Should New Hampshire follow Iowa?
By Christopher Larimer
Associate Professor of Political Science, University of Northern Iowa
The ever-expanding GOP field has made for interesting polling. Some candidates fare well Iowa or New Hampshire, but few candidates poll well in both states. Such inconsistency is often presented by the media, pollsters, voters, and even some candidates as evidence that Iowa’s leadoff position in the nomination process skews reality.
But perhaps this is not the fault of Iowa but rather the sequence of the process. Rather than asking, “Should Iowa come first?” perhaps we should be asking: Should New Hampshire come second?
To answer this question, I pooled results from the Iowa Caucuses and New Hampshire primary from 1976-2012 (results are based on caucus results as provided by the Des Moines Register). The percent support (or delegate strength for Democratic candidates in Iowa) for Democratic and Republican candidates in Iowa is matched with his or her support in the subsequent New Hampshire primary.
The results show clear differences by party, and further illustrate the significance of sequencing when it comes to presidential nominating events. (A similar discussion of the results but with more detailed statistical analysis can be found here.)
First, for the Democrats, the results show that percent delegate strength in Iowa on caucus night is a reasonable predictor of percent support in the subsequent New Hampshire primary. Approximately 42 percent of the variation in primary support observed in New Hampshire can be explained by the results from the Iowa Caucuses.
Upon closer examination, however, Iowa Senator Tom Harkin’s candidacy in 1992 stands out as a clear outlier. Recall that Harkin received over 75 percent support in Iowa but only 10 percent support in New Hampshire. Removing 1992 from the analysis, the correlation between Iowa and New Hampshire increases considerably. The revised analysis shows that 71 percent of the variation in New Hampshire support can be explained by what happens in Iowa. In other words, using results from the Iowa Caucuses to predict what will happen in New Hampshire makes sense from a statistical viewpoint. A Democratic candidate who does well in Iowa also does reasonably well in New Hampshire, and similarly for candidates who fare poorly in Iowa.
For the Republicans, the story is a bit different. Looking at results from the Iowa Caucuses and subsequent New Hampshire primaries from 1980 – 2012 (1976 is not included because results from the Iowa Caucuses from that year are not reliable), there is considerably less overlap compared to the Democratic Party.
What happens in Iowa is not a particularly good barometer for what will happen in New Hampshire for Republican presidential candidates. Less than one-fourth of the variation in New Hampshire primary results can be explained by what happens in Iowa. Put another way, if we were trying to predict the results for the New Hampshire primary, using results from the Iowa Caucuses as the only predictor would be a poor choice. Clearly, other factors (potentially that independents can vote in either primary in New Hampshire but must be registered with a particular party to participate in Iowa) are important.
But as with the Democrats, there is an outlier, in this case a single candidate in John McCain who received 5 percent support in Iowa in 2000 and 49 percent support in New Hampshire the following week; a similar pattern is evident for McCain in 2008 when he received 13 percent in Iowa and 37 percent in New Hampshire. Removing those two observations improves the correlation substantially, but even then, the explanatory power of the Republican Iowa Caucuses is just 60 percent of that for the Democratic Iowa Caucuses.
The mismatch between polling in Iowa and polling in New Hampshire for the 2016 field is not an anomaly. Rather, for Republicans, there is a clear historical trend of candidates doing well in Iowa or New Hampshire, but less often in both states, though clearly the inclusion of John McCain in the race affects the results. For Democrats, with the exception of 1992, the sequencing of Iowa and New Hampshire has been smooth. So while the Iowa Caucuses are a good omen for Democrats when it comes to predicting results in New Hampshire, the same cannot be said for Republicans.
Something to keep in mind on February 1, 2016.