It’s all relative

Yesterday, for the first time in many months, I visited my local audio store to check out the latest equipment and to see how they've been weathering the recession.  When I asked how business had been the owner told me that last year people were saying, "Flat is the new up." He says that this year they're saying, "Open is the new up."

This same sort of relativism seemed to be at work in this press release last week from Clarabridge, the text mining people, and Harris Interactive.  It compares results from an online Harris Poll on US healthcare reform with online chatter during the same period.   In its survey Harris found that support for the current healthcare legislation had fallen to 40 percent by mid-November.  The Clarabridge analysis of social media pegged support at 65 percent during the same period.   What exactly am I supposed to do with that? The best I could do is conclude that social media has the huge bias that we all fear it has.  That it's basically a qualitative method to be valued for its depth and richness rather than its accuracy.  That even though we can count things it's not quant research in the traditional sense and we should be really careful with any conclusions we might draw.

But there is a larger issue here as well and it's the matter of fact presentation of an online survey as "survey research" and the gold standard against which the accuracy of social media should be measured.  They might have been better served by a comparison to a Gallup Poll  done by telephone during the same period showing support for healthcare reform never dipping below 48 percent. (And, yes, Gallup called landlines and cell phones.) At least it's closer to their 65 percent.  It makes you wonder whether we can find the truth any more or, worse yet, if anyone cares.

In Decembers past I often have left the audio store with some sleek new piece of hi-fi gear.  This year I left with a single record.  Yes, a record, not a CD.  I own plenty of both but I'm mostly a vinyl guy, which probably explains a lot.