Crossover design in SPSS

[joannerachell]

I am doing crossover design to compare the effect of indoor and outdoor running on weightloss among obese people. I'm having a problem in the statistical analysis which, I didn't really know the right way to analyze it in SPSS. Do I use the paired sample t-test? How? Sorry, I'm still in the learning process and getting a little help from you guys should be great. Thank you in advance.

[Bruce Weaver]

Do you mean a simple 2x2 design like this?

Group 1:  Indoor first, outdoor second
Group 2:  Outdoor first, indoor second

If so, I would structure the data file like this:

ID  Order  Indoor  Outdoor
1     1          Y1         Y2
2     2          Y1         Y2
3     1          Y1         Y2
4     2          Y1         Y2
etc

Then, I would use GLM-Repeated Measures with Location (indoor v outdoor) as the within-Ss factor and Order as the between-Ss factor.  Your main questions, I assume, are whether there is a difference between indoor and outdoor, and whether it depends on the order in which they occurred.  Those questions are addressed by the F-tests for Location and the Order*Location interaction.  

Finally, note that there is a lot of variation in how people conceptualize and describe even this simple design--e.g., in some fields, people would talk about the "period" effect rather than the order, as I have done here.  Jerry Dallal talks about this in his notes on crossover studies:

  http://www.jerrydallal.com/LHSP/crossovr.htm

You'll want to use the conceptual approach that is familiar to people in your field.  

HTH.

joannerachell wrote

I am doing crossover design to compare the effect of indoor and outdoor running on weightloss among obese people. I'm having a problem in the statistical analysis which, I didn't really know the right way to analyze it in SPSS. Do I use the paired sample t-test? How? Sorry, I'm still in the learning process and getting a little help from you guys should be great. Thank you in advance.

[joannerachell]

Thank you guys. Yeap it is simple 2x2 design and I'm using the design for the first time so I really didn't know.

On May 13, 2015 9:36 PM, "Bruce Weaver [via SPSSX Discussion]" <[hidden email]> wrote:

Do you mean a simple 2x2 design like this?

Group 1:  Indoor first, outdoor second
Group 2:  Outdoor first, indoor second

If so, I would structure the data file like this:

ID  Order  Indoor  Outdoor
1     1          Y1         Y2
2     2          Y1         Y2
3     1          Y1         Y2
4     2          Y1         Y2
etc

Then, I would use GLM-Repeated Measures with Location (indoor v outdoor) as the within-Ss factor and Order as the between-Ss factor.  Your main questions, I assume, are whether there is a difference between indoor and outdoor, and whether it depends on the order in which they occurred.  Those questions are addressed by the F-tests for Location and the Order*Location interaction.  

Finally, note that there is a lot of variation in how people conceptualize and describe even this simple design--e.g., in some fields, people would talk about the "period" effect rather than the order, as I have done here.  Jerry Dallal talks about this in his notes on crossover studies:

  http://www.jerrydallal.com/LHSP/crossovr.htm

You'll want to use the conceptual approach that is familiar to people in your field.  

HTH.

joannerachell wrote

I am doing crossover design to compare the effect of indoor and outdoor running on weightloss among obese people. I'm having a problem in the statistical analysis which, I didn't really know the right way to analyze it in SPSS. Do I use the paired sample t-test? How? Sorry, I'm still in the learning process and getting a little help from you guys should be great. Thank you in advance.

--
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.

