Jump to content

twin study ACE model - questions & concerns


MonDie

Recommended Posts

Maybe should have gone in biology... @#$%!

 


 

I posted this on another forum, but I'm hoping to get better responses here.

 

I got my understanding of the ACE model from Wikipedia, so it might be over simplistic or just wrong.

https://en.wikipedia.org/wiki/Twin_study#Methods

 

my understanding of the ACE model:

 

Pasted from the other forum website for convenience. I wrote it.

Dizygotes (fraternal twins) and monozygotes (identical twins) are used to calculate the relative contributions of "additive genetics" and "common environment". Here's how.

The equation assumes that all twins share 100% common environment ( C), but dizygotes only share ~50% of their genetics (½A). The monozygotes (A+C) exhibit a higher rate of conformity than the dizygotes (½A+C), and the difference between conformity in monozygotes and conformity in dizygotes (½A) is multiplied by two and divided by the total conformity of monozygotes (A+C) to derive the relative contribution of genetics.

 

Concerns:

 

Is there a model that controls for discrimination in response to genetically determined traits (e.g. skin color)?

One of my concerns is that the culture will induce certain environments in response to certain genetics. For example, you will probably experience more discrimination (enviornment) if you have naturally dark skin (genetic), which makes that extra discrimination an "A" genetic factor.

 

My other concern is that non-random mating will cause certain genetic factors to be detected as "C" common environment. If your parents both have some hypothetical intelligence gene—hence their attraction to one another—then all your siblings will have this gene regardless of whether they're your twin, hence the gene will be confused for an effect of common environment.

 

If you're confused, I explained these concerns in more detail in my third and final post in the link above.

Edited by MonDie
Link to comment
Share on other sites

Control - no I would not really think so as you already have control in that external influences (such as discrimination due to skin colour) are likely to equal. It is possible that non-identical twins could have features such that one is a target of discrimination and the other is not - but seems highly unlikely. If, as would seem most probable, both twins would attract the same amount of discrimination (averaged out etc.) then this in itself is the control. If you compare Child A with random Child B - you have to be very careful not to be fooled by external influences that affect A and not B; but with the studies you are suggesting as far as possible the only difference is slightly altered genetics. If there are obvious factors that might cause the assumption that environment is shared to be falsified then the pairing would most likely fail to meet selection criteria.


Second concern. My understanding is that you are looking for differences not similarities. What you say is correct - but already envisaged.

 

You have a common environment (so difference cannot be due to surrounding), you then have two groups idents and non-idents; and it is this you investigate. You are investigating question A do twinned siblings exhibit same or different behaviour in this circumstance - if idents have a ratio of 95% similar between siblings and 5% different between siblings, and non-idents have a 50% similar 50% dis-similar then you have the makings of a conclusion. ie by your choice you have factored out common environment and th common portion of genes and the only factor that may cause an interesting result is the changed portion of the genome.

Link to comment
Share on other sites

If there are obvious factors that might cause the assumption that environment is shared to be falsified then the pairing would most likely fail to meet selection criteria.

You're right, they could check that each non-identical pairing has roughly the same skin-tone, leveling out whatever wasn't leveled out by non-random mating. As described in my intelligence gene example, however, this would cause the skin-tone discrimination to be detected as C rather than A, which would still be misleading to the layman who will likely equate "common environment" with "family environment".

 

 

Second concern. My understanding is that you are looking for differences not similarities. What you say is correct - but already envisaged.

Meh, similarity and difference are perfectly reciprocal (hence interchangeable) concepts.

 

 

You have a common environment (so difference cannot be due to surrounding), you then have two groups idents and non-idents; and it is this you investigate. You are investigating question A do twinned siblings exhibit same or different behaviour in this circumstance - if idents have a ratio of 95% similar between siblings and 5% different between siblings, and non-idents have a 50% similar 50% dis-similar then you have the makings of a conclusion. ie by your choice you have factored out common environment and th common portion of genes and the only factor that may cause an interesting result is the changed portion of the genome.

