hi Nuala

One could argue (on the assumption that both groups have equal amount of members) that group h got the same group result with one member slacking, therefore the other members of group h had to work harder to get the same group result. This would justify the higher marks.

…just my 2 cents…

best wishes,

Frank Sierens
university college Ghent
faculty of science and technology

 

 

I’d be grateful for some advice.

 

In a recent assessment we have run we have seen large differences in the WebPA multipliers generated and are puzzling about how to fairly moderate the results.

 

In this case the assessment was run with marking as split 100 (a slip up – we had meant to use Likert).  This has generated some large ranges of WebPA scores, especially evident in groups where members have marked one person down significantly.

 

Here’s an example of scores from two groups g and h (ordered in terms of their webPA score). Both groups had the same group mark, but the first member of group h was given a really low score, thereby giving more of the pie to the remaining members, which I suspect explains their higher scores.  The graph shows the effect of a WebPA weighting of 50%. 

 

 

 

 

(the WebPA raw scores range from 0.96 to 1.03 for group g and 0.3 to 1.23 for group h).   

 

What we are not sure about is how to moderate this.  It’s likely that under this assessment members of group h who performed well will get bigger scores than those in group g who performed at the same level.  My suspicion is that if we removed the lowest performing member from group h and reran the assessment that the range of scores would significantly moderated.   

 

We could take the webpa weighting right down to minimise the amplitude, but this wouldn’t adjust the difference between the two groups.

 

So, any thought on what we could do now?

 

Best wishes

 

Nuala

ISS, Newcastle University