| |
 |
|
|
Science Forum Index » Statistics - Education Forum » Correlation Problem: Need Advice
Page 1 of 1
|
| Author |
Message |
| Guest |
 Posted: Tue Dec 12, 2006 12:16 pm |
|
|
|
|
Dear Group:
I'm looking for a rigorous, statistic that will correlate
(validate??) computed risk values and an observed incidence rate on a
monthly time step. My hypothesis is as the monthly incidence rate
increases the computed risk average would increase (i.e. the risks
values in the frequency distribution for a given month would be skewed
right favoring higher risk). My perceived problem is the differences in
data type and dealing with the computed risk's frequency distribution.
The monthly incidence value is one value expressed as a ratio ranging
from 0.0 to 1.0. However, the computed, monthly risk values, which are
mathematically extrapolated from many data sources, are interval values
ranging from a minimum of 1 (low risk) to a maximum of 343 (highest
risk).
Where I'm having difficulty is how do I compare a single ratio value
(incidence) to a histogram of computed risk values in interval format
for any given month?
This is what I'm thinking so far:
1) Rescaling the 1 to 343 distribution of risk values to 0.0 to 1.0 to
match the scale of the incidence value.
2) For each month, compare the mean value of the risk distribution to
the incidence ratio value using a Pearson Correlation Coefficient,
thereby determining how closely the mean risk follows the increases or
decreases in the incidence ratio.
However, I'm not sure if what I'm thinking is even close to a
meaningful analysis given the issues of data type and distribution or
what kind of conclusion one could draw; yet, any help would be greatly
appreciated. Thanks so much.
Kind Regards-
J. |
|
|
| Back to top |
|
| |
|
Page 1 of 1
All times are GMT - 5 Hours
The time now is Wed Dec 03, 2008 9:09 pm
|
|