Phil McDSome people are hanging onto every little twitch of the Gallup during this political season. Inevitably this brings up the question of how accurate are polls. To be sure there is the statistical error say plus or minus 2% or whatever. However sometimes this masks a more fundamental uncertainty in the poll. Namely the underlying data may be inherently unstable or highly cyclical. This can render the supposed statistical error irrelevant because the inherent instability of the data dominates.

One of my favorite examples is of the Gallup Poll's Measure of Daily Mood. For example if the poll showed a reading of 60% last Friday and then it fell to 45% this Monday one might be tempted to conclude that the mood of the public was adversely effected by the financial turmoil over the weekend. Nothing could be further from the truth. The reality is that the Daily Mood is totally swamped by what day of the week it is. Every single week the mood swings from 45% to 60% without fail. It is governed by how close we are to the weekend and little more. See a visual of this on Gallup's web site.

Dr. McDonnell is the author of Optimal Portfolio Modeling, Wiley, 2008


WordPress database error: [Table './dailyspeculations_com_@002d_dailywordpress/wp_comments' is marked as crashed and last (automatic?) repair failed]
SELECT * FROM wp_comments WHERE comment_post_ID = '3134' AND comment_approved = '1' ORDER BY comment_date




Speak your mind


Resources & Links