Hi mate,
If I have a mixed method explanatory sequential research. I have chosen the pragmatic philosophy and I have no theory or hypothesis. The study will explore Math's teachers attitudes towards one type of the evolution they get to ensure their quality. The first phase will be through questionnaire (100 teachers) the sample will be chosen using stratified approach. the second phase interviews (10 teachers) the sample will be (purposive) based on their responses to the questionnaire. The population will be math teachers in one city.
My questions:
1- I do not know the sample of the population so what is the limitation here? could it be representative? can I generalize the findings on the population? or is the study will be exploratory and I can not generalize the findings?
2- What limitations could be in the size of the sample?
thank you all in advance
I mean I am not sure about the number of the population (math teachers in the city) so is it possible to generalize the findings as the 100 sample will be considered as representative? If not what could be the limitations? Is that i am not sure the number of the population will be a problem?
Without describing the quantitative and qualitative methods, you cannot decide what sample size will be needed.
Here are four easy points to use to decide sample size:
1. Define the population and then pick a sample frame. Collect demographic information to sufficiently describe the sample so you can show representation to whatever population is chosen.
2. Data adequacy is what to look for in designing a sample size. The quantitative and qualitative methods determine sample size singularly and together. For quantitative research, power analysis and rules of thumb can be guides. Generally, observations/variable come into play. For example, the sample size of a paired t-test will be much smaller than a confirmatory factor analysis of a 100-item questionnaire. For qualitative research, the methodology and methods impact sample size (e.g., a homogeneous phenomenological study will be vastly different than if the researcher examined extreme samples using grounded theory, etc.). How both samples relate and produce representativeness together must then be considered.
3. Attrition, maturation, and unforeseen problems. Whatever sample size, add 20-50% to improve the likelihood of meeting minimum specifications. I often see mistakes where 300 is the a priori sample size, and one selects exactly 300. Then 50 people dropped out, and because of statistics with a small effect, the study is now underpowered. Generally, small samples probably need a higher percentage safety net than large samples.
4. Confounders. Exciting research on the quantitative side is examining what degree of a sample could confound the results (e.g., in one study with a sample of 1000, if 70 extremes were different, the main effect disappears, etc.). Few researchers examine confounders in qualitative research, but that could be a new direction.
Where to go? Read about sampling in quant and qual for the methods you pick, then you need to read examples to see what others are doing. There are free sample size generators online, and there are many articles about qualitative sample size. If you have access to a statistican, that would be preferable.
David Coker thank you for your answer.
I would ask you about representative does that mean generalization? though I can not access now to the number of population. It is a project proposal so I may make some changes in future. Can I highlight that as a limitation, I mean I do not know about the number of the population. For the qualitative stage I have chosen 10 teachers though I know I should stop when I reach data saturation. However, the time is limited, could be also be as a limitation?
It is a proposal and I need to talk about the limitation, therefore, How it will be addressed in regard (generalization and sample size).
thank you very much for answering my questions
Representativeness aids in generalizability/transferability. Population is what group you are wanting to research. In your example, if you want to know how math teachers think in your location, all the math teachers would be your population. Does the sample represent the population? There are many demographic variables to consider; there will be nested samples as well. How that location relates to math teachers in general, and if of interest, would aid in generalizability and transferability.
10 might or might not be enough for qualitative research. Propose a range. Saturation depends on what you want to know and how in depth. As a rule of thumb, 10-15 for a thematic analysis of a homogeneous sample will probably be sufficient. How you analyze the survey determines the sample size.
- Badges
- Science topic
- Similar topics
- Psychology