"As I watch President Bush's poll numbers supposedly drop every day, with the media continually bashing him with glee, please let me make four quick observations.
First, by a very high percentage, I assume, the pollsters making these calls are liberals, with the questions slanted in their biased way.
Second, any logic would presume that the pollsters are reaching a heavily skewed audience of Democrats at home. (We Republicans are out making money, and hiring people, to keep the economy going.)
Third, I'm 49 years old and still waiting for a pollster to call me. For some reason I don't think I'll ever get the call.
Last, and most important, if an election were held tomorrow with Bush against any Democrat, GW would win convincingly and overwhelmingly. The American people would go into the voting booth and, through common sense, realize that the economy is growing and in great shape and that President Bush is killing terrorists before they kill us.
It's as simple as that. Period. End of story......
http://www.philly.com/mld/philly/14614602.htm---------------------
This sort of nonsense drives me insane - especially since he's not questioning ONE poll or another (as we often do here on DU, but ALL the polls which show an unpopular President. He's basically questioning the whole concept of survey sampling.
------------
My letter back to him:
I respect your opinion about Mr. Bush expressed in your letter to the Inquirer. But it's hard to understand how a 49 year old adult "businessman" has gotten this far in life and in business without understanding survey research, which is taught to freshmen at every business school in the country.
First, you say that "pollsters making these calls are liberals, with the questions slanted in their biased way". It might be called "CBS News" poll, but the polling is done by huge market research companies run by the same business people who are out making money and hiring people to keep the economy going. The people who actually make the calls don't get to write the questions, so their politics are irrelevant - they read the question off a computer screen, and are monitored to be sure they do not stray from the script. If you read beyond the headlines about the poll numbers, you can often read the question exactly as it was asked, and judge for yourself if it's biased. I do that all the time: when I see a poll, I want to know if respondents were asked, "How would you rate X's performance on a scale of 1-10" or "How awful is X's performance: horrible, bad, or not-so-great" ). In addition to looking at the question, look up who did the poll, and make sure it is a reputable organization such as one of these:
http://www.library.yale.edu/socsci/opinion/pollingorganizations.html. Second, you say "pollster are reaching a heavily skewed audience of Democrats at home". Survey researchers use quotas, which means that they are looking for a representative sample of men and women; blacks, whites and Hispanics, Democrats and Republicans, and any other categories specified by those commissioning the polls (employed full time vs. retired vs. homemaker, income level, level of education complete, etc.) so that the poll reflects the population being surveyed. If Republicans outnumber Democrats in the US by 5:4, then those designing the poll will set the quota so that Republicans outnumber Democrats by 5:4. If, for some reason, pollsters are not able to reach an adequate number of Republicans (since they were out making money), the answers from the Republicans they were able to reach will be weighted more heavily - so they may count 1.5 times and those self-identifying as Democrats (or taken from voter rolls) may count as only .75 or .5 times or whatever it takes to make Republican answers outnumber Democratic answers by exactly 5:4..
Third, you're "49 years old and still waiting for a pollster to call me". Do you know how many people are required for a poll to be valid? Only a little over a thousand, for a national poll - that's 1,000 out of 200,000,000 people. Not high odds that any one person will be called. Years of experience and statistical analysis has shown that once you hit 1000, asking more people does not increase the accuracy of the poll.
As the folks at ABCNew Polls say " If you don't believe in random sampling, next time you go to the doctor for a blood test, have him take it all."Lastly, you're right in implying that survey results CAN be manipulated (that's why I always want to see exactly how the question was asked, and so should you) but the large survey research companies doing this polling stake their reputation on these polls be at least conducted in a statistically valid way.
In order to learn more about the business of survey research, you may want to check out information from the American Statistical Association (not exactly a commie/liberal group) who explain things better than I can at
http://www.whatisasurvey.info. ABCNews Polls has an excellent section explaining polls at
http://abcnews.go.com/sections/politics/DailyNews/POLL_EXPLAINER.html, or read one of the many books available in any college bookstore near you.
And yes, of course, I've worked in this industry since 1985.