Weighting example

 

You might need to ‘weight’ certain variables in your data to adjust a distribution to reflect some ‘target’ distribution.

 

For example, assume that the U.S. population is 50% male and 50% female and that our survey data, collected in the U.S., reports a sample size of 300: 120 male and 180 female. The actual distribution in this example is:

  

 

Male

Female

 

====

======

Frequency

120

180

Percent

40%

60%

 

But the target distribution you want to reflect is based on the known population:

  

 

Male

Female

 

====

======

Frequency

150

150

Percent

50%

50%

 

The desired weight is calculated by dividing the target distribution by the actual distribution. For example, for males, you divide the 150 target distribution by the 120 actual distribution, for a weight of 1.25. For females, you divide the 150 target distribution by the 180 actual distribution, for a weight of .8333.

 

To calculate a weight factor use the formula: desired N / actual N = weight.

Related topics:

Simple Weighting

Sample Balancing