I don't think this answers the question. The non-idents won't be dissimilar in the relevant DNA sequence if the parents tend to have matching phenotypes and thus matching genotypes. The identical twin similarity due to genes will be fully attributed to A only if non-idents vary at least as often as you would expect by chance with random mating.

Edited by MonDie
Link to comment
Share on other sites

You're right, they could check that each non-identical pairing has roughly the same skin-tone, leveling out whatever wasn't leveled out by non-random mating. As described in my intelligence gene example, however, this would cause the skin-tone discrimination to be detected as C rather than A, which would still be misleading to the layman who will likely equate "common environment" with "family environment".

 

Any huge difference may be ruled our - or if you have a decent trial size then it would not matter. A well designed experiment would have minimal possibility for a genetic factor to be confused. Not sure what you are getting at about non-random mating.

 

 

Meh, similarity and difference are perfectly reciprocal (hence interchangeable) concepts.

 

No you couldn't be more wrong. You can be different in a myriad of ways - you can only be similar as one. And in this particular set-up yu have engineered a situation in which EVERYTHING except "one thing" is the same between the twins - you therefore look for differences which you can try to correlate with that "one thing"; there are many things that are the same (ie all environmental and some genetic factors) so similarities tell you much much less.

I don't think this answers the question. The non-idents won't be dissimilar in the relevant DNA sequence if the parents tend to have matching phenotypes and thus matching genotypes. The identical twin similarity due to genes will be fully attributed to A only if non-idents vary at least as often as you would expect by chance with random mating.

 

You are trying to get into more detail than the experiment allows. IF there is a significant difference between twin A and twin B if they are non-ident but not if they are ident; then you can say that this is likely to be genetic in basis - nothing else can be covered.

 

Link to comment
Share on other sites

back to your previous post

It is possible that non-identical twins could have features such that one is a target of discrimination and the other is not - but seems highly unlikely. If, as would seem most probable, both twins would attract the same amount of discrimination (averaged out etc.) then this in itself is the control.


highly unlikely = infrequent
The non-idents will infrquently experience different levels of discrimination, but sometimes they will, and those instances may or may not significantly impact the results depending on how relevant that discrimination is to the dependent variable (DV).

If you compare Child A with random Child B - you have to be very careful not to be fooled by external influences that affect A and not B; but with the studies you are suggesting as far as possible the only difference is slightly altered genetics. If there are obvious factors that might cause the assumption that environment is shared to be falsified then the pairing would most likely fail to meet selection criteria.


I didn't realize that the pasted text didn't mention E, unique environment. E is the rest of the difference that can't be accounted for by A or C. Any differences in environment between the non-idents will be unique environment.




your most recent post

No you couldn't be more wrong. You can be different in a myriad of ways - you can only be similar as one. And in this particular set-up yu have engineered a situation in which EVERYTHING except "one thing" is the same between the twins - you therefore look for differences which you can try to correlate with that "one thing"; there are many things that are the same (ie all environmental and some genetic factors) so similarities tell you much much less.


I'm assuming a focus on some dependent variables (DVs) that can be measured along a one-dimensional axis. I'm not talking about an exploratory study to discover which traits are more genetic and which are more environmental; I'm talking about an attempt to precisely quantify the relative contributions of genetic and environment to a specific DV.

You are trying to get into more detail than the experiment allows. IF there is a significant difference between twin A and twin B if they are non-ident but not if they are ident; then you can say that this is likely to be genetic in basis - nothing else can be covered.


This renders the experiment useless.

 

If my goal is to gradually distinguish and quantify all contributing factors, starting with the broad categorizations of A, C and E, that still leaves the ACE quantities rather useless on their own.

If my goal is to quantify the relative influence of genetics and environment, the problem is that my results could be skewed in either direction by non-random mating, depending on whether mates tend toward similarity in the DV (IQ, skin tone, etc.) or difference in the DV (Major Histocompatibility Complex, S&M, etc.)—I'm talking about difference relative to the average difference, of course.
I suppose one could investigate the overall mating tendency to establish whether A is an upper limit or lower limit. For example, if mates tend to be similar in the DV, the researchers could say that the contribution of genetics is ≤ A. That is still terribly blunt.
Perhaps a good control would be to (1) check that all parents are the biological parents, (2) survey the biological parents to measure how similar or disparate each couple is in the DV, and (3) make sure that each twin-pair sired by a similar couple is matched with a twin-pair sired by an equally disparate couple.


