repeated measure with some people quit in mid-way

[Rongjin Guan]

Hello Everyone,

I have a repeated measure data, about 150 people in the beginning(time1), and then
at time2 only 110 people got measured, and at time3, only 85 left, and at time4,
only 60 people measured.

Time1 is before treatment, and time2-4 are 3 different time points after treatment.
We want to know if the treatment is effective or not.

My question is: in this scenario, the best method to choose is mixed model, not the
repeated measure ANOVA, isn't it?

If I use mixed model, do I need to impute the missing with LOCF method?


Thank you.

Rongjin  Guan
=====================
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]

What is LOCF method? LOCF is not a word used in the syntax reference.
Gene Maguin

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Rongjin Guan
Sent: Thursday, August 27, 2015 4:09 PM
To: [hidden email]
Subject: repeated measure with some people quit in mid-way

Hello Everyone,

I have a repeated measure data, about 150 people in the beginning(time1), and then at time2 only 110 people got measured, and at time3, only 85 left, and at time4, only 60 people measured.

Time1 is before treatment, and time2-4 are 3 different time points after treatment.
We want to know if the treatment is effective or not.

My question is: in this scenario, the best method to choose is mixed model, not the repeated measure ANOVA, isn't it?

If I use mixed model, do I need to impute the missing with LOCF method?


Thank you.

Rongjin  Guan
=====================
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

[Rongjin Guan]

it is "last observation carrying forward". It is often used in clinical studies.
sorry for the abbr. that is not frequently used here.


________________________________________
From: SPSSX(r) Discussion [[hidden email]] On Behalf Of Maguin, Eugene [[hidden email]]
Sent: Thursday, August 27, 2015 4:18 PM
To: [hidden email]
Subject: Re: repeated measure with some people quit in mid-way

What is LOCF method? LOCF is not a word used in the syntax reference.
Gene Maguin

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Rongjin Guan
Sent: Thursday, August 27, 2015 4:09 PM
To: [hidden email]
Subject: repeated measure with some people quit in mid-way

Hello Everyone,

I have a repeated measure data, about 150 people in the beginning(time1), and then at time2 only 110 people got measured, and at time3, only 85 left, and at time4, only 60 people measured.

Time1 is before treatment, and time2-4 are 3 different time points after treatment.
We want to know if the treatment is effective or not.

My question is: in this scenario, the best method to choose is mixed model, not the repeated measure ANOVA, isn't it?

If I use mixed model, do I need to impute the missing with LOCF method?


Thank you.

Rongjin  Guan
=====================
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

=====================
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

[Ryan]

Responses below.

Sent from my iPhone

> On Aug 27, 2015, at 4:08 PM, Rongjin Guan <[hidden email]> wrote:
>
> Hello Everyone,
>
> I have a repeated measure data, about 150 people in the beginning(time1), and then
> at time2 only 110 people got measured, and at time3, only 85 left, and at time4,
> only 60 people measured.

Attrition is not atypical. The question you should ask yourself is whether or not the pattern is random.

>
> Time1 is before treatment, and time2-4 are 3 different time points after treatment.
> We want to know if the treatment is effective or not.
>
> My question is: in this scenario, the best method to choose is mixed model, not the
> repeated measure ANOVA, isn't it?

Yes, a mixed model would be preferable since you do not lose all data from a subject just because data are missing from one or more time points when estimating the parameters. You should know, however, that a mixed model assumes the data are missing completely at random or at least missing at random.
>
> If I use mixed model, do I need to impute the missing with LOCF method?

1. You do not "need" to impute data to employ a mixed model when there is missing data.
2. If you are going to impute data, multiple imputation is preferable to LOCF.

>
>
> Thank you.
>
> Rongjin  Guan
> =====================
> 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

[Maguin, Eugene]

In addition to what Ryan just said, I urge you to so some background reading on modern missing data methods. A number of people have written articles, chapters and books on the topic.
Gene Maguin


-----Original Message-----
From: Rongjin Guan [mailto:[hidden email]]
Sent: Thursday, August 27, 2015 4:21 PM
To: Maguin, Eugene <[hidden email]>; [hidden email]
Subject: RE: repeated measure with some people quit in mid-way

it is "last observation carrying forward". It is often used in clinical studies.
sorry for the abbr. that is not frequently used here.


________________________________________
From: SPSSX(r) Discussion [[hidden email]] On Behalf Of Maguin, Eugene [[hidden email]]
Sent: Thursday, August 27, 2015 4:18 PM
To: [hidden email]
Subject: Re: repeated measure with some people quit in mid-way

What is LOCF method? LOCF is not a word used in the syntax reference.
Gene Maguin

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Rongjin Guan
Sent: Thursday, August 27, 2015 4:09 PM
To: [hidden email]
Subject: repeated measure with some people quit in mid-way

Hello Everyone,

I have a repeated measure data, about 150 people in the beginning(time1), and then at time2 only 110 people got measured, and at time3, only 85 left, and at time4, only 60 people measured.

Time1 is before treatment, and time2-4 are 3 different time points after treatment.
We want to know if the treatment is effective or not.

My question is: in this scenario, the best method to choose is mixed model, not the repeated measure ANOVA, isn't it?

If I use mixed model, do I need to impute the missing with LOCF method?


Thank you.

Rongjin  Guan
=====================
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

=====================
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]

Here is one article (by David Streiner) that addresses LOCF specifically.

https://www.researchgate.net/publication/5628871_Missing_data_and_the_trouble_with_LOCF

Maguin, Eugene wrote

In addition to what Ryan just said, I urge you to so some background reading on modern missing data methods. A number of people have written articles, chapters and books on the topic.
Gene Maguin


-----Original Message-----
From: Rongjin Guan [mailto:[hidden email]]
Sent: Thursday, August 27, 2015 4:21 PM
To: Maguin, Eugene <[hidden email]>; [hidden email]
Subject: RE: repeated measure with some people quit in mid-way

it is "last observation carrying forward". It is often used in clinical studies.
sorry for the abbr. that is not frequently used here.


________________________________________
From: SPSSX(r) Discussion [[hidden email]] On Behalf Of Maguin, Eugene [[hidden email]]
Sent: Thursday, August 27, 2015 4:18 PM
To: [hidden email]
Subject: Re: repeated measure with some people quit in mid-way

What is LOCF method? LOCF is not a word used in the syntax reference.
Gene Maguin

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Rongjin Guan
Sent: Thursday, August 27, 2015 4:09 PM
To: [hidden email]
Subject: repeated measure with some people quit in mid-way

Hello Everyone,

I have a repeated measure data, about 150 people in the beginning(time1), and then at time2 only 110 people got measured, and at time3, only 85 left, and at time4, only 60 people measured.

Time1 is before treatment, and time2-4 are 3 different time points after treatment.
We want to know if the treatment is effective or not.

My question is: in this scenario, the best method to choose is mixed model, not the repeated measure ANOVA, isn't it?

If I use mixed model, do I need to impute the missing with LOCF method?


Thank you.

Rongjin  Guan
=====================
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

=====================
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