Stuart Mckelvie writes on 1 Apr 99,:
> By the way, I would not use the label "correlational", because I
> like to reserve that term for one of the kinds of nonexperimental
> research (also called descriptive research). However, some writers
> allow that "correlational" be used in two ways: as equivalent to
> descriptive and as a sub-category of descriptive (a study in which a
> correlation coefficient is calculated as the major data).
I would use the term "correlational" because it shows that the statistical
technique used in the analysis does not directly determine the nature of
the conclusions that can be drawn. My initial foray into meta-analysis
last year allowed me to understand just how closely related many
statistical techniques are. For example, should you do a t-test or a
point-biserial correlation? One can easily be translated into the other.
With bivariate data we have r and r squared and with ANOVA (F), we
have eta correlations and eta squared. There is no strict demarcation in
statistical techniques between those used for correlational and causal
interpretations.
> Also: Should we have another name for the "independent" variable when it
> is not truly experimentally manipulated?
I think many people call them correlational variables or, in this case,
subject variables.
> And should we have another name
> for the "dependent" variable in this situation?
I think we should but I usually also call them correlational variables.
Rick
Dr. Rick Froman
Psychology Department
Box 3055
John Brown University
Siloam Springs, AR 72761
[EMAIL PROTECTED]
http://www.jbu.edu/sbs/psych
Office: (501)524-7295
Fax: (501)524-9548
"The plural of anecdote is not data."
- Roger Brinner, Economist, Data Resources International
