Hi, Statalisters:
I am not so much good in posting on the forum, please excuse me for not adhering the rules.
My setup (Windows 10, Stata 16.1 MP)
My objectives:
  1. To see the moderating effect of wealth status on the relationship between Intimate Partner Violence (IPV) and antenatal care (ANC) service
  2. to see the average marginal effects of the models
  3. To stratify by wealth status in order to see the presence of variation in subgroups (Low wealth vs. High wealth) in the associations between IPV and ANC
I have attached the regression output results in the word file attached with this forum below.

A. Moderating effect of wealth status output

Finings:
In the interaction model above, the household wealth status (AOR = 1.08, p = 0.754), had no a significant moderation effect on the relationship between exposure to emotional IPV and ANC service utilization.

B. the average marginal effects of the models

Findings: The effect of exposure to IPV on ANC is non-significant in the negative direction in both sub-groups. Maternal exposure to emotional IPV variable makes a difference in ANC probability of approximately 3% points lower in the high wealth sub-group, and about 5% points lower in the low wealth sub-group.

C. Stratified by wealth status (Low wealth vs. High wealth) in the associations between IPV and ANC - This is the output for low wealth status

N.B: Models adjusted for: Mother’s age, birth order, mother’s and husband’s education, media exposure, decision making autonomy, place of residence, and contextual regions - for all models

Findings: In the stratified analysis above, an association between spousal emotional IPV exposure and utilization of ANC service was demonstrated in women living in household’s with the lowest wealth (AOR = 0.71, (95% CI:0.52 - 0.97, P = 0.031)).

My question: I found non-significance in the interaction model and in the marginal model but found the significance in the stratified model (shown in the above regression output).
So, I don't understand why the results in the interaction model and in the marginal model don't show the significance that found significance in the stratified model.

Any possible explanation you have, please.