Apple iPad Consumer Disinterest Study Another Case of Bogus Data

(MoneyWatch)  I will admit to finding myself going back and forth on the Apple (AAPL) iPad in terms of its potential today. But I noticed another "study" making the rounds of blogs. This one comes from Retrevo, an online electronics marketplace. Unfortunately, the details suggest that there's nothing you can really extrapolate from them, meaning that this particular effort of trying to extrapolate the potential success of the iPad is useless.

The study claims all sorts of numbers that are supposed to represent consumer sentiment before and after the iPad announcement. Essentially, it claims that "iPad hoopla fails to convince buyers." For example, the company claims that 26 percent of people had heard rumors of the device but were uninterested in purchasing one, versus after the announcement, when 52 percent heard of it but were uninterested. Or there's the 49 percent prior to the announcement thinking that they wouldn't need an iPad versus the 61 percent after. It sounds interesting, until you get to the details of the study methodology:

Data for this report came from a study of more than 1000 randomly selected Retrevo users. A first random sample of users were surveyed between Jan 16-20, 2010 before the Apple Tablet was announced. A second random sample of users were surveyed between Jan 27 ?€" Feb 3, 2010, after the announcement. The sample was distributed across gender, age, income and location in the United States. Most questions had a confidence interval of 4% at a 95% confidence level.

One problem is that because the survey respondents were all users of the Retrevo site, at best the survey would only predict how the group of all Retrevo users might react. However, it likely isn't even that meaningful. One of the factors that is important in survey work is the percentage of people approached to take the survey that actually do. If the percentage is low, then you're getting less of a random sample, which is critical to statistical analysis, and more of a self-selecting group. The methodology description doesn't even mention this number. In addition, "most" questions had a 4 percent confidence interval at a 95 percent confidence level. Which ones? How about the ones they mentioned? Why not say what the confidence interval and level of those were?

Even if the sampling was meaningful, Retrevo focused on a negative interpretation that would likely give it the most play in the press. One number it didn't emphasize was that the percentage of people who said that they wanted to buy an iPad went from 3 percent before the announcement to 9 percent after. Given the anticipation and the fact that those who wanted to buy were probably the Apple loyalists, this would suggest an extensive expansion of potential customers. Apple has made a strong business out of the core alone, so if anything, you might equally point to this study as suggesting that the iPad would be a hit. That is, if the study were something you could take with a grain of mathematical seriousness, which it doesn't appear to be.

Image via stock.xchng user lusi, site standard license.