How to run a GEE in SPSS with a Tweedie distribution?

[lsummer]

Hi,
I like to run a GEE as I have panel data of movement behaviour of people
(per person I have multiple trips, for example the detour ratio). I like to
estimate the effects of the direct distance on the detour ratio, while
taking into account that there might be differences between first trip ..
fourth trip -and so on.
The problem is that detour ratio is often 0, and in a few cases it can be
very high, thus a tweedie distribution is the most appropriate to use. To
start I used a Normal distribution with Identity function, which gave
results (but as expected, not very good ones).
However, I receive an error when I use the tweedie distribution: "There is
at least one invalid case in the last iteration. A case is invalid if there
are errors in computing the inverse identity link function, the
log-likelihood, the gradient, or the Hessian matrix in the iterative
process. Only the iteration history is displayed."
Is there somebody with experience in GEE, tweedie distribution, and SPSS?
Any tips would be of great help!

PS: I originally posted this question on Research gate, where they referred
me to this website.
https://www.researchgate.net/post/How_to_run_a_GEE_in_SPSS_with_a_Tweedie_distribution

Thank you!



--
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

[lsummer]

Hi,

I like to run a GEE as I have panel data of movement behaviour of people
(per person I have multiple trips, for example the detour ratio). I like to
estimate the effect of the direct distance on the detour ratio, while taking
into account that there might be differences between first trip .. fourth
trip -and so on.

The problem is that detour ratio is often 0, and in a few cases it can be
very high, thus a tweedie distribution is the most appropriate to use. To
start I used a Normal distribution with Identity function, which gave
results (but as expected, not very good ones).
However, I receive an error when I use the tweedie distribution: "There is
at least one invalid case in the last iteration. A case is invalid if there
are errors in computing the inverse identity link function, the
log-likelihood, the gradient, or the Hessian matrix in the iterative
process. Only the iteration history is displayed."
Is there somebody with experience in GEE, tweedie distribution, and SPSS?
Any tips would be of great help!

Thank you very much!
Lara Zomer

PS: originally I posted this question on RG:
https://www.researchgate.net/post/How_to_run_a_GEE_in_SPSS_with_a_Tweedie_distribution#view=5d2ceac5979fdc914b41bc7c




--
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

[Bruce Weaver]

Hello Lara.  I have used GENLIN with GEE, but never with the Tweedie
distribution.  If you post your command syntax (i.e., your GENLIN command)
and a small sample of what the data file looks like, that might help other
members of this forum to understand better what you are doing.  The DATA
LIST command is a convenient way to create a small sample data set that
others will be able to read in quickly.  E.g.,

DATA LIST LIST / {list of variable names here}.
BEGIN DATA
{ your data here, one row per unit of observation }
END DATA.

Based on what you said, I expect you will have multiple rows per ID with
Trip as an index number within ID.  

Meanwhile, that error message--"There is at least one invalid case in the
last iteration"--is making me think that one could maybe write a macro to
loop through all cases in the file, leaving one out each time to try to
identify the offending case.  (This would only work if there is only one
offending case, I suppose.)  

HTH.



lsummer wrote

> Hi,
>
> I like to run a GEE as I have panel data of movement behaviour of people
> (per person I have multiple trips, for example the detour ratio). I like
> to
> estimate the effect of the direct distance on the detour ratio, while
> taking
> into account that there might be differences between first trip .. fourth
> trip -and so on.
>
> The problem is that detour ratio is often 0, and in a few cases it can be
> very high, thus a tweedie distribution is the most appropriate to use. To
> start I used a Normal distribution with Identity function, which gave
> results (but as expected, not very good ones).
> However, I receive an error when I use the tweedie distribution: "There is
> at least one invalid case in the last iteration. A case is invalid if
> there
> are errors in computing the inverse identity link function, the
> log-likelihood, the gradient, or the Hessian matrix in the iterative
> process. Only the iteration history is displayed."
> Is there somebody with experience in GEE, tweedie distribution, and SPSS?
> Any tips would be of great help!
>
> Thank you very much!
> Lara Zomer
>
> PS: originally I posted this question on RG:
> https://www.researchgate.net/post/How_to_run_a_GEE_in_SPSS_with_a_Tweedie_distribution#view=5d2ceac5979fdc914b41bc7c
>
>
>
>
> --
> Sent from: http://spssx-discussion.1045642.n5.nabble.com/
>
> =====================
> To manage your subscription to SPSSX-L, send a message to

> LISTSERV@.UGA

>  (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





-----
--
Bruce Weaver
[hidden email]
http://sites.google.com/a/lakeheadu.ca/bweaver/

"When all else fails, RTFM."

NOTE: My Hotmail account is not monitored regularly.
To send me an e-mail, please use the address shown above.

--
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