In the ACE Model:What are the criteria for setting the start values？ |
- Date: 2020/12/14 19:19
- Name: Zhenzhen Luo
<luozhenzhen0419@gmail.com>
- Dear openmx developers:
I use your excellent software package OpenMx, and have one question about parametric setting.
In the ACE Model: aTwin Analysis (OpenMxUserGuide: P74), you wrote that: “These are set to be free and given a start value of 20”.
We are confused about the “values = 20” when you used mxPath function. According to our tests and understanding, the mean values of both bmi1 and bmi2 were near to 21, and you gave the start values of 20 which is smaller than true mean values.
For this question, we did the following two tests:
1.For your twinData, no matter what start value is, the results are similar under different start values.
values a^2
-0.1 0.7810503
0 0.7810504
0.05 0.7810506
0.1 0.7810506
0.3 0.7810506
1 0.7810506
2 0.7810506
20 0.7810506
21 0.7810505
2.For our data, taking different mean values has a impact on our results.
Ourdata1： Ourdata2：
values a^2 values a^2
-0.1 0.1084373 -0.1 0.0612887
0 0.50061187** 0 0.0612891
0.05 0.1084363 0.05 0.0612891
0.1 0.50061187** 0.1 0.0612893
0.3 0.1084433 0.3 0.0612891
1 0.1084363 1 0.0612894
2 0.1084362 2 0.0612892
20 0.1084363 20 0.0612892
21 0.1084363 21 0.0612893
What are the criteria for setting the start values?
Can we choose the mean of two variables, mean of any of two variables, or a smaller mean of two variables?
We are looking forward to your reply. Thank you!
Sincerely yours,
Zhenzhen Luo
 | |
This thread is locked.Only browsing is available.
Re: In the ACE Model:What are the criteria for setting the start values？