Intercept variance (Mixed modeling)

[Oliver]

Hi everyone,

I am running a multilevel linear regression with one continuous Level 1
independent varialbe (x). As can be seen in the syntax below, a random
intercept is specified.  I have a question regarding the interpretation of
this random effect, because the value generated in the output (Intercept
[subject = ID] Variance is substantially greater than the intercept value
from the fixed effect. I understand that the intercept value from the fixed
effect represents the average outcome value when X=0 and that the random
effect is different, but the difference in these values is extremely large
(i.e., fixed intercept value = *4.4*;  Intercept variance = *385.6*).
Perhaps this huge intercept variance simply reflects the huge between-person
differences in mean outcome values, but I'm still wondering if this type of
difference can happen.

Here's the syntax below:

MIXED y WITH x
/PRINT=SOLUTION TESTCOV
/METHOD=ML
/FIXED= x
/RANDOM INTERCEPT | SUBJECT(ID) COVTYPE (ID)
/REPEATED=Wave | SUBJECT(ID) COVTYPE(AR1) .

Thanks in advance for your assistance.
O.



--
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

[Maguin, Eugene]

Oliver,
I'm curious. Why do you have a repeated specification as well as a random specification? What is the structure of your data?
Gene Maguin

-----Original Message-----
From: SPSSX(r) Discussion <[hidden email]> On Behalf Of Oliver
Sent: Thursday, October 29, 2020 10:02 AM
To: [hidden email]
Subject: Intercept variance (Mixed modeling)

Hi everyone,

I am running a multilevel linear regression with one continuous Level 1 independent varialbe (x). As can be seen in the syntax below, a random intercept is specified.  I have a question regarding the interpretation of this random effect, because the value generated in the output (Intercept [subject = ID] Variance is substantially greater than the intercept value from the fixed effect. I understand that the intercept value from the fixed effect represents the average outcome value when X=0 and that the random effect is different, but the difference in these values is extremely large (i.e., fixed intercept value = *4.4*;  Intercept variance = *385.6*).
Perhaps this huge intercept variance simply reflects the huge between-person differences in mean outcome values, but I'm still wondering if this type of difference can happen.

Here's the syntax below:

MIXED y WITH x
/PRINT=SOLUTION TESTCOV
/METHOD=ML
/FIXED= x
/RANDOM INTERCEPT | SUBJECT(ID) COVTYPE (ID) /REPEATED=Wave | SUBJECT(ID) COVTYPE(AR1) .

Thanks in advance for your assistance.
O.



--
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

[Oliver]

Hi Gene,

Thanks for getting back to me.  I have probably not provided enough details
about the study design, but the outcome is assessed repeatedly (i.e., 14
times) for each ID (i.e., within-person). Same for the Level 1 independent
variable. The repeated subcommand was thus used to specify the repeated
measures.  
Thanks again for your assistance.
O.



--
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