My Dear Mark

no problem in your decision  when 3 variables (or any varibales) acceptance this mean that these variables are efftect on response(incom) and other variables not effect(3 var.)

in ganeral you can test the main hypothesis if it is signficant(the all 6 variables ) are effect in this case you can use decompasion test (test B for each a variable )

With Best Regrad

(send me feedbake)

Many Thanks, Fadhil, I know that there's no issue statistically. It has occurred in my work many times over the years. It was purely how to write this into my table of Hypotheses Accepted and Hypotheses Rejected, given the 50/50 situation (which I've not experienced before and find it hard to believe that I can't just Google up a simple answer).

Taking there to be those 6 variables in relationship, if the statistical analysis discovered say:

1. Accepted in 5 cases, Rejected in 1 case, I would write Partially Accepted, then note the one exception.

2. Rejected in 4 cases, Accepted in 2 cases, I would write Partially Rejected, then note the two exceptions.

3. Rejected in 3 cases, Accepted in 3 cases. What do I write? Partially Accepted or Partially Rejected, then noting the exceptions?

It's purely a writing-up, not a statistical issue.

Many Thanks

Mark

Hi Mark,

Interesting question.

May I ask why you preferred to analyze the relationships to income separately (i.e. 6 separate analyses), instead of running one multivariable regression model with all 6 independent variables?

Best,

Faith

This is multivariate cases... Mmm

There seems to be a confusion throughout in that I'm only talking about writing up, a presentational issue, not a statistical issue. I have pulled one problem finding - problem in a presentational, not a statistical sense - out of a multivariate run.  Anyway, the answer appears to be, as I guessed, one of always stressing the positive ie that where you have that highly unusual 50/50 split between Accept and Reject, you record as a Partial Acceptance, noting the exceptions.

I'm ease with classical philosophy of Statistics... Based only one value of test statistic value ...leading to a very important conclusion for a  population....