Professional > Data management scripting > Working with the Weight component > Rim weighting
 
Rim weighting
You use rim weighting (sometimes referred to as rim target weighting) when:
You want to weight according to various characteristics, but you do not know the relationship of the various combinations of those characteristics. For example, when you want to weight by age, gender, and income and you know the targets for each age, gender, and income group, but you do not know the targets for some or all of the combinations, such as men under 25 who earn more than 25,000, women between 25 and 44 who earn 10,000-20,000, etc.
You want to weight using target weighting but there are no respondents who fulfill the requirements of one or more of the cells of the target weighting table. This is more likely to happen when you base the weight on several variables and so have a multidimensional weighting table that has a large number of cells.
Rim weighting is designed to attempt to weight all of the characteristics at the same time. The accuracy of the weighting will depend on how well your sample matches the target population. If the sample is a good match, then it is likely that the Weight component will generate acceptable weights. If the sample is not a good match it is possible that the weights will be acceptable for some of the subgroups but not for others, which will, of course, mean that the weighting will be unacceptable overall.
As the rim weighting process runs, it tries to distort each variable as little as possible while still trying to attain all of the desired proportions among the characteristics. The Weight component produces a root mean square figure, which tells you how much distortion has been introduced, and therefore how reliable your sample is. The larger the figure, the greater the distortion and therefore the less accurate your sample is. The root mean square figure is recorded in the weighting report (see Weighting report).
Another powerful feature of rim weighting is that it automatically rescales all of the target values to the same base. For instance, suppose you have a sample of 5,000 respondents and you define rim weighting parameters as follows:
A weighted total of 10,000
Weighting targets for age in percentages
Weighting targets for gender that add up to 758
Weighting targets for occupation that add up to 1134
The Weight component calculates the weights for these characteristics, using the figures given, and then adjusts them to the required total. If you do not define a total, the weights are adjusted to the sum of the targets given for the first variable specified.
As you can see from this simple example, rim weighting can be used when you have weights coming from different sources, and when those weights are expressed in different ways.
See also
Working with the Weight component