Hi!
I’ll be a student in Day 3 session B, and I have several questions about the material. I thought I’d ask them here ahead of time so I don’t slow things down asking too many questions in the tutorial.
The questions in this post refer to the first video, “Modeling/Estimating genetic and environmental components”:
Edit: Some of my questions in this post were answered during the tutorial. I am including the answers as I think I understand them below, plus one remaining question I’m still confused about at the bottom.
Questions that were answered:
- In the OpenMx example of the
Height ~ b0 + b1*SNP + errormodel, why is it necessary to drop SNP from the model to test the significance of b1 (instead of just using the standard error of b1)?
The confidence intervals calculated by OpenMx are asymmetric. So while OpenMx gives you a confidence interval, the existence of a confidence interval doesn’t mean you have a standard error and the ability to use it to do a statistical test that will give you a p-value. So when you want a p-value, you use a fit test instead. (But the confidence intervals are still important information because they tell you more about the possible range of values and not just significant vs. not significant.)
Questions 1 and 2 were related to definition variables and have been moved to a separate thread.
Questions I am still confused about:
- I see you have a 1 in a triangle that has arrows from it to other things, and that the video says triangle are means. What is 1 the mean of?
- Do models ever put a value other than 1 in a triangle, or is it always 1?
- Is it just a matter of convenience whether a model has one triangle with a 1 in it and multiple things making use of that triangle vs. the model having two different triangles used by different parts of the model? Or does that difference mean something?
- Eg on slide 12 of the lecture slides there’s a model with two separate triangles with 1s, but then on slide 13 there’s a model where one triangle has arrows pointing to multiple variables.