Please login or register to join the discussion.
Very trendy post. Interesting, thanks!
Jeff, I have done a full study of the subject covering over 300 companies. I did it first in 2017 and repeated it earlier this year. You can find my article on it at the address below, and that includes access to the data set. In short, there is no relationship between the two for industries where employees have little or no direct customer contact, and quite a high relationship for industries like hotels where there is a lot of direct customer contact.
Interesting research, Maurice!
I must admit, I usually yawn and lose interest when someone pulls out a regression analysis, but your study is really compelling. Thanks for sharing!
I re-read your post recently, and couldn't reconcile some of the math. I think your ultimate conclusion may be true, and I'm the first to admit that basic stats sometime elude me, so maybe you can clear this up:
If the R2 for Glassdoor vs. ACSI rating for the full group of 345 companies is .0438, how can the R2 for two subsets of that group be .0963 and .0514 if those two subsets comprise the entire group?
Full group: 345 companies (R2=.0438)
Subset 1 (high-touch): 125 companies (R2=.0963)
Subset 2 (low-touch): 220 companies (R2=.0514)
Well, that's the nature of R2 Jeff. It is a measure of variation. It is to be expected that if you take two relatively homogenous subsets, the combination of the two will have more variation than each individual subset. More variation means a lower R2 number.
A random example occurs to me, and I have not checked it for validity. Lets suppose you have an R2 number for the relationship between iron levels in carrots for each month of the year, and another R2 number for the relationship between iron levels in broccoli and the months of the year. The R2 number would probably be relatively high for each, meaning low variation in numbers. Combine the two with similar scores for other vegetables to have a set of scores for 'vegetables' overall, and there will be far more variation, so a lower R2 number.
Assign one of your company’s values to a certain employee, every month, based on a peer-voting process. The person who best represented that value can be set as an example and be publicly acknowledged for his actions. Employee portals like MyWegmansConnect One also yield very well and also improve employees enagement.