| Message/Author |
|
|
|
|
hello,
I am running a two level path analysis model.
the number of observations per cluster ranges between 10-20.
sum clusters are identical in the second level variable.
under which conditions should i consider pooling similar clusters in order to increase the per-cluster sample size,will it improve the models power? |
|
|
|
|
| I would not recommend pooling clusters. The cluster size you have is not bad. |
|
|
|
|
Hi
I have about 950 individuals nested within 800 neighborhoods in my data set.
I am looking at the effects of neighborhood disadvantage on educational outcomes and it looks like I have some dependent variables which have some levels of ICCs. Does it make sense to run multilevel modeling in this case, even though I have only one individual per neighborhood for many neighborhoods? Would you please recommend an alternative to mlm if multilevel modeling would not be ideal in this case? Thank you. |
|
|
|
|
| I would try the multilevel modeling. It is not a problem to have only one individual in some clusters. |
|
|
| Back to top |
