Why are reviewers so hard on the z?

Red__Zed

I know you've seen my discussion about the statistical irrelevance of true delta.

Self reported, small sample size statistics are inherently flawed and essentially useless.

UNKNOWN_370

actually i havn't but i know i fill out all true delta reports every 3 months in full detail when necessary. Send me a link to that report.

Red__Zed

All the info you need is here:


Sampling bias - Wikipedia, the free encyclopedia

Sample size determination - Wikipedia, the free encyclopedia

UNKNOWN_370

I see what you mean. But sampling is not meant to give exacts anyway, but an idea. At least you get a small society of people in your situation. You take it as a general guide. it's like polls in politics. They give you a general idea, as long as they aren't rigged. But they don't determine who will be the next president.
Nielsen ratings cancelled many a good show based on sampling. I know where you're coming from. But car samples, though you can't get biblical with them, do hold some type of value. at least an 1/8 value to your total research. The forum is a sample bias as well... But i won't negate it as a 1/8 soursce as well.
The problem with corporate sponsored ratings sytem like CR vs Edmunds is that CR may declare unreliability due to a sample set of minor TSB issues vs independents who look for things that the common public consider substantial issues. Long term tests from 8 magazines would be sample testing. C&D brake failure... Sample test. Everything is a sample test in media.

Red__Zed

Right, and this present a small glimpse into why self-reporting is dangerous as well. With exit polls, certain types of people are more likely to report who they voted for, skewing the results. This gets even more extreme when you talk cars and reliability.



yeah, based on flawed sampling...bad sampling processes are the problem. And if you think Nielsen's practices are bad, look at TD...


I don't think you are getting the point.