Interesting 2011 auto quality survey

the only thing even vaguely "interesting" about any j.d. power "survey" is their methodology. years ago, i used to work for a company that did various product surveys. i'm telling you how it used to work because i have no reason that jdp's methodology is any different.

1. the noobs would survey a sample of users and tabulate the results.

1. middle management would review and, um, "adjust" the numbers based on what our competition's numbers said, and of course, their "gut".

2. the corner office would review these "draft" numbers over lunch with the highest paying client du jour and return with the "finals" which then went to print.

being the young and naive stage 1 guy at that time, i asked why we didn't save the time and money with the stage 1 survey and just go straight to stage 2 or even 3. answer: "so that we're seen to be doing real survey work".

so, there you have it. jdp does real survey work. but i'll bet you a week old donut hole that those survey numbers don't end up as the published numbers without "additional input" from the people whose products are being surveyed and whose ads are making the publication profitable.

Hm, sounds pretty cynical. But believable. Is there any survey we can believe in?

the process i outlined is direct personal experience. it may sound cynical, but it's true.

depends what's at stake. if it's auto-generated, from a near 100% sample rate and close to real-time, sure. if it's a small percentage of potential population, not real time, has massive general implication, and results get "reviewed" my management sensitive to the reaction of recipients, then pretty much "no".

Vladimir Uljanov, better known as Lenin, has said in his time: "There are 3 kinds of lie: an ordinary lie, a big lie and statistics." :-)

I doubt that quote is from Lenin because something eerily similar used to be attributed to Disraeli ("There are three kinds of lies: lies, damned lies, and statistics." )