Variance partition coeffecient using GENLINMIXED

classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|

Variance partition coeffecient using GENLINMIXED

nina
Dear all,

it is not clear to me how to compute a variance partition coefficient for a
dichotomous dependent variable using GENLINMIXED in SPSS 24.

For example, assume I receive the score 1.13 for the intercept variance of a
null model (1.13 is contained in the table "Random effect", first column
labelled "Random effect covariance", second  column labelled "estimate", row
labelled "Var [Intercept]").

Would it be correct to consider this score to (a) indicate the random
variance of the intercepts var(u0j)?

Then I would  calculate the VPC as var(u0j) / [var(u0j) + (π2 / 3)] = 1.13 /
(1.13 + 3.29) = 25.5%? (e.g. Snijders & Bosker 1999)?

Best,
Nina



--
Sent from: http://spssx-discussion.1045642.n5.nabble.com/

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD
Reply | Threaded
Open this post in threaded view
|

Re: Variance partition coeffecient using GENLINMIXED

Ryan Black
Looks right to me. -Ryan

> On Nov 25, 2018, at 6:02 AM, nina <[hidden email]> wrote:
>
> Dear all,
>
> it is not clear to me how to compute a variance partition coefficient for a
> dichotomous dependent variable using GENLINMIXED in SPSS 24.
>
> For example, assume I receive the score 1.13 for the intercept variance of a
> null model (1.13 is contained in the table "Random effect", first column
> labelled "Random effect covariance", second  column labelled "estimate", row
> labelled "Var [Intercept]").
>
> Would it be correct to consider this score to (a) indicate the random
> variance of the intercepts var(u0j)?
>
> Then I would  calculate the VPC as var(u0j) / [var(u0j) + (π2 / 3)] = 1.13 /
> (1.13 + 3.29) = 25.5%? (e.g. Snijders & Bosker 1999)?
>
> Best,
> Nina
>
>
>
> --
> Sent from: http://spssx-discussion.1045642.n5.nabble.com/
>
> =====================
> To manage your subscription to SPSSX-L, send a message to
> [hidden email] (not to SPSSX-L), with no body text except the
> command. To leave the list, send the command
> SIGNOFF SPSSX-L
> For a list of commands to manage subscriptions, send the command
> INFO REFCARD

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD