I finally got fed up with being bombarded with Harbor Freight coupons that were only good in their stores, since I'm about 5 hours from the nearest one. If you go to their unsubscribe page, however, you can cancel just that portion of their emails. Since it's been a long time since they've had any specials I cared about I was going to nix the whole thing, but they also allow you to control the frequency of emails, so I changed it to once a month. FYI.
The poll has a mean satisfaction of 31.54%, but given how pathetically small our sample size is, that's really misleading. Because people's experiences are based on an unknown number of products ordered (and over an unknown period of time when quality could've changed), it's not possible to do a really in depth analysis without making all sorts of hand waving assumptions. But to try and get a feel for how uncertain out data is (and because I already had the spreadsheet set up to do it), I ran the poll results through to analyze them with a 5% confidence interval to generate a cumulative distribution:
The column graph attempts to show two cumulative probability distributions, bracketing the likelihood of one choice being 95% likely based on the data.
For example, the chance that a randomly selected person who's had dealings with Harbor Freight will have 80-100% satisfaction could be as low as 2.6% and as high as 27% (with 95% confidence). The chance that they'd have between a 60-100% satisfaction could be as low as 10.4% and as high as 36.1%. Etc. The last column doesn't add a whole lot, because all it says is that the chance of satisfaction being between 0-100% is 100%. No kidding?!
Anyhow, the hugely sized spreads show that, indeed, our sample size isn't very statistically significant.
I think I've done a lousy job explaining this, but if anybody has an interest in such stuff, I'd be happy to try again. Also, suggestions for easier ways to communicate the results would be greatly appreciated. I made this spreadsheet to interpret Amazon product ratings. Sometimes you can get enough data for things to firm up pretty well.
I can come up with convincing arguments for that last column being all red or all blue. Driving me nuts!
__________________