umx multigroup not printing out CIs for all parameters, but only for one group?
lf-araujo
Joined: 11/25/2020
| Attachment | Size |
|---|---|
| dom.csv | 35.37 KB |
Hopefully the hangover from Boulder has passed when you read this :)
I want the CIs from this multigroup SEM, but it is only printing out the paths for males (I think). Here is the MWE, data in the attached:
m1 <- " # RI-CLPM with 3 timepoints
# latents
psy_0 =~ saps_0 + sans_0
psy_24 =~ saps_24 + sans_24
psy_48 =~ saps_48 + sans_48
# random intercepts
ri_psy =~ 1*psy_0 + 1*psy_24 + 1*psy_48
ri_vol =~ 1*total_gray_vol_0 + 1*total_gray_vol_24 + 1*total_gray_vol_48
# causal
total_gray_vol_24 ~ psy_0
psy_24 ~ total_gray_vol_0
total_gray_vol_48 ~ psy_24
psy_48 ~ total_gray_vol_24
# autoregressive
total_gray_vol_24 ~ total_gray_vol_0
total_gray_vol_48 ~ total_gray_vol_24
psy_24 ~ psy_0
psy_48 ~ psy_24
# immediate
psy_0 ~~ total_gray_vol_0
# total_gray_vol_0 ~ psychosis_0
psy_24 ~~ total_gray_vol_24
#total_gray_vol_24 ~ psychosis_24
psy_48 ~~ total_gray_vol_48
#total_gray_vol_48 ~ psychosis_48
"
ma <- umxRAM(m1, name="males", data = dom %>% filter(sex=="male"), std.lv = F,
optimizer = "SLSQP", tryHard = "yes", autoRun = "no")
mb <- umxRAM(m1, name="females", data = dom %>% filter(sex=="female"), std.lv = F,
optimizer = "SLSQP",tryHard = "yes", autoRun = "no")
mgo <- umxSuperModel("Multigroup", ma, mb, tryHard = "yes")
mgo <- umxCI(mgo, run = "yes",
which="ALL")
Which results in:
Running Multigroup with 38 parameters
[1/1] MxComputeGradientDescent(SLSQP) evaluations 13833 fit -0.9224 cha
[6/1] MxComputeGradientDescent(SLSQP) evaluations 43557 fit -0.203474 c
[9/1] MxComputeGradientDescent(SLSQP) evaluations 74415 fit 0.0158719 ch
lbound estimate ubound
total_gray_vol_0_with_total_gray_vol_0 0.139 0.407 0.935
total_gray_vol_0_to_total_gray_vol_24 -0.275 -0.068 0.478
total_gray_vol_0_to_psy_24 -0.459 -0.111 0.117
total_gray_vol_24_to_total_gray_vol_48 0.021 0.285 0.875
total_gray_vol_24_to_psy_48 -0.611 -0.229 -0.014
psy_0_to_total_gray_vol_24 -1.378 -1.022 -0.649
psy_0_to_sans_0 0.935 1.187 1.510
psy_0_to_psy_24 0.384 0.922 1.278
psy_24_to_total_gray_vol_48 -1.291 -0.958 -0.650
psy_24_to_sans_24 1.025 1.237 1.516
psy_24_to_psy_48 -0.199 0.051 0.339
psy_48_to_sans_48 -18.689 3.711 NA
total_gray_vol_24_with_total_gray_vol_24 0.181 0.259 0.398
total_gray_vol_48_with_total_gray_vol_48 0.047 0.147 0.435
sans_0_with_sans_0 0.155 0.284 0.428
saps_0_with_saps_0 0.371 0.500 0.661
sans_24_with_sans_24 0.042 0.140 0.252
saps_24_with_saps_24 0.305 0.416 0.559
sans_48_with_sans_48 -27.883 -0.978 0.090
saps_48_with_saps_48 0.543 0.843 1.347
psy_0_with_total_gray_vol_0 -0.676 -0.463 -0.304
psy_0_with_psy_0 0.248 0.458 0.673
psy_24_with_total_gray_vol_24 -0.081 -0.019 0.058
psy_24_with_psy_24 -0.110 0.024 0.123
psy_48_with_total_gray_vol_48 -0.154 -0.016 0.107
psy_48_with_psy_48 0.006 0.119 0.357
ri_psy_with_ri_psy -0.021 0.009 0.816
ri_psy_with_ri_vol -0.007 0.118 0.286
ri_vol_with_ri_vol 0.098 0.580 0.839
one_to_total_gray_vol_0 -0.117 0.015 0.147
one_to_total_gray_vol_24 -0.068 0.082 0.233
one_to_total_gray_vol_48 0.428 0.575 0.741
one_to_sans_0 -0.136 0.005 0.146
one_to_saps_0 -0.154 -0.013 0.128
one_to_sans_24 -0.144 -0.003 0.138
one_to_saps_24 -0.173 -0.025 0.123
one_to_sans_48 -0.410 -0.135 0.141
one_to_saps_48 -0.336 -0.036 0.258
Now the above are only the males CIs plus the means. Where is the females CIs?
- What am I missing
Many thanks,
Luis
Solution
It is related to how umx labels the paths, you just have to change them accordingly:
ma <- umxRAM(m1, name="males", data = dom %>% filter(sex=="male"), std.lv = F,
tryHard = "yes", autoRun = F, optimizer = "CSOLNP") %>%
umxModify( regex="*", newlabels ="m_\\1" , name = "males")
mb <- umxRAM(m1, name="females", data = dom %>% filter(sex=="female"), std.lv = F,
tryHard = "yes", autoRun = F, optimizer = "CSOLNP") %>%
umxModify(regex="*", newlabels ="f_\\1" , name = "females")
mgo <- umxSuperModel("Multigroup", ma, mb, tryHard = "yes")
Log in or register to post comments