84% Hung up on Robo-poll: Another Test Raises Serious Questions
Frequent readers of The Edge know that we take every opportunity to test the claim that robo-polling is on par with surveys conducted by live interviewers. Our latest experiement came on election eve last November. A client wanted to get a quick read on how the national election results might be affecting local turnout. The idea was to make thousands of calls in a very narrow window at very low cost. Enter robo calls.
For this project the client hired professional voiceover talent to record just three simple questions. Of the 2,637 numbers dialed, a human (the targeted voter on the list?) picked up the phone 1,379 times. A penetration rate of slightly more than half is, frankly, not too bad. But here is the big problem: Of those who answered the phone 1,103 hung up before responding to ANY of the questions. That’s 80% who refused to take the survey! Another 60 people responded to only one or two questions so the actual refusal rate was 84%. In comparison, CERC typically gets cooperation from about 80% of the folks it contacts for a 3-question survey. Could a legitimate pollster possibly trust data when more than half the people contacted do not participate? It’s time robo-polling firms start posting their true cooperation rates.