[Rasch] negative alpha

Mike Linacre mike at winsteps.com
Sat Dec 6 08:31:39 EST 2014


Parisa:

In your report, the Std. Dev. (observed standard deviation) is less than 
the SEM (standard error of measurement). The dispersion of the examinees 
is less than what is expected by chance.

Alpha (Reliability) = (Std. Dev.**2 - SEM**2) / Std. Dev.**2

Mike L.

On 12/6/2014 6:54 AM, Parisa Daftari Fard wrote:
> Dear Friends,
>
> I have run itemman on a set of data and come up with the following 
> result. Why alpha turned to be negative? I appreciate your incisive 
> comments on the result.
>
> N of Items           8
> N of Examinees      22
> Mean             2.864
> Variance         1.027
> Std. Dev.        1.013
> Skew             1.061
> Kurtosis         0.006
> Minimum          2.000
> Maximum          5.000
> Median           3.000
> Alpha           -0.754
> SEM              1.342
> Mean P           0.358
> Mean Item-Tot.   0.263
> Mean Biserial    0.337
> Max Score (Low)      2
> N (Low Group)       10
> Min Score (High)     3
> N (High Group)      12
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: https://mailinglist.acer.edu.au/pipermail/rasch/attachments/20141206/1226a21f/attachment.html 


More information about the Rasch mailing list