Between June 12 and 16 the New York Times and CBS News did a poll showing the level of support for President Obama's health care plan. The results showed a very strong 72% support. That makes for a pretty good story, doesn't it?
But suppose support wasn't really that strong? And further suppose that your organizations also like the President's plan and want it to succeed? If you really wanted the poll to work out, what could you do? You could poll an unbalanced number of people who would tend to agree with Obama. And they did.
This poll sampled almost twice as many who voted for Obama as those who voted for McCain. That's one way to look for an imbalance. And the poll drastically undersampled Republicans and oversampled Independents based on voter registrations.
The charts are below. I think the pictures speak loudly. And you can read more detail on the CNS News Site
Tuesday, June 30, 2009
Media Bias - CBS & NY Times Distort Health Care Poll
Subscribe to:
Post Comments (Atom)
1 comment:
This is a great example that anytime statistics are presented, whether in the media or not, the methodology is faulty literally over 90% of the time. There are almost always vested interests, whether it's the media to promote an agenda, or providing the results the funding source wants.
Results can be manipulated very easily. When I was in college taking research class, my final big research project didn't produce the percentage spread I needed. So, without changing the numbers, I just plugged the results into several formulas in the textbook until I got the difference I needed. I thought at the time that was really something, & has cast a shadow of doubt for me on statistics ever since.
Your post is a good caution flag to all - ALWAYS check out the methodology before accepting any polls or statistical analysis.
Dirk
THE FIRST AMENDMENT, NOT POLITICALLY CORRECT II
http://tsalagiman2.blogspot.com/
Post a Comment