spssforum.com

I am currently using SPSS 15.0.

My study has one dependent variable (quantitative),
one categorical indipendent variable at level 1, individual(gender, value are 0 and 1)
one categorical indipendent variable at level 2, groups (nature of team, value can be 0,1,2).

Each group is composed by six individuals. My hypotheses ISthat, according to the nature of group, male and females scores different values of the dependent variable (in some groups men are better than women, in other group women are better than men). Since observation are dependant (Intra Class Correlation is about 30%), I have to use MultiLevel models for a study.

I am not sure of the correct form of the syntax for my models (which should take into account the interaction between team nature and gender).

I used this code:



The outcomes are strange because if I run the simpler model




condition factor is significative, while in the most complex model it is not.

Thank you in advance for any suggestion or correction you can provide me.

I am the author of the previous post (sorry if I figured out as a guest).
I modified the code because I cannot use condition as a covariate (it is a categorical variable). So, the code I run is the following:



Is it the right code? My hypotheses are specifically addressed to estimate if gender behaviour is different accordin to the nature of the group.
Do you know I can also made post-hoc comparisons among different condition (e.g. there is no gender difference in condition 1 while women performs better in condition 2), in case interaction is significative?

Hi,

I know this is a really old post but it relates to my question so I was hoping you might have found the answer in the mean time.

In my experiment I measure pupil dilation when subjects observe different stimuli.
between subject factor:
species_subject=2 (0-1)
within subject factor:
same_different=2 (0-1)
increase_decrease_stimulus=2 (0-1)


For my experiment, I asked adivice from a statistician. He proposed the model below but I myself would have chosen for 'by' instead of 'with'. But I am only a beginner to multilevel, so perhaps you could give me an advice?


mixed pupilsize_subject with int lin quadr species_subject same_different increase_decrease_stimulus
/ fixed int luminance_of_stimulus
lin quadr
species_subject same_different increase_decrease_stimulus
species_subject*lin same_different*lin increase_decrease_stimulus*lin
species_subject*quadr same_different*quadr increase_decrease_stimulus*quadr
species_subject*same_different
species_subject*increase_decrease_stimulus
same_different*increase_decrease_stimulus
species_subject *same_different*increase_decrease_stimulus
species_subject*same_different*lin
species_subject*increase_decrease_stimulus*lin
same_different*increase_decrease_stimulus*lin
species_subject*same_different*quadr
species_subject*increase_decrease_stimulus*quadr
same_different*increase_decrease_stimulus*quadr
species_subject *same_different*increase_decrease_stimulus*lin
species_subject *same_different*increase_decrease_stimulus*quadr
| noint
/ random int lin | subject(id) covtype(DIAG)
/ random int lin | subject(id*condition) covtype(DIAG)
/ repeated = time | subject(id*condition) covtype(AR1)
/ print solution testcov r.


best wishes,
Mariska