I have two data series to compare, one from the Federal government, the other from the Gallup organisation. Both are concerned with charting unemployment and the combination of unemployment and underemployment on a monthly basis.
They use different methodologies and, I am sure, somewhat different question and answer formats in their search for the same basic information, so it is no surprise that they do not provide exactly the same results. However, any divergence over time does catch my attention.
The Federal data is from the Bureau of labour Statistics and the Gallup data is from their latest report. Note: Gallup polls on this question regularly and uses 30-day rolling averages. They report once at mid-month and once at end-month. I use their end-month stats to compare with the BLS stats.
The difference between Gallup’s report of 10.3% unemployment for February clashes with the BLS report of 8.9% and has caused me to look back over the last fourteen months to see if there was anything unusual going on. It seems that the last four months have shown a growing divergence between the two reports for both unemployment alone, and when combined with underemployment.
It has taken me some time to put these graphs together and I face a series of meetings today, beginning shortly, so I am just putting them up for the moment without much commentary. The divergence may be explained by differences in methodology alone or perhaps it is temporary, but it is noticeable, so note it, I do. I will be keeping an eye on it in coming months.
Although I have no time for a detailed look at the two methodologies, I do wonder whether one factor, emotion, might make a difference during times when unemployment and underemployment are both unusually high for much longer than average periods of time. I have commissioned a couple dozen national surveys for various purposes from private services like those provided by Gallup, although I have used Harris and Zogby. So I also know from experience that who asks the question can have an impact on the response. I ask myself, is it possible that respondents respond differently to a government surveyor than they do to a private surveyor under these potentially emotion-driven circumstances? I can’t answer that question, but it does linger in the back of my mind.
Normally, I would not raise a question like this without first doing my own full analysis, but I know that is not going to happen given my current workload. So I am doing something irregular. I am posting these graphs just to share with you what I have noticed. If someone out there wants to do the necessary research to comment analytically, more power to you! In the meantime, I will watch this and see if the divergence continues and perhaps grows. It may be telling us something of significance or it may not, but it is interesting.