Question about analysis of longitudinal data

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

Question about analysis of longitudinal data

Norberto Hernandez
Hello everybody!! I'm asking for your kind advice because I have a consolidated weekly data of hospital admissions for the las 2 years (104 observation points) of 32 cities. I'm looking to estimate the mean growth of hospital admissions over this period, but a visual inspection of the data suggest that some cities shows a growth in hospital admission but others has stable records over time (admissions did not vary over time)

My first idea was use LGM, but i have no access to Mplus or Lisrel (I have no licence of this packages), so I want tos ask you suggest me any idea for analyse this set.

Any help or reference will be gratefully received.

Kind Regards
Norberto
===================== 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: Question about analysis of longitudinal data

Dr Chris Stride
Longitudinal MLM would be the most appropriate choice, observations (timepts) nested within cities. Will give equivalent results to LGM at most basic level.

From: [hidden email]
Sent: ‎24/‎10/‎2016 17:37
To: [hidden email]
Subject: Question about analysis of longitudinal data

Hello everybody!! I'm asking for your kind advice because I have a consolidated weekly data of hospital admissions for the las 2 years (104 observation points) of 32 cities. I'm looking to estimate the mean growth of hospital admissions over this period, but a visual inspection of the data suggest that some cities shows a growth in hospital admission but others has stable records over time (admissions did not vary over time)

My first idea was use LGM, but i have no access to Mplus or Lisrel (I have no licence of this packages), so I want tos ask you suggest me any idea for analyse this set.

Any help or reference will be gratefully received.

Kind Regards
Norberto
===================== 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
Reply | Threaded
Open this post in threaded view
|

Re: Question about analysis of longitudinal data

Norberto Hernandez
Thank you very much Chris, but I am not sure if MLM could be used. I "knew" that MLM relies in the a priori specification of the different groups, and in my case the hospitals with "growing admission" and "stable admissions" are not specified. Am I correct?? Or I need to classify hospitals in "growing admission" and "stable admissions" before MLM analisys.

Thanks a lot

2016-10-24 12:01 GMT-05:00 Chris Stride <[hidden email]>:
Longitudinal MLM would be the most appropriate choice, observations (timepts) nested within cities. Will give equivalent results to LGM at most basic level.

From: [hidden email]
Sent: ‎24/‎10/‎2016 17:37
To: [hidden email]
Subject: Question about analysis of longitudinal data

Hello everybody!! I'm asking for your kind advice because I have a consolidated weekly data of hospital admissions for the las 2 years (104 observation points) of 32 cities. I'm looking to estimate the mean growth of hospital admissions over this period, but a visual inspection of the data suggest that some cities shows a growth in hospital admission but others has stable records over time (admissions did not vary over time)

My first idea was use LGM, but i have no access to Mplus or Lisrel (I have no licence of this packages), so I want tos ask you suggest me any idea for analyse this set.

Any help or reference will be gratefully received.

Kind Regards
Norberto
===================== 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
Reply | Threaded
Open this post in threaded view
|

Re: Question about analysis of longitudinal data

Dr Chris Stride

Your hospital IDs are your higher level grouping variable. Hospital type (e.g. growing vs stable) could be modelled using this as dichotomous hospital-level predictor

e.g.

MIXED ADMISSIONS BY TYPE WITH TIMEPOINT
 /FIXED = TIMEPOINT TYPE
 /RANDOM = INTERCEPT | SUBJECT(HOSPID)
 /REPEATED = TIMEPOINT | SUBJECT(HOSPID) COVTYPE(AR1)
 /PRINT SOLUTION TESTCOV.

is an example of the sort of syntax you need... but you need to go away read about MLM, longitudinal MLM, issues around variance partitioning, centering, the sequence of models to run, modelling autocorrelation patterns (that's what the /repeated command does here), etc, etc




On 24/10/2016 18:39, Norberto Hernandez wrote:
Thank you very much Chris, but I am not sure if MLM could be used. I "knew" that MLM relies in the a priori specification of the different groups, and in my case the hospitals with "growing admission" and "stable admissions" are not specified. Am I correct?? Or I need to classify hospitals in "growing admission" and "stable admissions" before MLM analisys.

Thanks a lot

2016-10-24 12:01 GMT-05:00 Chris Stride <[hidden email]>:
Longitudinal MLM would be the most appropriate choice, observations (timepts) nested within cities. Will give equivalent results to LGM at most basic level.

From: [hidden email]
Sent: ‎24/‎10/‎2016 17:37
To: [hidden email]
Subject: Question about analysis of longitudinal data

Hello everybody!! I'm asking for your kind advice because I have a consolidated weekly data of hospital admissions for the las 2 years (104 observation points) of 32 cities. I'm looking to estimate the mean growth of hospital admissions over this period, but a visual inspection of the data suggest that some cities shows a growth in hospital admission but others has stable records over time (admissions did not vary over time)

My first idea was use LGM, but i have no access to Mplus or Lisrel (I have no licence of this packages), so I want tos ask you suggest me any idea for analyse this set.

Any help or reference will be gratefully received.

