[IPAC-List] How significant is significant enough?

[Winfred Arthur, Jr.]

. . . assuming that i am understanding you correctly, a couple of points:

1.  if the .36 represents a criterion-related validity estimate, that it 
is really not small in the context of what one generally sees in the 
literature and practice.  of course it is based on a relatively small 
sample size and so it may be susceptible to some sampling error biases.

2.  based on your description of the sample, it would seem you may have 
range restriction issues as well and so the .36 is probably an 
underestimate.

3.  merely increasing the sample size is likely to increase, well, 
really decrease the p-value but is unlikely to systematically increase 
the point estimate (i.e., the correlation) if the additional cases are 
drawn from the same "pool" as the present sample.

hope this is of some help.

- winfred

Jason Bowling wrote:

> Hi all,

>

>  

>

> I recently joined this list-serv and greatly enjoyed the conversations

> thus far. 

>

>  

>

> So, I have a question and would appreciate any thoughts.

>

>  

>

> I am considering administering a test of basic math skills to candidates

> for apprentice electrical lineman. Actually, prior to my starting with

> the company, a math test was SOP for candidates to these jobs, but no

> one ever asked if the test was a valid predictor. Furthermore, the

> weight placed on test results was always ambiguous. We seemed to be

> taking a pragmatic approach to our employment processes... that is, it

> always seemed to work in the past.

>

>  

>

> Our apprenticeship does include testing components, and math does play a

> large role in those tests. Furthermore, the job description does

> describe math skills to be an essential competency, and we have had to

> let apprentices go for failing these apprenticeship tests. 

>

>  

>

> So, I gave the test to our current apprentices and recent graduates

> (N=30), and compared their test results to their average apprenticeship

> test results (r=.36, and p=.02). The correlation seems weak to me, but I

> have no basis of comparison. 

>

>  

>

> I could probably get more subjects to bolster the significance, but

> again, I'm not sure at what point I should feel satisfied.

>

>  

>

> Any help?

>

>  

>

> Thanking you I am,

>

>  

>

>  

>

> Jason Bowling, PHR

>

> Human Resources Manager

>

> S.S.V.E.C

>

> Ofc: 520.515.3480

>

> Cel: 520.220.6563

>

>  

>

> _______________________________________________________

> IPAC-List

> IPAC-List at ipacweb.org

> http://www.ipacweb.org/mailman/listinfo/ipac-list

>