anecdotally, there are huge differences in the enthusiasm between McCain and Obama supporters on campus. 
The Obama campaign has consistently had volunteers on campus, manning registration booths on a near daily basis, while the McCain effort is virtually non-existent. In classes, there appears to be an even greater level of support difference between the two candidates. Thus, if 2008 turns out to be a mobilizing election in Nevada, then the decision of the Republicans to try to squeeze out one more win from the old playbook may leave them flat-footed and on the losing end come Nov. 4, particularly if the economy continues to dominate the agenda, minimizing the ability of the GOP to use cultural issues to peel off working class whites.

David Damore is a political scientist at the University of Nevada at Las Vegas.