 

btw

If the researcher is surveying e.g. skin color to assure that it's controlled for, they might as well use the same-skinned and different-skinned non-ident pairs to quantify the relative contribution of skin-color. I'll call it the CASE model!

Edited by MonDie
Link to comment
Share on other sites

 

I didn't realize that the pasted text didn't mention E, unique environment. E is the rest of the difference that can't be accounted for by A or C. Any differences in environment between the non-idents identicals will be unique environment.

 

fixed it

Link to comment
Share on other sites

  • 3 months later...

No you couldn't be more wrong. You can be different in a myriad of ways - you can only be similar as one. And in this particular set-up yu have engineered a situation in which EVERYTHING except "one thing" is the same between the twins - you therefore look for differences which you can try to correlate with that "one thing"; there are many things that are the same (ie all environmental and some genetic factors) so similarities tell you much much less.

 

Perhaps a good point is that they are using simarity/difference on the ratio level, for they say that non-identical twins are half different (presumably with respect to the average degree of difference between people).

 

If you're lost:

The quantifiability of variables is sometimes split into the "interval" and "ratio" levels of measurement. The former allows for the quantification of differences between values, but only the latter allows for ratios since there is a point where the property is absent, assigned the value zero.

 

In the case of genetic differences, it would be extremely rare that any two people would be different in every way possible (within the confines of their gene pool), and the degree of difference will vary with how closely related they are. Therefore it only makes sense to assign the value of zero to absolute similarity, and to speak of ratios between degrees of difference.

 


 

Returning on the 17th. :)

Edited by MonDie
Link to comment
Share on other sites

 

Perhaps a good point is that they are using simarity/difference on the ratio level, for they say that non-identical twins are half different (presumably with respect to the average degree of difference between people).

 

If you're lost:

The quantifiability of variables is sometimes split into the "interval" and "ratio" levels of measurement. The former allows for the quantification of differences between values, but only the latter allows for ratios since there is a point where the property is absent, assigned the value zero.

 

In the case of genetic differences, it would be extremely rare that any two people would be different in every way possible (within the confines of their gene pool), and the degree of difference will vary with how closely related they are. Therefore it only makes sense to assign the value of zero to absolute similarity, and to speak of ratios between degrees of difference.

 


 

Returning on the 17th. :)

 

 

No you are lost. Half different in this context means that on average for a large enough sample fraternal twins will share half their genes - they are not saying that two fraternal twins are half alike; that would be nonsense. 100pct of idents' genes will be shared and 50 percent of fraternals' will be shared.

 

You get results as follows

 

idents different / fraternals same = screwed up experiment or you need to really think hard about this one

 

idents different / fraternals different = most likely unique to each twin environment is most likely cause

 

idents same / fraternals different = most likely genetic

 

idents same / fraternals same = most likely due to shared environment

Link to comment
Share on other sites

  • 1 month later...

I should acknowledge in this thread that I probably misrepresented and oversimplified the calculations. My separate thread concerning only the maths may be merged at your discretion. http://www.scienceforums.net/topic/86520-twin-data-math-clarify-rijsdijk-sham/

 

However, Rijsdijk & Sham, 2002 say that twin methods rest on several assumptions which pertain to the discussion here.

http://bib.oxfordjournals.org/content/3/2/119.long

The assumptions include the following:

 

• MZ and DZ twin pairs share their environments to the same extent.

 

• Gene–environment correlations and interactions are minimal for the trait.

 

• Twins are no different from the general population in terms of the trait.

 

• Matings in the population occur at random (no assortment).

 

I also want to note that MZ and DZ twins would also share cohort effects, which would make cohort effects shared-environment too.

 


 

Fru ̈hling V. Rijsdijk and Pak C. Sham. Analytic approaches to twin data using structural equation models. Briefings in Bioinformatics. 2002 Jun;3(2):119-33.

Edited by MonDie
Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
×
  • Create New...

Important Information

We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue.