Push Polls Are Very Rare

Dear Editor,
    I don't have a dog in the fight for Travis County judge, but Michael King's shoddy journalism ["Brown vs. Eckhardt vs. … McCombs?" Newsdesk, Jan. 18] needs to be addressed.
    For the record, I've worked directly on many campaigns and have consulted on dozens, from local to federal, and I worked with pollsters on all of them. What Mr. King fails to understand about legitimate polling is that the motive for conducting – and paying for – a poll is less about getting a horse-race number and more about testing messages. Those messages include positives on the candidate conducting the poll ("Would you be more likely to vote for Ms. Eckhardt if you knew she …?") and negatives on the opponent, to see what moves likely voters, or voters in targeted demographics. If it turns out that a certain position the opponent took moves more likely voters (or moderate women, aged 40-55), then that issue might be used in direct mail, TV ads, voter-to-voter communications, etc.
    It's understandable that a private citizen receiving a call where they hear negative questions about the opponent would assume that it's a push poll. But for a push poll to work, it would need to be blasted out to enough of the voter universe (perhaps a majority) to influence the entire electorate, rather than just a random sampling. Mr. King presents no evidence that the poll was received by an abnormally large number of Austinites.
    Mr. King displays a fundamental misunderstanding of polling, which is surprising and disappointing for someone who has been covering local politics as long as he has. As I said, a private citizen can be forgiven for failing to grasp the distinction between a (rare) push poll and a legitimate and common poll of likely voters that tests negative messages. But a journalist is subject to higher standards. Had Mr. King asked any political consultant in Austin not connected to either candidate, they would have quickly explained why this call is not a push poll. Instead, even after updating his post, he resorted to snark that doubled down on his error, rather than trying to confirm facts.

   [Michael King responds: Based on his stated experience, Drew Dupuy should presumably know that campaigns do not normally release the details of their own polling, neither the target demographic nor the presumed numbers of people polled. In this case, the Eckhardt campaign declined even to confirm the identity of the pollsters operating its computerized poll. By Dupuy's definition, if a campaign claims there was no push poll, there was no push poll. In this instance, I'll believe my own ears.]