**Date:** Thu, 2 Jul 2009 22:38:59 -0300
**Reply-To:** Hector Maletta <hmaletta@fibertel.com.ar>
**Sender:** "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
**From:** Hector Maletta <hmaletta@fibertel.com.ar>
**Subject:** Re: percent increase vs regression line as a predictor
**In-Reply-To:** <97D6F0A82A6E894DAF44B9F575305CC9093D4ADF@HCAMAIL03.ochca.com>
**Content-Type:** multipart/alternative;
Then my second paragraph applies: the crude average past percentage rate of
growth is usually a poor predictor, except if you have some objective
grounds to expect a constant rate of growth.

Why the number of prescriptions per month should be an increasing (or
decreasing) function of time? Time starting when? Are you talking of time
counted since some condition is diagnosed, or something similar, or the mere
passing of time? If time is the only predictor for the NUMBER of
prescriptions, the PERCENTAGE GROWTH of prescriptions has little to do with
it. If the number of prescriptions is a function of time, then the
proportional increase in the number of prescriptions would be (by
definition) a function of the logarithm of time.

Hector

_____

From: SPSSX(r) Discussion [mailto:SPSSX-L@LISTSERV.UGA.EDU] On Behalf Of
Pirritano, Matthew
Sent: 02 July 2009 20:25
To: SPSSX-L@LISTSERV.UGA.EDU
Subject: Re: percent increase vs regression line as a predictor

The variables are not dichotomous in this case. I'm looking at number of
prescriptions per month. Number of prescriptions is the DV and time is the
IV.

Matthew Pirritano, Ph.D.

Research Analyst IV

Medical Services Initiative (MSI)

Orange County Health Care Agency

(714) 568-5648

_____

From: SPSSX(r) Discussion [mailto:SPSSX-L@LISTSERV.UGA.EDU] On Behalf Of
Hector Maletta
Sent: Thursday, July 02, 2009 3:17 PM
To: SPSSX-L@LISTSERV.UGA.EDU
Subject: Re: percent increase vs regression line as a predictor

For a pair of dichotomous variables, the percent difference in Y between the
two values of X is mathematically equivalent to a regression coefficient.

Average crude percentage change in the past (independently of predictors)
may be a very poor forecasting tool. To mention just a famous example,
remember the (in)famous Fisher blunder in 1929, predicting continuous growth
in the stock exchange by simply projecting average past increases, even
after the initial crash. The future does not always repeat the past.

Hector

_____

From: SPSSX(r) Discussion [mailto:SPSSX-L@LISTSERV.UGA.EDU] On Behalf Of
Pirritano, Matthew
Sent: 02 July 2009 17:44
To: SPSSX-L@LISTSERV.UGA.EDU
Subject: percent increase vs regression line as a predictor

Pawsers,

Stats question. I've not done any forecasting before other than with
multiple regression analyses. What is the difference between multiple
regression and using average past percent change to make predictions? Based
on one scenario I'm dealing with it looks like percent change results in a
positive curvilinear (possibly logistic?) relationship.

Thanks,

matt

Matthew Pirritano, Ph.D.

Research Analyst IV

Medical Services Initiative (MSI)

Orange County Health Care Agency

(714) 568-5648

[text/html]