Z-score for normal and reversed variables

Question

I'm comparing multiple countries across many indicators (with different scales and units). I want to develop a score for each country and I'll do that by taking the mean of the z-scores for each indicator. But some of my indicators are reversed. So, for example, for -Life Expectancy- less is worse, while for -Obesity- less is better. How can I statistically account for that? Can I just multiply the z-scores for those reversed variables by -1 so when I take the mean of the z-scores it actually means something?

See also points 2-7 [here](https://stats.stackexchange.com/a/9361/17230). — Scortchi - Reinstate Monica, Aug 23 '17 at 14:10

score 2 · Accepted Answer · edited Aug 23 '17 at 09:05

2

It's perfectly acceptable to do this, so that a negative z-score implies a negative result, just as long as you state that this is what you have done somewhere. This way they all mean the same thing.

Have you thought about doing J-charts for this too?

edited Aug 23 '17 at 09:05

Nick Cox

48,377
8
110
156

answered Aug 23 '17 at 08:43

Beavis

216
1
5

Z-score for normal and reversed variables

1 Answers1