1

Hi I have data (weight in kg) for samples from different heights (e.g. 5.5 ft, 5.8ft, 6ft). I want to normalize their weight with height (to exclude the effect of height). I am not very familiar with statistics. Example data are as follows:

e.g. patient A has 5.5ft and 60 kg, patient B has 5.8ft and 73 kg, patient C has 6 ft and 100 kg.

In that case, I would like to normalize patient B and C with patient A who has lowest height 5.5 ft.

I use this formula: Normalized weight value = (real weight value * minimum height)/real height.

So, Patient B (normalized) = (73 * 5.5)/5.8 = 73.017 kg

Is my way of calculation correct? If not, please advise the correct ways.Thanks.

Teddy
  • 11
  • 1
  • 2
  • Could you explain what you are trying to accomplish with this "normalization"? It sounds a little like you might want to [control for height](http://stats.stackexchange.com/questions/17336) in a regression involving weight. – whuber Aug 24 '16 at 21:24
  • Hi I would like to compare the weight among those groups (with different heights), but I don't want the height as confounding factor. So, I need to consider what will be their weight if they have the same height. Thanks. – Teddy Aug 24 '16 at 21:27
  • Then you are indeed asking how to regress weight against height. I'm sure we have [many threads about regression](http://stats.stackexchange.com/questions/tagged/regression) that can help you! – whuber Aug 24 '16 at 21:28
  • The relationship between height & weight shouldn't necessarily be a straight line. Are you familiar with the Body-Mass Index ([BMI](https://en.wikipedia.org/wiki/Body_mass_index))? – gung - Reinstate Monica Aug 24 '16 at 21:53
  • No i dont need the relationship between height and weight. I like to exclude the effect on height and just compare the weights among the group. But to compare their weights, i need to make sure that the higher weight is not because of higher height. And i know BMI, but i cant use it as I need to know weight for specific reason. Thanks. – Teddy Aug 24 '16 at 21:58
  • Are you planning on using the "normalized" weight to predict something else (e.g. "risk of heart disease")? – GeoMatt22 Aug 24 '16 at 23:13
  • Yes, kind of. So i need that normalized weight value. Can you please suggest how to do that. Thanks! – Teddy Aug 24 '16 at 23:16
  • 2
    If you are using weight to predict something else, then as @whuber said, really you "want to *control* for height in a regression involving weight". Aside from the thread whuber linked, [this answer](http://stats.stackexchange.com/a/78830/127790) explains the concept. – GeoMatt22 Aug 25 '16 at 00:09
  • 2
    The point is that weight and height may interact in different ways, depending on what third variable you are comparing to. So the appropriate "normalization" could differ as well. – GeoMatt22 Aug 25 '16 at 00:15

0 Answers0