Developed a model identifying potential Donald Trump supporters who had not previously voted in a Republican primary. Logistic regression identifying 142,000 individuals was developed from polling data. The model predicted a 72 percent propensity to support Trump. Test Universe received a total of seven contacts (5 email 1 phone call and 1 text message).
By focusing on a modeled universe; not only did the test generate first time primary voters for the desired candidate; but it also was done at lest than half the cost of the next lowest expenditure per vote.