From: <rbaer_at_atsu.edu>

Date: Thu 28 Oct 2004 - 00:25:34 EST

Rationale

I have the recurring problem of students misinterpreting the alternative hypothesis statement as an interpretation of the t-test p-value not as a "tail decision". The "quick students" think R has made a mistake. The "slow students" (please save me) accept the alternative hypothesis with huge p-values "because the program told them to". All students seem confused. If this had affected a single group of students, I'd let it go, but this is a recurring theme with each new crop.

R-devel@stat.math.ethz.ch mailing list

https://stat.ethz.ch/mailman/listinfo/r-devel Received on Thu Oct 28 00:38:53 2004

Date: Thu 28 Oct 2004 - 00:25:34 EST

Full_Name: Rob Baer

Version: 2.0.0

OS: Windows

Submission from: (NULL) (198.209.172.106)

Request/wish summary: state the alternative hypothesis before the t-test and p-value

A typical cor.test output looks like the following:

*> x=runif(30)*10
*

> y=runif(30)*10

> cor.test(x,y)

Pearson's product-moment correlation

data: x and y

t = 0.6161, df = 28, p-value = 0.5428

alternative hypothesis: true correlation is not equal to 0
95 percent confidence interval:

-0.2552588 0.4568795

sample estimates:

cor

0.1156454

Rationale

I have the recurring problem of students misinterpreting the alternative hypothesis statement as an interpretation of the t-test p-value not as a "tail decision". The "quick students" think R has made a mistake. The "slow students" (please save me) accept the alternative hypothesis with huge p-values "because the program told them to". All students seem confused. If this had affected a single group of students, I'd let it go, but this is a recurring theme with each new crop.

Proposed fix. Generaly, we decide on an alternative hypothesis (or should) as an a priori statement. I think that there would be MUCH LESS confusing if the printed hypothesis statement came BEFORE the t-test and p-value output. This should put it in conceptual sequence; it would not be so easily confused with an interpretation.

This same problem happens in some other test output as well. For these, a similar order change could greatly reduce confusion about what is being stated. If the output code is shared, the fix would only have to be done once.

Thanks for considering it.

R-devel@stat.math.ethz.ch mailing list

https://stat.ethz.ch/mailman/listinfo/r-devel Received on Thu Oct 28 00:38:53 2004

*
This archive was generated by hypermail 2.1.8
: Wed 03 Nov 2004 - 22:45:26 EST
*