Dapresy supports cell and RIM (Random Iterative Method). 


RIM weighting: R.I.M.weighting logic allows for the application to weight on multiple variables in the data so that after a predefined maximum number of loops (iterations) it matches the target value of all of selected variables. It's often used in two cases:

  • when you have many variables you want the sample to be representative for (and the table used for cell weighting becomes too big and you will get empty cells in your sample)
  • you don't know how the variables interfere, because it's using the totals only, this is not a problem to have the unknowns in the table below


Example of representative numbers of gender age:

gender/age
< 35
>=35
total
men
?
?
0.61
women
?
?
0.39
total
0.58
0.42
1


Below you see a summary of the logic when computing a new weight variable based on the R.I.M method:


  • The Target distribution can be set for up to six variables
  • “Filters” can be applied to the weight table which makes it possible to apply different target values in different sub groups, as an example the target distributions might be different indifferent countries in a multi country brand tracker
  • The iterative weight process runs until the target distribution is fulfilled for all ingoing variables
  • A “Result Tolerance value” can be set. If the Target is set to 20% and the tolerance is set to 0.1% it means that 19.9% -20.1% will be treated as an achieved target
  • A max number of iterations can be set in case the defined targets are not achievable. In practice, most iterative weighting schemes are achieved within 2-20 iterations. The Max number of iterations can be set from 2 to 100
  • An option for automatically rescaling the Targets in case of the actual distribution is zero respondents is available. The option will remove the targets for the “zero count options” and rescale the targets for the remaining options in the same variable
  • The weighting can be made within all data in the project or within each data set or within a specified interval like Week, Months, Quarter or year
  • A project can contain multiple weight variables in case of having the need to for example having booth weekly and monthly weights in a tracker.
  • The weight calculations are automatically applied to newly imported data to support an automated process


Here we have an example of the weight table when the R.I.M method is used, as shown the Target distributions are setup individually per variable (Age and Gender in this example) and Country has been used as filter which allows different targets per Country.