So this morning I log in to MyCentralJersey.com and see the front page article titled “Poll: Gov. Christie takes most blame for teacher layoffs, not unions or school boards.” Now why would that be?
Of course the first line written by Andrea Clurfeld is all “analysis” (read: think the way I tell you to, not the way logic shows you) than answering the Six Questions of Journalism.
Gov. Chris Christie is in danger of losing the first major battle he’s waging as governor, according to a new poll.
But I read through the marginal effort to rework the press release from the Monmouth University Polling Institute (MUPI). But one thing was glaringly missing! Why Andrea did not put in the raw numbers which tell the whole story! But one page three of the press release, you suddenly see why it isn’t included. First, the poll was conducted with 804 “New Jersey adults.” Perhaps MUPI would do well to understand just how many people actually live in the state. According to City-Data, New Jersey had an estimated population of over 8 million in 2006. So that means less than 1 millionth of the population was actually surveyed.Wow, that’s like looking at the first inch of your lawn from the curb which has die-back from the salt applied to the road during the winter and deciding that you need to replace the entire lawn with new sod.
Then I saw: This sample has a margin of error of +- 3.5 percent. Why is that important? To put it simply, the larger the number, the less statistical confidence there is in the polling that was done. This means there could be errors in how the results were collected, errors in how the questions were read to participants, or errors in how participants understood the questions. Yet, there are so-called “statisticians” who think this is perfectly acceptable. Among legitimate statisticians 3.0 percent is a stretch at best for polls to be considered legitimate. Amazingly, MUPI actually states the following on page 8 of the press release:
For results based on the total sample, one can say with 95% confidence that the error attributable to sampling
has a maximum margin of plus or minus 3.5 percentage points. Sampling error increases as the sample size
decreases [emphasis mine], so statements based on various population subgroups, such as separate figures reported by gender
or party identification, are subject to more error than are statements based on the total sample. In addition to
sampling error, one should bear in mind that question wording and practical difficulties in conducting surveys
can introduce error or bias into the findings of opinion polls.
So they are even willing to admit to the issue. I’m left to suppose that Ms. Clurfeld was either was told not to include this in the article or she included it, but it was deleted by an editor. Either way, it is a significant part of the content that was omitted. Thus the general public was not informed fully.
In the end, I decided it was far better to pillory the pollsters than to argue with Mr. Clurfeld who was tasked with making a gob of garbage stick. My comment to the article is as follows:
Gannett sure didn’t get it’s money’s worth on this poll. According to City-Data, NJ had an estimated population
of 8,724,560 and Monmouth University Polling Institute only called 804 New Jersey Adults. That’s barely
.00010! With a margin of error of 3.5 percent, the validity of this poll is statistically worthless.Come one Gannett, stop being the pipe organ for the NJEA and start supporting the taxpayers. After all, you need
to boost your subscriptions so you don’t fold like the rest of birdcage liners.
