Good question! This is a common scenario when designing research studies with small courses. I would agree that a one-group design makes sense rather than splitting the group. The nature of your questions also would likely result in more meaningful findings if you methods were qualitative. I would recommend Sharan Merriam's or Robert Stake's work on conducting a case study. Good luck!

Hi! It is the gold standard to have a control group for any evaluation study and use a randomized trial with a pre-test post-test control group design. Due to feasibility reasons in the real world scenario, however, you may conduct a pre- and post- intervention design in which individuals serve as their own control. It is essential though to have comparative measurements before and after the intervention to check if there are significant differences in the learners' competencies. For educational interventions, it is best to look at the different levels of learning and ensure that measurements are taken per level. You may check models of learning (e.g. Kirkpatrick model) and assign variables that will measure your learning outcomes.

Dear Kharlo Subrio ,

As you don't have a benchmark to measure against you will need to measure one group against another. Thus having a control group would seem to be a good idea. If you are measuring scale data then a t-test can still provide a significant result provided that your effect size is large. However, you should still test its assumptions. Otherwise, if your data is categorical, it is probably wise to reduce it to two categories with approximately equal frequenciees. You can then either carry out a Chi-squared test or a Fisher's exact test. Please see my study guides on independent t tests and the chi-squared test.

Hello. Since your sample is very small, the power of the statistical test would be low. Another way to increase the sample size would be to use the present class as a control group and collect data on it for the criterion variables and use the next class as experimental group and compare their scores with this group. As the students are from the same college the groups can be assumed to be equivalent.

As to me it difficult to decide research design by looking only the number of the participants rather research design depends on the basic questions of the study as well as the philosophy of the researcher. Based on this, the research design may be developmental research design, case study, phenomenological, maybe grounded theory.

Splitting the group into two makes the sample in each group even smaller. As mentioned by other colleagues this limits the power of statistical test although t-test was designed for such small samples.

The idea of approaching the study qualitatively is a viable one. The evaluation committee proposed a control group probably because your research questions are about cause and effect. So, alternatively you may focus on investigating qualitatively what is actually happening and experienced when OBTL is in use.

Hi! This is an interesting question. Based on your research objective, you need to measure the enhancement of OBTL against your other variables. I think having two groups is a good idea. And a t-test can be used here for small samples.

Hi Kharlo,

My suggestion is first to identify your approach? Qualitative, Quantitative or mixed methods. If you select a qualitative approach, then you have selected a small set from the 16 members in your class. A phenomenology study would be ideal based on the length of time for your study. Quantitative would not capture the thoughts of your participants adequately. Mixed methods could be considered but again what is the timeframe to complete the study.

You have received a lot of views so choose the best option that is comfortable for you.

All the best