SPSS Statistics

SPSS Statistics

Your hub for statistical analysis, data management, and data documentation. Connect, learn, and share with your peers! 

 View Only
Back to discussions
Expand all | Collapse all

Interrupted Time Series Analysis

  • 1.  Interrupted Time Series Analysis

    Posted Mon October 18, 2021 10:08 AM
    Edited by System Admin Fri January 20, 2023 04:08 PM

    Hi all. I have conducted an Interrupted Time Series Analysis using the forecasting option/ARIMA model.

    For some reason the predictor variable is being omitted from the model in the output. I am pretty sure I have done everything correctly although do not understand why this is happening and wondering if anyone could help? Does this mean the predictor does not fit the model? Any input is much welcomed!



    ------------------------------
    Carolina Antonini
    ------------------------------
    #SPSSStatistics


  • 2.  RE: Interrupted Time Series Analysis

    Posted Mon October 18, 2021 02:37 PM
    Hi, Carolina. Would it be possible for you to share either a sample or simulation of your data and the syntax you used? Thanks.

    ------------------------------
    Rick Marcantonio
    Quality Assurance
    IBM
    ------------------------------



  • 3.  RE: Interrupted Time Series Analysis

    Posted Mon October 18, 2021 03:03 PM
    What you see in the output will depend on the type of model and method.  Be sure to check the coefficients box under Output to see the model details.

    ------------------------------
    Jon Peck
    ------------------------------

    Original Message


  • 4.  RE: Interrupted Time Series Analysis

    Posted Mon October 18, 2021 03:43 PM
    Hi Jon. I suspect I am looking at this the wrong way as not familiar with Interrupted Time Series Analysis with ARIMA models. I want to know if my intervention had an effect of incident rates but the model in the output "kicked out" my predictor (i.e. intervention) so not too sure whether I have done something wrong.
    I have attached the output.

    ------------------------------
    Carolina Antonini
    ------------------------------

    Original Message


  • 5.  RE: Interrupted Time Series Analysis

    Posted Mon October 18, 2021 03:37 PM
      |   view attached
    Hi Rick. Thank you for your answer!
    I am evaluating an intervention and looking to find out whether levels of incidents (e.g. verbal abuse, harassment, etc.) have reduced following the intervention. I recorded the number of incidents on a weekly basis for 6 months prior to and following the intervention.
    So the DV is number of incidents and the IV is is the intervention. I have attached the output - as you will see, there is no predictor (i.e. intervention) in the model despite having included this in the analysis. 
    I have conducted the same analysis (step by step) with another DV (i.e. incidents of harassment) and the predictor (intervention) is retained in the model.
    I am new to ITSA so I may be looking at this the wrong way!





    ------------------------------
    Carolina Antonini
    ------------------------------

    Attachment(s)

    docx
    SPSS Output.docx   60 KB 1 version
    Original Message


  • 6.  RE: Interrupted Time Series Analysis

    Posted Mon October 18, 2021 03:48 PM
    I mocked up an ITS analysis and used:

    TSMODEL
   /MODELSUMMARY  PRINT=[MODELFIT]
   /MODELSTATISTICS  DISPLAY=YES MODELFIT=[ SRSQUARE]
   /MODELDETAILS PRINT=[PARAMETERS]
   /SERIESPLOT OBSERVED FORECAST
   /OUTPUTFILTER DISPLAY=ALLMODELS
   /AUXILIARY  CILEVEL=95 MAXACFLAGS=24
   /MISSING USERMISSING=EXCLUDE
   /MODEL DEPENDENT=time INDEPENDENT=drug  PREFIX='Model'
   /ARIMA AR=[0]  DIFF=0  MA=[0]  TRANSFORM=NONE  CONSTANT=YES
   /AUTOOUTLIER DETECT=OFF.​

    Do you have /MODELDETAILS PRINT=[PARAMETERS]? I didn't see that table in your output.

    ------------------------------
    Rick Marcantonio
    Quality Assurance
    IBM
    ------------------------------

    Original Message


  • 7.  RE: Interrupted Time Series Analysis

    Posted Mon October 18, 2021 03:59 PM
    urgh. For some reason the parameters are not showing up..I checked all the relevant boxes but couldn't address the issue. I have attached the actual output - maybe that will help.
    ps:I am only using ARIMA models.

    ------------------------------
    Carolina Antonini
    ------------------------------

    Original Message


  • 8.  RE: Interrupted Time Series Analysis

    Posted Mon October 18, 2021 04:40 PM
    OK. I don't know what's going on. Can you please paste the syntax and send that to me so I can look at it? I have tried various things but the independent variable always shows up in the Model Parameters table, whether significant or not.

    ------------------------------
    Rick Marcantonio
    Quality Assurance
    IBM
    ------------------------------

    Original Message


  • 9.  RE: Interrupted Time Series Analysis

    Posted Wed April 02, 2025 08:19 AM

    When I use this syntax, I get no estimates for my event independent variable. Please advice how I can estimate and test the effect of my "event".

    TSMODEL
       /MODELSUMMARY  PRINT=[MODELFIT]
       /MODELSTATISTICS  DISPLAY=NO 
       /MODELDETAILS  PRINT=[ PARAMETERS RESIDACF RESIDPACF]  PLOT=[ RESIDACF RESIDPACF]
       /SERIESPLOT OBSERVED
       /OUTPUTFILTER DISPLAY=ALLMODELS
       /AUXILIARY  CILEVEL=95 MAXACFLAGS=24
       /MISSING USERMISSING=EXCLUDE
       /MODEL DEPENDENT=SAkgha INDEPENDENT=GMevent [E] 
          PREFIX='Model'
       /EXPERTMODELER TYPE=[ARIMA]
       /AUTOOUTLIER  DETECT=OFF.



    ------------------------------
    Andrew Chapman
    ------------------------------

    Original Message


  • 10.  RE: Interrupted Time Series Analysis

    Posted Wed April 02, 2025 08:20 AM

    Hi All, I have a time series with an "event" independent variable which just gets ignored in the parameter table. I have tried different ways to use the options. Please help me. My basic syntax is this:

    TSMODEL
       /MODELSUMMARY  PRINT=[MODELFIT]
       /MODELSTATISTICS  DISPLAY=NO 
       /MODELDETAILS  PRINT=[ PARAMETERS RESIDACF RESIDPACF]  PLOT=[ RESIDACF RESIDPACF]
       /SERIESPLOT OBSERVED
       /OUTPUTFILTER DISPLAY=ALLMODELS
       /AUXILIARY  CILEVEL=95 MAXACFLAGS=24
       /MISSING USERMISSING=EXCLUDE
       /MODEL DEPENDENT=SAkgha INDEPENDENT=GMevent [E] 
          PREFIX='Model'
       /EXPERTMODELER TYPE=[ARIMA]
       /AUTOOUTLIER  DETECT=OFF.



    ------------------------------
    Andrew Chapman
    ------------------------------

    Original Message