February 10, 2016
I would kill for some data. Nevada is when we really find out whether the Bern is what happens if you hold a lit match for an extra half second or a four mile trip over hot coals. The campaigns know something. We know NOTHING.
Here’s another problem. Even when we start getting polls, there is great potential for noise. Iowa, New Hampshire, and South Carolina have comparatively lengthy histories of mattering.
National pollsters are used to surveying there. Local universities and media outlets have plenty of practice. There’s no Ann Selzer of Nevada.
The Silver State became important in 2008. It wasn’t fully contested on the GOP side in 2008 or 2012, as Mitt Romney and Ron Paul were the only two candidates to make a major push in a more crowded field. President Obama wasn’t opposed in 2012. That gives us a total of one race in one party in one year as a real sample.
To complicate things further, this is a caucus. Like in Iowa, Republicans release raw vote totals, Democrats do not. We get to see delegate equivalents again. Most observers believe Bernie had more voters in Iowa. Hillary won more delegate equivalents.
There’s no polling-to-delegate equivalent conversion metric. Perhaps by 2028 or 2032 enough examples will exist to finesse something like this, but not yet. There are three possible outcomes:
Clear Victory for Hillary: Clinton has a definite majority of actual voters, a majority of state delegate equivalents (the measure the cable networks will use on Caucus Night), and a majority of convention delegates from Nevada.
Clear Victory for Bernie: The same, except in his favor.
2008 Nevada/2016 Iowa: A mess. Back in 2008, Hillary won the state delegate equivalent count, but Obama won more convention delegates due to outperforming in the sparsely populated rural counties. He got to claim a victory of sorts. We know what just happened in Iowa.
So with bad tools, we need to use polling data to try to determine which of those three outcomes, each of which have a very different meaning going forward, are more likely.
It gets worse. The GOP primary in South Carolina, sharing a 2/20 voting date with the Democrats in Nevada, has three distinct polling cycles. A full packet of surveys will hit before the debate, making it possible to see if the event made a difference.
Not here. Bernie and Hillary debate Thursday night. You’ll get a couple of surveys that were mostly taken before they met, but most of what you see over the weekend will include before and after data. Tracking polls were very useful in New Hampshire. Don’t expect to see as many (if any) in Nevada.
In Iowa, there was a wide range in how various pollsters saw the race. Most were consistently pro-Hillary or pro-Bernie. Because they surveyed the Hawkeye State for months, we were able to determine how the candidates were doing by seeing how well they progressed against their previous results with the same pollster.
Not an option here. Even in New Hampshire, some pollsters had Bernie up by 20 to 30 points, others had the race in single digits. Again, there was enough repetition to play the Iowa game. It’s worth noting that in both states, the truth was about 75% of the way to the Bernie-favorable polls.
If the same is true in Nevada, I’d be comfortable making the appropriate adjustment, but reaching conclusions based on two data points is dangerous. However, all things being equal, it’s worth giving Bernie the benefit of the doubt. He’s still the insurgent challenger, and undecided voters traditionally break that direction.
My complaining aside, Nevada will provide us a few decent clues over the next week. Here they are:
Percentage of Hillary’s “definite” voters: There are New Hampshire voters who were definite for Hillary 9 months ago and voted for Bernie yesterday. He doesn’t have that kind of time here. If polls show her up 15 points, with 80% definite voters, Bernie will need huge turnout. A seven point advantage with 70% certainty and it’s another story.
Percentage of undecided voters: Some pollsters force voters to choose. In those surveys, see the above. Others allow for an undecided option. If you see Hillary ahead 48/38, Bernie is in good shape. In order to win, he needs to have a fair amount of voters at least open to him. That’s good enough.
Bernie’s favorability ratings among non-white voters: If you see Bernie running anywhere near even with Hillary among Latino voters in Nevada, she’s cooked. More likely is him finding a batch of voters who are at least willing to give him plenty of consideration over the next week.
The first indication of that is favorability numbers. Though Bernie’s numbers are weaker than Hillary’s, few non-white voters have an unfavorable view. The problem is many were unsure/didn’t have enough info. In an October or December survey, no big deal.
With limited time left, Bernie is unlikely to convert unsure favorability to caucus votes. His opportunity for expansion are voters who like both Hillary and Bernie but favor her for now.
Hillary’s favorability ratings among younger voters: Same thing. We know Hillary does better with voters of color, Bernie with younger voters. If someone has an unfavorable view of Hillary today, they aren’t caucusing for her in just over a week.
If they do like her, she has a chance to pull from his strength. We still don’t know what younger, non-white voters will wind up doing.
If they act like their age cohort, Hillary will lose. If they act like their ethnic cohort, Bernie will lose. Favorability gaps are the best way to see if voters are willing to consider both.
Sample breakdown comparisons between surveys kindest to Bernie and Hillary: Wide differences in poll result are usually due to disparity in who is being sampled.
Internet surveys and those which don’t screen ahead for voter registration require a large turnout to wind up accurate. So far, voters have shown up and made these look decently on point.
A pollster also needs to estimate the ratio of each demographic group to attempt to match the breakdown to the actual voter group. If one pollster figures on a higher percentage of young voters, it favors Bernie.
A higher percentage of Hispanic voters favors Hillary. You get the idea.
Not every pollster will show most of the above, so find your breadcrumbs where you can.
Nevada shapes up as the toughest polling call of the campaign. Take out your Ouija board.