There are three classes A, B and C at the same year of study. I just need one class for treatment and one class for control, so I selected classes A and B as the objects of my research. What is the sampling technique applied in this situation?
Hello Hadiyanto,
Sounds like a convenience sample to me. As you don't indicate how you came to select classes "A" and "B" (and omit "C"), or how you assigned A & B to condition, it sounds as if nothing was random (neither selection nor assignment). As assignment was for the class, not the individual student, your effective sample size (based on classes as sampling units) is 2. Students are nested within classes (and therefore their resulting outcomes or scores are not likely to be completely independent).
Good luck with your work.
Thank you for your contribution. Your answer is truly useful for improving my paper.
Sampling is just the path chosen to be able to discuss any appreciable difference due to intervention. What is the outcome you are willing to measure? by simplifying, convenience sample is suitable for qualitative approaches as individual perception and insights are will be the focus, but need to be carefully managed in quantitative approaches. As such sampling unit and unit of analyses have to be discussed. I believe you will be sampling classes, being the child your unit of analyses.The quasi-experimental design needs to be adapted to children, study power will depend on that. Then you will need to check for differences between or among groups (in case you will decide for two groups of control). Care for self selection (as this can be a confounder). Care for contamination, intervention can dilute over non-participating classes.
Some will not agree with but check for Pawson and Tilley (1997) Realistic evaluation and further use, that will help.
At the moment, you decided that cluster a, b and c are going to be part of the study. You have the participants or sample because the groups must be homogenious. Next, you chose a method to take a and b, so this method can be a convience sample because they are similar or the participants acepted. Another sample if you selected them ramdonly. I think you must focus in the research level that you have longitudinal study because you must implement pre and post test. Then the type of data (categorical or numerical) which leads to the statiscal formula Chi square or t student to determine if the implementation outcomes are significance.
Another point, when you chose a, b or c to be the groups is convinience sample, but the particpants can be chosen using a simple random probabilistic method if some particpants are including to be analized using that inclusion criteria (sampling frame)
Thanks to everyone here for your enlightment, especially to @Hadiyanto Hadiyanto who has raised this question. This is also an issue that has been becoming my question.
Let me expand the case already mentioned by Hardiyanto Hardiyanto. Imagine the researcher intends to do a quasi experiment (the effect of mindmap use toward student writing competency) toward 8th graders. In the school, at the 8th grade there are 4 classes: A, B, C, and D. Simple random sampling is not allowed (intact classes). The school authority only allows two classes to be taken. The researcher then administers a test to get two classes that are not significantly different (let's say via a t-test). I imagine two possible manners in which the samples will be selected,i.e.:
1. The test is administered to all the four classes and let's say the results reveal that B, C, and D are not significantly different to one another. Two classes will be randomly picked up as the samples (then, randomly assigned as the control and experimental group).
2. The test is directly administered to two classes that the school informs as being equal to one another (let's say the school conditons that best students are put as Class A and B, while poorly performing students as Class C and D). The school just asks the researcher to do the test to Class C and D first, and let's say the result shows C and D are not significantly different (equal) and then, they will be randomly assigned as the control and experimental group.
To the above situations, please help me:
a. Is each of option 1 and 2 above correct to be carried out for a quasi experimental research?
b. Often, in the Methods Chapter, students are asked to explain in detail what is their population and sample. Is a population subsection needed in the methods chapter of a quasi-experiment research report; or, can it just be omitted?
c. If option 1 and 2 above are both correct and if a population sub-section is a must in a quasi-experiment research report, what should be reported as the research population for option 1 and 2 above and what is the name of the sampling technique for each of them?
A great deal of thanks for your kind help and contribution.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Sampling Design
- Sample Size