April 24, 2015 — CNN, along with their polling partner, ORC International, conducted a nationwide poll of the presidential contest and, as happens from time to time in modern-day national political polling, the result does not likely reflect the state of the actual electorate.
The poll (April 16-19; 1,018 American adults; 435 self-identified Republicans and Independents who lean Republican; 458 self-identified Democrats and Independents who lean Democratic) projects Hillary Clinton to be holding huge leads over the major Republican candidates in hypothetical general election pairings.
In the GOP primary, a very tight race is forecast with former Florida Gov. Jeb Bush leading Wisconsin Gov. Scott Walker by five points, and senators Rand Paul (R-KY) and Marco Rubio (R-FL) together by six. Another 13 candidates were tested, but all of these attracted only single-digit preference. But, what is consistent in all national polls, as was found here, even when Bush leads the pack, he is still generally below 20 percent (in this case, 17 percent). This, for a candidate having virtually universal name identification with the vast majority of respondents expressing an opinion of him.
Proceeding through the election cycle, we are likely to see more of this type of polling, so it is important to pay attention to the methodology behind each study. This particular survey yields flawed information, and for several reasons.
First, as you will notice in the national sample configuration, the pollsters sampled 1,018 adults. This means the respondents were not screened as registered voters. Though the individuals have political opinions and were willing to participate in the survey, their voting pattern, or lack thereof, is not necessarily reflective of how likely voters would perform.
Secondly, on the primary polling question, the respondents are likewise sampled on a national basis. Since there is no such thing as a national primary, how a small sample (about half of the requisite number when breaking down by party) ranks the candidates on ballot test questions is irrelevant because at no time will a vote be taken in this manner. As we all know, the only votes that matter are those within each state. The most significant data derived from early polling comes from the first four voting states, Iowa (caucus), New Hampshire (primary), Nevada (caucus), and South Carolina (primary), and not the country as a whole.
Third, the polling sample is not reflective of the country’s partisan divisions. The CNN-ORC sampling universe contains 27 percent self-identified Democrats, 21 percent who call themselves Republicans, and a whopping 52 percent who say they are Independent or members of a minor political party. According to the most recent Pew Research Center study of partisan affiliation (released in early April, and part of an annual research project that has been ongoing for the past 23 consecutive years), it portrays the most current national party electorate as 32 percent Democrat, 23 percent Republican, and 39 percent Independent. Therefore, the CNN party split is not close to the actual national breakdown.
Given that the CNN sample contains what must be a large segment of non-voters, it is not surprising that both Clinton and Bush are leading their respective primary fields, and that the former Secretary of State and First Lady enjoys such large leads on the general election ballot tests. Since the non-voter, or casual participant, is included in this sample it is more likely that they would be attracted to the most well-known candidates. Again, the presence of these individuals skews the results and deteriorates the reliability factor.
According to the results, Clinton would lead Bush 56-39 percent; Gov. Chris Christie (R-NJ) 58-39 percent; Sen. Rubio 55-41 percent; and former Gov. Mike Huckabee (R-AR) 58-37 percent. With so little variation, the results tell us that this is largely a party preference poll derived from a skewed sample.