Create Weighted Values from Percentages

[Mony]

Looking for advice to see if this method has any relevance when creating Values for a Weighting Model

If I have 3 variables in the Sample I.e.

A = 5
B = 10
C = 85
Total = 100

For my weighting model I want to apply a greatest value to A and lowest to C.

What I did was invert the percentage breakdown.

A = 1 - (.05) = .95
B = 1- (.10) = .9
C = 1- (.85) = .15

I then applied these values to create my weighting values. Has this method any statistical significance?

 

[tkhunny]

The linear nature of your adjustment is a bit odd. It MAY be what you want. A more traditional approach would be the reciprocals (particularly if you are talking about standard deviations or some related concept), but that may not be what you want. When you said "invert", I thought you meant the reciprocal. Careful with words.

A-Weight: 1/5 = 0.20
B-Weight: 1/10 = 0.10
C-Weight: 1/85 = 0.011765

You must weight based on your needs and how your model operates.

You may also wish to scale whatever weights you are using:

From your linear adjustments:
0.95 + 0.90 + 0.15 = 2.00, so divide then all by 2 to give
0.475
0.450
0.075
And 0.475 + 0.450 + 0.075 = 1.000

From my reciprocal adjustments
0.20 + 0.10 + 0.011765 = 0.311765, giving
0.641509
0.320754
0.037737
And 0.641509 + 0.320754 + 0.037737 = 1.000000

Making the sum 1.00 may or may not be a natural fit for your model.

To sum up: Do what works and makes sense to you.