# Small samples, multiple predictors

#### Noahm

##### New Member
Let us consider a song competition. There are 3 global criteria weighted A worth 40 points, B worth 30 points, and C worth 30 points, equaling a total of 100 points to determine the Winner. The 40-point criterion has 4 sub-dimensions, each with weightings of 5 pts, 15 ots, 15 pts, and 5 pts. The 2nd criterion has 3 sub-dimensions, with weightings of 15 pts, 10 pts, and 5 pts. The 3rd criterion has 3 sub-dimensions with weightings of 15 pts, 10 pts, and 5 pts. There are 30 competitors. There are 7 judges. Can regression analysis be used to determine which sub-dimension is the greatest predictor of their respective global criterion? Can it also be used to determine which global criterion is the greatest predictor of contestants' final scores? The concern is the limited data points with only 7 judges. And the uneven weightings of the 3 global criteria and their and sub dimensions. That is, 40 pts, 30 pts, and 30 points globally, then 5, 15, 15, 5 for the 2nt criterion and so on. Thanks in advance for your insight.