Kind Regards
Norberto
===================== 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

-- 

--

Dr Chris Stride, C. Stat, Statistician, Institute of Work Psychology,
University of Sheffield
Telephone: 0114 2223262
Fax: 0114 2727206

"Figure It Out"
Statistical Consultancy and Training Service for Social Scientists

Visit www.figureitout.org.uk for details of my consultancy services, and
forthcoming training courses, which are also available on an in-house basis:

 - Data management using SPSS syntax
 - Advanced SPSS syntax and SPSS macros
 - Testing for Mediation and Moderation using SPSS
 - Multi-level Modelling using SPSS
 - Introduction to Structural Equation Modelling using Mplus
 - Testing for Mediation and Moderation using Mplus
 - Multi-level Modelling using Mplus
 - Latent Growth Curve Modelling using Mplus 
===================== 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: Question about analysis of longitudinal data

Rich Ulrich
In reply to this post by Norberto Hernandez

The advisors here seem to be jumping ahead of the measurements.


A paired t-test for year 1 to year 2, for each hospital, will give the 32 averages, per week, for admissions, along with a

t-test whose p-value gives an indication of growth (positive or negative) and how systematic it is.  Comparing year 1 to

year 2 seems like an obvious way to present the data, so this is a start.  The t-test is useful because you can't readily

say that an "increase of 1.1"  per week is as meaningful in a large city as in a small one.


Auto-regression is a popular way to model, but it might be overkill at the start.  Do you care if there are seasonal trends?

--
Rich Ulrich


From: SPSSX(r) Discussion <[hidden email]> on behalf of Norberto Hernandez <[hidden email]>
Sent: Monday, October 24, 2016 1:39 PM
To: [hidden email]
Subject: Re: Question about analysis of longitudinal data
 
Thank you very much Chris, but I am not sure if MLM could be used. I "knew" that MLM relies in the a priori specification of the different groups, and in my case the hospitals with "growing admission" and "stable admissions" are not specified. Am I correct?? Or I need to classify hospitals in "growing admission" and "stable admissions" before MLM analisys.

Thanks a lot

2016-10-24 12:01 GMT-05:00 Chris Stride <[hidden email]>:
Longitudinal MLM would be the most appropriate choice, observations (timepts) nested within cities. Will give equivalent results to LGM at most basic level.

From: [hidden email]
Sent: ‎24/‎10/‎2016 17:37
To: [hidden email]
Subject: Question about analysis of longitudinal data

Hello everybody!! I'm asking for your kind advice because I have a consolidated weekly data of hospital admissions for the las 2 years (104 observation points) of 32 cities. I'm looking to estimate the mean growth of hospital admissions over this period, but a visual inspection of the data suggest that some cities shows a growth in hospital admission but others has stable records over time (admissions did not vary over time)

My first idea was use LGM, but i have no access to Mplus or Lisrel (I have no licence of this packages), so I want tos ask you suggest me any idea for analyse this set.

Any help or reference will be gratefully received.

Kind Regards
Norberto
===================== 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
Reply | Threaded
Open this post in threaded view
|

Re: Question about analysis of longitudinal data

Norberto Hernandez
Yhank you very much Chris, Davis and Rich. Actually it is possible to have seasonal trend, I'm thinking using ARIMA for the description of overall trend, but I don't know if ARIMA may be used for compare admission by year.

2016-10-24 16:59 GMT-05:00 Rich Ulrich <[hidden email]>:

The advisors here seem to be jumping ahead of the measurements.


A paired t-test for year 1 to year 2, for each hospital, will give the 32 averages, per week, for admissions, along with a

t-test whose p-value gives an indication of growth (positive or negative) and how systematic it is.  Comparing year 1 to

year 2 seems like an obvious way to present the data, so this is a start.  The t-test is useful because you can't readily

say that an "increase of 1.1"  per week is as meaningful in a large city as in a small one.


Auto-regression is a popular way to model, but it might be overkill at the start.  Do you care if there are seasonal trends?

--
Rich Ulrich


From: SPSSX(r) Discussion <[hidden email]> on behalf of Norberto Hernandez <[hidden email]>
Sent: Monday, October 24, 2016 1:39 PM
To: [hidden email]
Subject: Re: Question about analysis of longitudinal data
 
Thank you very much Chris, but I am not sure if MLM could be used. I "knew" that MLM relies in the a priori specification of the different groups, and in my case the hospitals with "growing admission" and "stable admissions" are not specified. Am I correct?? Or I need to classify hospitals in "growing admission" and "stable admissions" before MLM analisys.

Thanks a lot

2016-10-24 12:01 GMT-05:00 Chris Stride <[hidden email]>:
Longitudinal MLM would be the most appropriate choice, observations (timepts) nested within cities. Will give equivalent results to LGM at most basic level.

From: [hidden email]
Sent: ‎24/‎10/‎2016 17:37
To: [hidden email]
Subject: Question about analysis of longitudinal data

Hello everybody!! I'm asking for your kind advice because I have a consolidated weekly data of hospital admissions for the las 2 years (104 observation points) of 32 cities. I'm looking to estimate the mean growth of hospital admissions over this period, but a visual inspection of the data suggest that some cities shows a growth in hospital admission but others has stable records over time (admissions did not vary over time)

My first idea was use LGM, but i have no access to Mplus or Lisrel (I have no licence of this packages), so I want tos ask you suggest me any idea for analyse this set.

Any help or reference will be gratefully received.

Kind Regards
Norberto
===================== 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

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