---

If you reply to this email, your message will be added to the discussion below:

http://spssx-discussion.1045642.n5.nabble.com/Crossover-design-in-SPSS-tp5729553p5729554.html

To unsubscribe from Crossover design in SPSS, click here.
NAML

[parisec]

I found this to be a good article that describes what Bruce is suggesting. I was able to input the data in the article and replicate the results.

http://ligarto.org/rdiaz/Papers/cross-over-animal-behaviour.pdf

[gtripepi]

Dear Bruce,
I am dealing with a cross-over trial in which patients are allocated in
random order to a standard therapy and after that to a new therapy or
viceversa (according to the random order, 0: standard therapy first; 1=new
therapy first). For each patient, I have 3 measurements of vitamin D during
standard therapy and 3 measurements of vitamin D during new therapy (see
below, a short example of the file)

ID     order  VitD_t0_standard VitD_t1_standard VitD_t2_standard VitD_t0_new
VitD_t1_new   VitD_t2_new
1         1          35                    36                        37                          
40            41                43
2         1          34                    34                        33                          
45            44                49
3         0          34                    34                        33                          
45            44                49
............................................................................................................................................

I would use GLM-Repeated Measures with VitD as the within-Ss factor and
Order as the between-Ss factor.  
My main question is whether there is a difference between the changes of
vitamin D during standard treatment and the new treatment , and whether it
depends on the order in which they occurred.  

These questions are addressed by the F-tests for Vitd and the Order*VitD
interaction.

Do you agree ?
Thank you very much in advance !
Giovanni  





--
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]

You (Giovanni) wrote:

"My main question is whether there is a difference between the changes of
vitamin D during standard treatment and the new treatment,"

Your null hypothesis is not stated very clearly here.  I can think of 3 null
hypotheses that you might want to test:

1. The effect of time (whatever it is) is the same for both treatments
(standard and new).  
2. There is no effect of time for the standard treatment.
3. There is no effect of time for the new treatment.

Use the Treatment x Time interaction for 1.  

For 2 and 3, use tests for the simple main effects of time at each level of
Treatment.  These can be obtained via EMMEANS with COMPARE(Time)--see the
sample syntax below.  

You also wrote:

"and whether it depends on the order in which they occurred."

If null hypothesis 1 was what you intended above, then you want the Order x
Treatment x Time interaction here.  It tests the null hypothesis that the
nature of the Treatment x Time interaction (whatever it is) does NOT depend
on the order in which treatments were received.  

The GLM command will look something like this:

GLM st0 st1 st2 new0 new1 new2 BY order
  /WSFACTOR=Treat 2 Time 3
  /METHOD=SSTYPE(3)
  /EMMEANS=TABLES(Treat)
  /EMMEANS=TABLES(Treat*Time) COMPARE(Time)
  /WSDESIGN=Treat Time Treat*Time
  /DESIGN=order.
 
Notice that I replaced your (very) long variable names with shorter (more
manageable) names.  The COMPARE(Time) option on that second EMMEANS
subommand should give you tests two 2-df tests of the simple main effects of
Time at each level of Treatment, if that is what interests you.  

HTH.


gtripepi wrote

> Dear Bruce,
> I am dealing with a cross-over trial in which patients are allocated in
> random order to a standard therapy and after that to a new therapy or
> viceversa (according to the random order, 0: standard therapy first; 1=new
> therapy first). For each patient, I have 3 measurements of vitamin D
> during
> standard therapy and 3 measurements of vitamin D during new therapy (see
> below, a short example of the file)
>
> ID     order  VitD_t0_standard VitD_t1_standard VitD_t2_standard
> VitD_t0_new
> VitD_t1_new   VitD_t2_new
> 1         1          35                    36                        37                          
> 40            41                43
> 2         1          34                    34                        33                          
> 45            44                49
> 3         0          34                    34                        33                          
> 45            44                49
> ............................................................................................................................................
>
> I would use GLM-Repeated Measures with VitD as the within-Ss factor and
> Order as the between-Ss factor.  
> My main question is whether there is a difference between the changes of
> vitamin D during standard treatment and the new treatment , and whether it
> depends on the order in which they occurred.  
>
> These questions are addressed by the F-tests for Vitd and the Order*VitD
> interaction.
>
> Do you agree ?
> Thank you very much in advance !
> Giovanni  
>
>
>
>
>
> --
> 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