The University of Michigan has been gathering data on consumer sentiment for 60 years, with a new Consumer Sentiment Index released monthly. The questions included in the survey look like the following, when they are asked by the University of Michigan on the telephone:
Now we see these same questions showing up on the Do-It-Yourself research tool, Google Consumer Surveys (#gsurveys). Do you think these (fundamentally identical) questions are being fielded by someone who wants to replace the University of Michigan as the nation’s official arbiter of Consumer Sentiment? Do you think these questions are being fielded by the University of Michigan, so web results can be compared to the University’s own telephone results? Or do you think the questions are being asked for some other reasons? Here’s what’s appearing now on the Google Consumer Surveys tool:
The beauty of what the University of Michigan does is that it hasn’t changed in 60 years, making it possible to create a graph like the following, which goes back to 1952.
The Consumer Sentiment Index (on which the above graph is drawn) is based on the answers to a number of questions, and is calculated using a formula. But even the answers to single questions within the questionnaire have unique value. Consider this question, which is appearing on Google Consumer Surveys now …
Here is how the residents of New Hampshire have answered that same question on 30 separate occasions, over a period of 8 years, as measured by the Granite State Poll, conducted by University of New Hampshire Survey Center.
When do you think fundamental public policy measurements, such as the University of Michigan Consumer Sentiment Index, will switch from telephone (as the way in which the data are gathered) to Internet? Will that happen in the next 12 months? In the next 24 months? In the next 36 months? Will it take longer than that? Or will it never happen?
Could a DIY tool such as Google Consumer Surveys ever produce meaningful data on this topic? Or is any data produced by Google Consumer Surveys inherently suspect, no matter how many ongoing refinements Google makes to the tool?