A Data Science Central Community
As a coincidence, I noticed from Google Insights for Search that there was some sort of correlation between web search interests in "American Idol" (singing competition on US television) and "iPhone" (Apple smartphone). Looking at the time series graph, it appears that American Idol searches are seasonal from January to May each year, which correspond to the months which this television series is broadcast. As soon as American Idol interest drops in June, searches for iPhone spike in the same month. The annual June spike is noticeable above the general upward trend in iPhone search interest.
In our household, we were speculating if there were relationships in the American Idol and iPhone search trends. Perhaps people's interests turn to iPhones when they lose interest in summer television reruns in June. Perhaps people are interested in iPhones and other mobile devices as they start their summer travels. Most likely though, there are confounding factors contributing to make this a spurious relationship. In recent years, Apple seems to release new iPhone versions in June (we're due soon for the iPhone 5, or at least an intermediate release like the 4s according to the rumors). The June timing of Apple iPhone release dates doesn't have anything to do with consumer television viewing habits, or does it?
I have 2 questions from out of this potential insight. In general practice, what steps can be done systematically to identify confounding factors? Also, what best practices can you give for determining optimal product launch dates?