Announcement

Collapse
No announcement yet.
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • Multilevel modeling- testing further the cross-level interaction terms after mnbreg

    Hi,
    I I highly appreciate any help in probing the significant cross-level interaction terms in this model. I have a multilevel model( individuals nested in organizational units). The IV1 is at unit level and moderators and two DVs are at individual level. I use multilevel SEM with Negative binomial link function in stata to test it. DV s are count variables (and have possibility of correlations).

    gsem ( DV1 <- IVs Moderator IV1*Moderator M1[unit]) ( DV2 <- IVs Moderator IV1*Moderator M1[unit]), nbreg

    A cross-level direct effect (effect of IV at unit level is significant on DV at individual level) and a cross-level interaction( IV at unit level, moderator and DVs at individual level) are significant. I have been asked to explain further and test this cross level interaction after running this multilevel negative binomial model.
    1. My approach with an ordinary regression is to test and plot the simple slopes for the relationship between IV and DV at one standard deviation above and below the mean of the moderator. Is this correct after negative binomial regression? Do I have to do anything further related to multilevel nature of the model?
    2. How can I get the predicted values for two DVs after gsem,nbreg? or after mnbreg?
    3. We can use margin command in stata for logit probit models in stata to test the interactions and plot them. 2-a)Can I use the such a command after multilevel negative binomial? 2-b) what command I can use after gsem with nbreg link to get the margins?
    Thanks in advance,
    Saeede
    Last edited by Saeedeh Ahmadi; 24 Jan 2020, 04:31. Reason: multilevel modelling , margins , cross-level interaction
Working...
X