Page 1 of 1

PGRE percentile advantages - taking it in April/October

Posted: Mon Jan 21, 2008 4:25 am
by Maxwells_Demon
Hi,

I've heard a couple times that taking the Physics GRE in April, it will be easier to obtain a higher percentile score, since most people "take it in October/September and less in April..." Does anyone know anything (like statistics) about this?
thanks,

-Maxwell's Demon

Posted: Mon Jan 21, 2008 11:30 am
by twistor
According to the ETS your score is scaled in such a way that your scaled score means the same thing from test to test. In other words, a score of 800 on the fall PGRE should be the same as an 800 on the April PGRE or from any PGRE given up until that point.

Posted: Mon Jan 21, 2008 5:04 pm
by quizivex
That's right... apparently they do some kind of statistical procedure (which they call equating) that makes a score of XYZ reflect the same level of performance regardless of what version of the test you took, what date you took it on, and who else took it the same day. I'm not sure how it works, I tried to find out, but I had similar concerns as you and was told it's not a problem.

Also, the score reports say "the percentile scales are updated annually to reflect the performance of a recent sample of test takers..." so the possibility of getting a different percentile for a given scaled score in April is discounted.

Posted: Mon Jan 21, 2008 5:40 pm
by marten
Yeah, ditto quizivex and twistor here.

Plus, if fewer people take the test in April, then the confidence level for any given interval will be lower, because statistically there is less data. So it it might be easier to score higher and score lower, the normal distribution won't be as well defined. But either way, I think that there are enough Physics GRE test takers on any set day for any given test version (there are different forms on each day) that the ETS is able to be confident enough, for all practical purposes. Plus, I think, most places ignore small score differences anyway.

Marten

Posted: Mon Jan 21, 2008 8:42 pm
by GCS
Also, the score reports say "the percentile scales are updated annually to reflect the performance of a recent sample of test takers..."
interestingly enough, when I got the report after sending my scores to different universities my general quant. and analytical writing percentages went up by 1% from the time I took it in May.

Posted: Mon Jan 21, 2008 8:46 pm
by quizivex
The general test works differently... the same thing happened to me.

Posted: Mon Jan 21, 2008 9:03 pm
by twistor
Maybe their scaling method is only accurate to within +/- 0.5 %. Then if they scaled your score and you ended up with 67.5 % instead of 67.1 % you would get 68 instead of 67 after rounding.