posted 05-20-2007 06:30 PM            

I would also be interested in any spreadsheets that contain formulas that analyze data. My least favorite subject in graduate school was statistics. I believe this was the case because I never really appreciated its use as I would now that I have read some of you geniuses discuss the importance of statistics in Polygraphy. Now I am quite interested in learning it. So, what I have decided to do is learn statistics at my own pace, and doing so in the realm of polygraph statistics, so that I will maintain an interest in it...

By the way, this forums discussion some time back motivated me to "interview" on significant reactions of comparison questions in the post-test, and it has lead to some disqualifying admissions! While I do not post as much as I would like to here, I read all of the postings. I wish that we had an upload page on here where I could upload my Lafayette poly charts for QA and discussion.

posted 05-20-2007 08:08 PM            

Having a project lends a lot of interest to the subject of stats. Beware though, most statistics books can cure insomnia. Some, I'm convinced, have some form of chloroform in the ink - so you may need a designated driver after you study.

You'll have to keep in mind that 3 and 7 position scale scoring are not really statistical tests, but threshold model tests. The ratios are rarely, if ever, empirically derived, and even decision thresholds are mostly set arbitrarily and subsequently investigated for their accuracy and utility. One exception to this is the OSS system, for which the ratios were derived from the analysis of data, and based both nonparametric and parametric theory.

Don Krapohl and Mark Handler wrote a paper on ratios last year - rethinking the traditional (arbitrary) 2:1 3:1 and 4:1 ratios, in consideration of their data.

If you are trying to automate the calculation of a three or seven position scale you could use a lookup table, or a nested series of nested IF statements. Both work equally well, and offer advantages only how you want to work with it. Statements seem more intuitive at first, but once you are familiar with them lookups are easier to modify.

Press F1 from within your spreadsheet and search for "IF" statements, and LOOKUP, VLOOKUP, and HLOOKUP functions.

Are you taking measurements by hand, or with an automated tool?

Also, what kind of decision model are you working toward? Is it based on traditional point thresholds, or do you have something else in mind?

r

------------------
"Gentlemen, you can't fight in here. This is the war room."
--(Stanley Kubrick/Peter Sellers - Dr. Strangelove, 1964)

posted 05-21-2007 06:59 PM            

This offer also goes out to anyone on this bulletin board interested in exploring measurements and scores.