Friday puzzle -- statistical data collection

Re: Friday puzzle -- statistical data collection

What if you used a denial as a positive answer. Instead of asking a question directly, phrase the question as a denial. Instead of "Have you ever ...?" Use "I have never..."

Re: Friday puzzle -- statistical data collection

I don't see how that cold mask that individual's experience.

Re: Friday puzzle -- statistical data collection

Do you know anyone that has used statistics?

Have you ever seen anyone using statistics?

Would you approve of your daughter dating a statistician?

Would you support legislation to outlaw statistics?

How many statisticians does it take to screw in a lightbulb?

Re: Friday puzzle -- statistical data collection

All good for getting impressions about the prevalence or acceptability of the practice of statistics, but we want to get data on how many people actually practice statistical analysis.

And that is your "practice no one would want to admit to"? I knew this would get ugly! ;-)

OK

I admit it. I have practiced statistics and am so ashamed ;-) Is there a 12 step process to become absolved?

Actually, all of the statistical analysis that I have been involved with was related to statistical process control. Will have to think about this one a bit.

Re: OK

I don't do statistics. I just program machines to do it for me. ;-)

Re: Friday puzzle -- statistical data collection

You could 'blind' the test by printing up a number of cards. Some percentage would have the "EQ" and the rest would have questions with known answers.

The subject would draw a card at random and (without showing it to you) tell you whether the statement on the card was true or false.

That would do it

You'd have to make sure that the respondent saw that there were multiple cards, so that if he got the EQ, he would know that you didn't know that.

The version I heard was similar: Tell the person to go into a room and flip a coin. If it comes up heads, write down the answer to the EQ. If it comes up tails, they are to flip again and answer the question "is the second flip a heads?". The if you question 2,000 people, throw out 500 "yes"es and 500 "no"s, and treat the rest as a sample of 1,000 respondents to the EQ.

Both of these methods allow you to estimate the statistic for the population, but introduce their own error, making the SD of the sample larger than a straight sample of EQ responses. That would be interesting to try to quantify, but I think that admitting I would find that interesting brands me!

Re: OK

And I'm an actuary; I just count dead people! ;-)

As the lawyer said to his

statician expert witness. "Just provide the data." And the statician replied, "Fine, but what do you want the data to indicate?"

There are truths, lies, and statistics.

In a democracy, one cannot ignore statistics.

I mean no slight to all those who know how to do statistical analysis.

Re: Friday puzzle -- statistical data collection

it takes two statisticians to screw in a lightbulb. the trick is getting them INSIDE the lightbulb...

Actually

Statistics can be very accurate. The problem is not the statistics, but rather the misuse of them. Not having a random sample, not disclosing the size of the basis, trying to apply population statistics to individuals etc.

Misuse of statistics

Well said, Dan. I like the way one person I know puts it:

Most people use statistics as a drunk uses a streetlight -- more for support than for illumination.

Link for Alex (and any others interested ;-)) *LINK*

Talk on the misuse/misunderstanding of statistics that is kind of interesting.

http://www.ted.com/index.php/talks/peter_donnelly_shows_how_stats_fool_juries.html


Talk on stat misuse