www.cmm.bristol.ac.uk/forum

Hi,

I have some questions about applying weights in MLwiN using runmlwin.

My data have two levels: individual and country. For each individual in each country, there is a weight assigned to it. For each country, the total of weight equals to 1. Individuals within countries are randomly sampled but the countries are fixed (i.e., no sampling for these countries among all countries in the world).

I was reading the document "weightings in mlwin" and section 2.2 mentioned that MLwiN requires the user to input both 'level 2 weights' and 'level 1 weights' while section 2.2.1 suggested that the weight used in MLwiN needs to be standardised.

Since I have only one weight variable in my dataset, I am confused with the level 1 and level 2 weights, as well as the standardised weights. Is the available weight a standardised weight, or is it only the level 1 weight?

Thank you very much.

Vivian

If you only have weights at 1 level then leave the weights at the other level as 'equal'. To use standardised weights just specify the weights column and check 'use standardised weights' in the menu box.
Harvey

Thanks Harvey.

I would also like to know if it is necessary to specify the weight natures (i.e., frequency weight, probability weight and analytic weights) in MLwiN? If not, does MLwiN deal with these weights differently?

Thanks again!