2
$\begingroup$

I am a psychology student using SPSS to analyse my data.

I have 2 conditions of the IV of treatment (treatment or no treatment) to test pain tolerance levels which is measured by the DV of pulse rate to measure pain.

The study used a repeated measrures design using a counterbalancing approach (e.g. 20 particpants in treatment condition first, wheras other 20 in no treatment condition first, and then in alternate condition in second part of experiment.

I've used paired samples t-tests to test for difference between both conditions, but now i want to test for order effects, to see if particpants who took part in whichever condition first, got higher scores.

Is there any ANOVa's i would use for this and how would i interpret it using what tests?

Improve this question
$\endgroup$
5
  • 1
    $\begingroup$ How long was the "washout" period between treatments? Also are you considering subjects as fixed or random in your experiment? $\endgroup$
    – StatsStudent
    Apr 2, 2015 at 21:02
  • $\begingroup$ hi thanks for replying - the treatments were short term like pain distraction techniques and each condition took place 5 minutes after the previos part, with each particpant taking part indidividualy in each timeslot $\endgroup$
    – Mohammed Badat
    Apr 2, 2015 at 21:09
  • $\begingroup$ so basically 2 iv conditions of treatment and no treatment, with 2 pulse rate mean figures each for treatment and no treatment groups, but i want to figure a further analysis of the order of the condition they underttok first to see whether there were any order effects $\endgroup$
    – Mohammed Badat
    Apr 2, 2015 at 21:11
  • 1
    $\begingroup$ You cannot directly test for order effects and simultaneously block on subjects. You have designed the experiment to be "counterbalance" to offset the order effects during the analysis. It would have been ideal if you could have waited a week or so and then re-run the second treatments on the subjects to help protect against against possible confounding of order effects. $\endgroup$
    – StatsStudent
    Apr 2, 2015 at 22:00
  • $\begingroup$ what process would i use if i were to replaace measuring order effects, using instead gender differences between both conditions etc $\endgroup$
    – Mohammed Badat
    Apr 3, 2015 at 10:14

1 Answer 1

Reset to default
1
$\begingroup$

You have a mixed design. Your IV of interest (treatment/notreatment) is within: all participants experience treatment and nontreatment. However, treatment order is between with two groups (t - nt, nt - t): each participant experiences only one of these orders, not both. You can compare the two groups, but what you probably want to do is test the interaction so you can see if, for example, t helps more when it comes first. You can do a 2x2 repeated measures anova testing the interaction, or you could use a mixed model specifying the subject as a random factor and possibly treatment as a random slope (if it converges). Here is a reference with some nice graphs of the interactions you might find

Improve this answer
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.