Today, we are inundated with presidential polls to assess which candidate is winning and who will become the next occupant of the White House.
Unfortunately, very few are valid barometers as to who is winning, simply because they are not scientific instruments. They are biased and somewhat promotional tactics to generate support for the sponsoring poll developer's respective candidate.
For a poll to be genuinely scientific, specific criteria increase the probability that the survey is valid, measuring which candidate is winning.
Very few polls, especially public polls, possess predictive validity. They are meant to promote a candidate rather than assess accurate, objective information regarding the candidates being measured.
Conversely, private polls that are internal to respective campaigns come closer to validity because not correctly assessing the candidate's strength would simply hurt the candidate by artificially inflating a contender's strength in the marketplace.
That said, here is a quick and dirty way to assess whether the poll results one is reading are valid and a meaningful tool to calculate a candidate's public relations strength.
- Does the poll use registered or likely voters? The promotional polls generally use registered voters. A better metric to use is likely voters. This is more expensive because it takes longer (which means money) to qualify as a likely voter. Validity Index: Use Likely Voters;
- Does the poll use a large enough sample size? Rule of thumb: the larger the sample, the more accurate the finding. A poll should use at least 1,000 respondents. Validity Index: Use at least 1,000 voters;
- Does the poll use an equal number of Democrats and Republicans? Most public polls oversample Democrats. Validity Index: Use equal number of Democrats and Republicans;
- Does the poll use an equal number of Men and Women? Most public polls oversample women. 52 to 48 favoring women is acceptable. Validity Index: Use 52 to 48 women to men;
- Does the poll use an error rate of more than 5 percentage points? A valid survey should have an error rate of no more than plus or minus 3.5 points. This means that if we did the poll 100 times, there would be a five percent chance our result would be higher or lower than 3.5 points. Validity Index: Use an error rate of plus or minus 3.5 points.
In polling, all scientific methodologies have two important constructs for the investigator's consideration. We have addressed validity; does the poll measure what it is supposed to measure?
Is the poll attempting to assess whether the candidate is winning and whether the survey can predict whether the principal under assessment will win a given race? Using the five ways outlined above increases the poll's validity significantly.
The other construct for pollsters is reliability, the ability of the poll to generate consistent results over time. If one considers the current state of this poll cycle, one can conclude that the polls are very reliable, with a difference of plus or minus 1 or 2 points separating the candidates.
The secret in assessing poll validity is understanding and evaluating the bias the poll's sponsor brings.
When one does this, along with following these five assessment tools, one will understand that it is always easier when you have marketing and branding in mind.
Dr. John Tantillo is a Marketing Professor at Lander College For Men at Touro University and the author of the popular book, "People Buy Brands Not Companies." Read More of his Reports - Here.