Stats terminology question

[Therapist4Chnge]

---

Members don't see this ad.

I am trying to write up the findings to a particular sub-test, and for the life of me I can't remember the correct term to describe the following:

The normed data positively skews strongly to one side, and a minor difference from the norm produces a larger than expected difference in how the person would compare to the norm so that the associated descriptor is not really reflective of the actual level of functioning. More simply....a score of 10/10 was achieved by the vast majority of the normed population, garnering a descriptor of "average", though a 9/10 drops the person all the way down to a "below-average" descriptor. There is somewhat of a ceiling effect, as I am sure there would be a more stratified sample if additional items were added and the difficulty increased, though as it stands I'm stuck with a poor descriptor. I'm going to describe this limitation, but I'm pretty sure there is a term that captures what I want to say....and I can't remember it. Yet another reason I dislike screeners with limited items.

Thoughts?

 

[erg923]

Truncated?

 

[Jegg]

I am trying to write up the findings to a particular sub-test, and for the life of me I can't remember the correct term to describe the following:

The normed data positively skews strongly to one side, and a minor difference from the norm produces a larger than expected difference in how the person would compare to the norm so that the associated descriptor is not really reflective of the actual level of functioning. More simply....a score of 10/10 was achieved by the vast majority of the normed population, garnering a descriptor of "average", though a 9/10 drops the person all the way down to a "below-average" descriptor. There is somewhat of a ceiling effect, as I am sure there would be a more stratified sample if additional items were added and the difficulty increased, though as it stands I'm stuck with a poor descriptor. I'm going to describe this limitation, but I'm pretty sure there is a term that captures what I want to say....and I can't remember it. Yet another reason I dislike screeners with limited items.

Thoughts?

It sounds like negative skew, not positive skew, if most of the numbers are clustered at 10 and the tail is produced by lower numbers (unless you are reflecting the results via some form of mathematical transformation).

An extreme negative skew with leptokurtosis and a lack of homogeneity of variance?

 

[JockNerd]

Truncated?

Seems like it to me.

 

[Ollie123]

Agree that sounds like a negative skew.

Truncated and/or range restriction seem fitting to me.

 

[Therapist4Chnge]

Yeah...I screwed up the skew.

 

[edified1]

Sounds like a restricted range.

 

[IOPsych]

Agree that sounds like a negative skew.

Truncated and/or range restriction seem fitting to me.

I agree with range restriction being an issue here, but it goes beyond that I think ceiling effect may be the proper term for this. I am not 100% sure though, it is like the opposite of a low base rate.