I am currently rather confused about the design type for my study.

I am looking at a number of measures pre and post intervention with control group in a community sample. The sample collected was not through randomization. Is the design type quasi-experimental? What kind of statistics can i use for this design.

Note: Sample size is 20 in each group before and after.

  • $\begingroup$ I may have prematurely edited your question. Were participants randomly assigned to treatment or control? $\endgroup$ – Jeromy Anglim Apr 18 '13 at 10:51
  • $\begingroup$ They are not randomly assigned for both groups. They have taken conveniently $\endgroup$ – user24502 Apr 18 '13 at 14:24
  • $\begingroup$ Okay. I've tweaked the question a little bit to make this clearer. the phrase "not through randomisation" is a little ambiguous in that it could refer to either (a) your sample was not randomly sampled from the population, or (b) you did not randomly assign your participants to groups. $\endgroup$ – Jeromy Anglim Apr 19 '13 at 0:47

If participants will be assigned to the control or treatment groups randomly, this is a run-of-the-mill experiment (not a quasi-experimental design) and you can use all of the usual statistics. On the other hand, if the control group is really a bunch of people who did not qualify for the intervention or were excluded by you or someone else based on some other criteria, then what you have is indeed a quasi-experiment.

Of course, if people registered voluntarily to participate in the study or were screened beforehand, its generalization to a broader population is open to discussion (as are all clinical trials in fact) but the “randomization” part in introductory psychology research method texts usually refers to treatment assignment, not to the fact that the participants form a random sample of anything. The difference therefore does not lie in the type of sample (random or community sample, treatment sample, convenience sample) but in treatment assignment (who goes to the control and who gets the intervention).

If you want more specifics on the analysis you will need to post a lot more details about the study (perhaps in a follow-up question).

  • $\begingroup$ There were both exclusion and inclusion criteria same for both groups except interms of taking intervention and non going for intervention. Can i use parametric statistics for a sample of 20. If so please help me finding reference of that. One more thing if parametric statistics are used which are more appropriate for this method. T or Anova or Ancova? or any other. $\endgroup$ – user24502 Apr 18 '13 at 14:22

I'm not sure I understood your phrase "not through randomisation". This answer assumes that you did not randomly assign participants to groups.

If there was no random assignment of participants to group you could call it quasi-experimental.

Thus, control and treatment groups may differ anyway. The fact that you have pre and post measurement hopefully reduces this issue because to some extent you are adjusting for baseline differences. That said, it doesn't resolve the issue completely.

Assuming that the dependent variable is numeric, you could look at the suggestions on this question about how to analyse such a design.

  • $\begingroup$ It's not clear to me from the description whether assignment is in fact random or not. $\endgroup$ – Gala Apr 18 '13 at 10:49
  • $\begingroup$ Good point. I read it a bit quickly. I guess it depends on what the OP means by "not through randomisation". I've asked for clarification. $\endgroup$ – Jeromy Anglim Apr 18 '13 at 10:50
  • $\begingroup$ I have taken it conveniently, so i called it as non randomized sample. $\endgroup$ – user24502 Apr 18 '13 at 14:19

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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