www.cmm.bristol.ac.uk/forum

Hello,

I have recently carried out analyses with MLwiN and now my partner is cross-checking it using the same database and executing exactly the same steps. Unfortunately the results we get differ. They are the same when it comes to the direction of the effects and their significance but the values of the estimates and SE are not exactly the same. I'll add that my partner is apparently using older version of the software. Has anyone had similar experience and knows what can possibly account for these differences? I would very much appreciate your support.

thanks,

Nina

This is possible depending on the versions of MLwiN and the particular estimation settings are used. Details of the changes that may have caused this are recorded on http://www.bristol.ac.uk/cmm/software/m ... fixes.html, however as I don't know the details of your model I cannot point you to the exact change.

Thank you. I guess the easiest way to check if that is the problem is for my partner to upgrade to new version. This is however not possible at the moment for various reasons and therefore I wanted to see if other people encountered similar problem. I hope that the fact that the variations there are only in terms of small differences in values indicates that this might be the upgrade issue. Anyway, thank you for the input and I'll gladly hear if anyone else had faced it.

Nina