In which Mr. Answer Person gets corrected for his comments about the City of Decorah survey

(Mr. Answer Person on Monday discussed the City of Decorah survey and stated, "A 10 percent response rate is considered statistically valid."  That statement prompted this e-mail from Luther College professor Loren Toussaint)

"I just wanted to quickly point out that a 10 percent response is not uncommon, but it is a LARGE leap to say it is statistically valid.  It is far from it.  Determining the validity or accuracy (probably better stated as accuracy) of a survey requires many things--not the least of which is appropriate response rate.

However, perhaps most important in the current case is that the survey was almost certainly not representative of the community at large.

The sample that responded was likely motivated for good or bad reasons to respond.  That leaves out a lot of folks in the middle.  The point is, survey samples are accurate in large part to the extent that you can demonstrate that the sample that completed the survey adequately represents the population of interest.  Unless those analyses are done and presented, then the accuracy of the survey will be open to question.  Sample size is important (especially for determining margin of error in estimates), but so too is representativeness of the sample."

Mr. Answer Person says: Loren is ABSOLUTELY CORRECT!  49.5 percent of those taking the survey were men and 50.5 percent were women.  So far, so good.  But nearly 46 percent of those taking the survey were 55 and older.  And--most glaringly--55 percent of the survey respondents were people who had lived in Decorah for at least 20 years.

So Loren brings up a good point when discussing survey results--they should be taken with a grain of salt.  A scientifically-based survey would have been much better--but city officials were watching taxpayer money and decided this informal survey would be good enough to get a sampling of public opinion.