6

I have seen statements like this in various places: "Such and such accounts for X percent of so and so". But it's not always clear what this means, how it is calculated or proven.

You will see this a lot, in particular, on 23andme.com, in your genetic report. If I'm 5' 11", and my genetics "accounts for 4%" of that...

Can one's genetics account for x% of their weight?

Randolf Richardson
  • 2,034
  • 2
  • 19
  • 21
Matt M.
  • 199
  • 2
  • Welcome to Skeptics: questions here should be about being skeptical of notable unreferenced claims. I've reworded your question to be on-topic. – Sklivvz Sep 27 '11 at 08:13
  • 2
    Your genome, existing in every single cell of your body, would reasonably comprise a certain % of your weight. But that's not really what's being asked, is it? – David Hedlund Sep 27 '11 at 08:28
  • Please given one concrete example. But my general feeling is that people usually blame genetics for factors they could actually control, like physical activity and diet (apart from the fact that weight loss should not be a goal, but rather fat loss should be a goal). – johanvdw Sep 27 '11 at 08:46
  • Thanks for rewording, Sklivvz. I didn't know whether to post this one on statistics, skeptics, or who-knows-what. But I already had this account created... – Matt M. Sep 27 '11 at 17:04
  • On a humorous note: if I was born a fruit fly, I'd weigh less! – horatio Sep 27 '11 at 21:45
  • Imho, you're right, that such numbers are meaningless. Humans without genes don't exist, so you can't perform arithmetics with those numbers. The same counts for talent-vs-training numbers, power-vs-technique comparisons and similar situations, where statistic numbers are just used, to pretend some deeper knowledge, to impress the audience. – user unknown Sep 28 '11 at 23:54

2 Answers2

2

“A is a gene for X” is a shortcut used by scientists when they actually mean to say “gene A modifies the expression of X, all other things being equal.” [1] In particular, it means that an allele of gene A influences X compared to another allele of the same gene.

This means that if you hear “gene A accounts for x% of trait T” then what is really meant is that “gene A accounts for x% of the difference in expression of trait T, all other things being equal”.

On the flip side, this means that if x% of some trait are determined by genes then the rest of the variability comes from outside factors, such as nutrition.

[1]: see for example Richard Dawkins, The Selfish Gene

Konrad Rudolph
  • 12,427
  • 3
  • 56
  • 84
  • Do you want to express, that 2 persons (or 2000), using the same nutrition, will vary by x% of body weight, just because of their genes? Independent from their behaviour, whether they perform sports, or how they heat their environment? But isn't the likelihood to do sports, and to decide, which nutrition to use, and how much, influenced from genes too? – user unknown Sep 28 '11 at 20:38
  • @user If those behaviours which influence trait X are influenced by genes then these genes are, by definition, also genes “for trait X”. Apart from that I said that the genes will have a *maximum* influence of x% on the trait (height). In fact, giving such a number is very hard for that very reason, because we don’t generally know which genes influence a trait indirectly. All we can measure (even in the best case) is the influence of *one* gene on a trait. – Konrad Rudolph Sep 29 '11 at 08:09
  • 1
    Your comment glosses over a point you made in the answer: If someone says a gene accounts for 15% of thumb-length, they generally mean it accounts for 15% of the *variance* in thumb-lengths, which is much less than the length of your thumb multiplied by 0.15. – Oddthinking Jan 05 '14 at 03:51
1

If I understand the question correctly, you are not asking for what the percentage heritability is, but whether it is valid to assign a percentage at all.

In situations like height, there is clearly a genetic factor (tall parents tend to have tall children) and an environmental factor (well-nourished generations tend to be taller than poorly-nourished generations). Weight has an even stronger environmental factor (as short-term weight-loss due to diets clearly shows).

In cases such as weight and height, while it may not be possible to predict the exact measure from genes alone, it is possible to account for a large part of the variability. So, yes, there are valid statistical techniques to establish what percentage of the variability can be accounted for by a single factor (or group of factors) like genetics.

I provide as evidence for this:

Oddthinking
  • 140,378
  • 46
  • 548
  • 638
  • Thanks for this answer. I understand what it means for some trait's variability to have a measurable genetic factor in general. But what does it mean to say *one person's* trait is accounted for by some unique percentage? – Matt M. Sep 27 '11 at 17:11
  • a) Tall parents having tall children can be taken as a social, cultural effect as well. b) I weight 80kg. If I stop eating, I will lose about 40 kg before dying, but if I eat, and eat, and eat, I might get 160 kg heavy. My genes can't do much against it - can they? – user unknown Sep 28 '11 at 20:21
  • @user unknown: re: Height: I explicitly didn't claim that height was only genetic. However, I thought it would be generally accepted that it is highly attributable to genetics. The heights of adopted children would match their biological parents much more closely than their adoptive parents. – Oddthinking Sep 29 '11 at 01:25
  • @user unknown: re: Weight. Agreed re: effects of under and over eating, but your genes could affect (i) how efficiently you convert your food to body mass, (ii) how much food you desire, (iii) the types of food you desire, (iv) how much activity you prefer to do/are capable of doing, (v) how much you are affected by peer pressure to adopt the socially-acceptable weight, (vi) how capable you are of overriding your desire to eat more/less and be more/less active, etc. – Oddthinking Sep 29 '11 at 01:29
  • @Oddthinking: re (height): My parents are smaller than me and my brother/sister, while they were bigger than their parents/my grandparents; probably due to problems with nutrition in the youth of them because of WWI and WWII. But I guess, under same circumstances, the lengths would be more similar, that's right. – user unknown Sep 29 '11 at 06:48
  • I understand cited papers, that they talk about populations, and the probability to have overweight with a special set of genes. But if you change the observed population (people/americans, americans/male americans, male persons/male persons between 30 and 40 and so on), the numbers will vary, but you can be part of different subgroups with different numbers. You can not individualize these numbers, imho. – user unknown Sep 29 '11 at 07:42
  • @User Unk: I'm afraid I don't understand your objection. Perhaps in the question you could give a very specific example of a claim you think is inappropriate. (The one about 5' 11" being *only* 4% genetic seems artificial to me.) Alternatively, let's move it to chat. – Oddthinking Sep 29 '11 at 13:13