I have a dataset with some multiple observations, but the observations are different for one specific covariate. The dataset looks something like this:
| ID | Salary | Provider | Field | Weight |
| 1 | 10000 | U1 | A | 0.33 |
| 1 | 10000 | U1 | B | 0.67 |
| 2 | 12000 | U1 | C | 0.5 |
| 2 | 12000 | U1 | A | 0.5 |
| 3 | 12500 | U3 | A | 1 |
| 4 | 14000 | U4 | C | 1 |
| 5 | 12000 | U1 | C | 0.33 |
| 5 | 12000 | U1 | B | 0.67 |
| 6 | 16000 | U4 | B | 1 |
Basically ID represents an individual that studied at University U in a certain field. Individual 1, instead of studying only field A, studied a combination of A and B, with the weights given to each subject being 0.33 and 0.67. What kind of weight should I use to take this into account?
I Know frequency weights cannot be non-integer, so I would wonder if I should just multiply everything by 100? Or is my understanding of fweight wrong, and that would give me something different, rather than adding the individuals so that they are not duplicate anymore?
0 Response to Multiple observations with some different covariates and non-integer weights
Post a Comment