More Thoughts on a WOM White Paper

Last week I blogged about a white paper researched and written by Walter Carl, a Northeastern University Asst. Professor. In my post I offered up a few questions related to the study and Walter posted a thorough reply on his WOM-Study blog.

Regarding my assertion about the BzzAgent sample most likely being more skillful at recognizing and reporting word-of-mouth activity than those in the convenience sample, Walter said …

First, I agree with you, it shouldn't be a surprising finding that BzzAgents report more WOM episodes than non-Agents. The amount of the difference was what surprised me (for example, Agents having twice as many of their total interactions include a WOM episode). Second, I don't have any direct measurement of whether or not BzzAgents are more or less skillful at recognizing WOM opportunities. My sense is, however, and like yours, that they probably are better able to recognize WOM opportunities.

WOW! BzzAgents reported 100% more word-of-mouth interactions than did those in the convenience sample. That discrepancy is empirical enough for me to know BzzAgents are indeed more skillful at recognizing and reporting word-of-mouth activities than are the everyday people in the convenience sample.

Walter also provided more specifics on the demographics of the BzzAgent sample compared with the convenience sample. There are significant discrepancies between the two samples. Walter said the BzzAgent sample was 83% female and nearly 50% were 30 years of age or older. While the convenience sample was 67% female and their average age was close to 20 years of age.

Given this clarification, the two samples appear too dissimilar to draw any empirical conclusions. That’s just my take. Read the WOM White Paper and Walter’s clarification for yourself and draw your own conclusions.

Comments

[Dustin]

"for example, Agents having twice as many of their total interactions include a WOM episode"

That would be 100% more, not 50% more... correct?

John, thanks for having the patience and chutzpah the rest of us don't have to ask these questions. Good stuff.

[johnmoore (from Brand Autopsy)]

Doh! 100% it is. My bad. Change has been made. Thanks for keeping me mathematically correct Dustin.

[Walter Carl]

Hi John,

I enjoyed reading your follow-up post about my WOM research project with BzzAgent. However I want to clear up a misunderstanding in your post. I did provide the additional demographic information on the BzzAgent sample, which was as you reported. However, in order to be able to make apple-to-apple comparisons I matched the BzzAgent sample to the "everyday people" convenience sample by age and education level. That is, from the total BzzAgent sample I created a sub-set of those agents with "Some College" and aged 18-29. This is reported in the white paper on Footnote #1 on page 3.

You are right in pointing out that there are is a higher representation of females in the BzzAgent sub-set that I used to compare with the convenience sample. However, among both the BzzAgent sub-set and the everyday people convenience sample there were no statistically significant differences for the number of interactions, number of WOM episodes, and the episode-to-interaction ratio (the percentage of total interactions that included a WOM episode) based on sex of the respondent.

Therefore, I think one can safely conclude that the BzzAgent sub-set and the everyday people convenience sample is an apple-to-apple comparison based on education level, age, and sex.

Thanks again for your interest in this research!

Walter Carl

[johnmoore (from Brand Autopsy)]

Cool. Thanks for the follow-up Carl and if the demos are indeed similar, then it will be more an 'apples to apples' comparison. However, Jonagold Apples taste different than Fuji Apples and Fuji Apples taste different than Granny Smith Apples. How 'bout them Apples?