Data Outliers in Iowa Prove Just That
Social Intelligence Data Slightly Off on Ramaswamy & Haley Numbers, Accurate on Trump and DeSantis
On Saturday, Jan. 13th we authored an analysis of Iowa data from our social intelligence monitoring system ahead of Monday nights caucus.
Our analysis focused on an outlier around Vivek Ramaswamy in the system vs. his traditional polling average. The data showed Ramaswamy with a Support Index (methodology here) between 17% & 20% — about 10%-13% above his final polling average.
The Support Index also showed Nikki Halley at 12%, below her final polling average by about 7%.
Final polling averages below here:
As we noted Saturday, and as we note in all our work with this type of data, a key factor is never assuming it always correct/incorrect vs. polling — but more hunting for trend-lines that either validate or expose outliers to examine further as part of an overall measurement mix.
In this case it is clear that social data inflated Vivek’s position and slightly diminished Haley’s — which actually presents an entirely different and interesting track worth discussing around why. But that is for another day.
Final Iowa results below:
Since the social data Support Index is based on 4 day rolling averages, below is a capture of what the data looked like last night as voting began — slightly different than the data from Saturday’s post.
Trump and DeSantis numbers line up nicely to actual results (Trump 51% to 51%, DeSantis 21.2% to 19%).
Vivek was 9% higher vs. actuals and Nikki was 7% lower in the social measurement.
Overall, traditional polling averages proved very accurate in Iowa with the biggest outlier being DeSantis over performing his average by about 6%.
Now, as they say, on to New Hampshire!
