Saturday, November 22, 2003

AARP Push Polling

Here's a description of the poll results Novelli was using to claim that 75% of members supported the legislation:

A look at the AARP survey itself (available in a PDF file) reveals that AARP misrepresented the results in its article. Of the 494 members surveyed in what was touted as a “nationally representative sample, with a margin of error plus or minus 4.4 percentage points,” it becomes clear, contrary to AARP’s claims, that those members surveyed did not conclude and could not have concluded that the new Medicare package would in any way help low-income elderly and those with high prescription drug costs.

The survey simply shows that AARP and its pollster worded the question in such a way that it creates the illusion of support. It brings to mind Karl Rove and his Rovian methodology.

First of all, the members polled could not have concluded that the Bill does in fact “help low-income elderly” because 62% of those polled said they were either completely unfamiliar with the Medicare Bill or were not very familiar with the specifics of the Bill. Only 2% felt they were very familiar with the Bill and 35% reported they were merely “somewhat familiar” and 1% refused to answer the question!

In order to ask a question of a group of people who are overwhelmingly ignorant of the subject, the pollsters had to educate the polled members—in other words, the pollster hand fed the answers they wanted to hear.

So the pollsters proceeded to remind those being polled that Medicare does not presently cover prescription drugs. Then the pollster boiled the complex bill down to three sentences. They tossed in another four sentences describing the benefits to the poor and those with high drug costs. Thus the pollsters created the “knowledge-base” for 97% of those who were polled.

Then the pollster asked the following question:

“Even if this plan won’t affect you personally either way, do you think it should be passed so that people with low-incomes or people with high drug costs can be helped?”

Seventy-five percent answered “Yes.”