December 1

Disclosure: I have an introductory college level of statistical background (but even that was a long time ago).

We put out a survey, essentially asking for value X at a time in the past and value X at the present. We did the same with value Y (i.e., asked what value Y was in the past and what value Y is now in the present). Approximately 100 people were surveyed, and each answered all four questions. A preliminary review of the data shows there is change in Y does not depend on change in X (My terminology may be inexact; I'm trying to say that ∆X does not explain ∆Y). However, my "preliminary review" holds little weight: I need statistical evidence to back it up.

This is where I'm at a loss. How do I show that ∆X does not explain ∆Y? Dtlyng (talk) 03:24, 1 December 2014 (UTC)[reply]

Am I right in presuming that each of the 100-odd people gave an accurate measure of whatever X and Y were/are, not just an estimate? If so, plotting a scatter diagram of ∆Y against ∆X may be sufficient evidence of the lack of correlation, i.e. a pattern like a swarm of bees isn't going to give anything definite, no matter what analysis is done. Calculating the product-moment correlation coefficient will give a numerical measure of the closeness of fit.→86.171.209.142 (talk) 21:53, 1 December 2014 (UTC)[reply]

What does person month mean? For example, the heart attack is 12 per 1000 person months? — Preceding unsigned comment added by Whereismylunch (talk • contribs) 03:27, 1 December 2014 (UTC)[reply]

A person month means the equivalent of one person for one month. It's often used to describe an amount of work but in "the heart attack is 12 per 1000 person months" it means an average of 12 heart attacks if 1000 persons are considered for a month. The actual data could for example be 24 heart attacks in a group of 2000 persons during a month, or 48 heart attacks in a group of 1000 persons during 4 months, or 12 heart attacks in a group of 500 persons during 2 months. PrimeHunter (talk) 04:03, 1 December 2014 (UTC)[reply]

This is analogous to the concept of a man hour. SemanticMantis (talk) 20:21, 1 December 2014 (UTC)[reply]

Yes, but months are inexact measures of time, varying from 28 days to 31, so they are a rather poor choice. I'd have gone with man-years. StuRat (talk) 02:44, 3 December 2014 (UTC)[reply]

Maybe in this context a month is defined as year/12? —Tamfang (talk) 21:42, 3 December 2014 (UTC)[reply]