I often read research articles that do not appropriate describe the sampling procedure. For example, the authors mention that the sampling was accidental or purposive but they don’t explain why they had only 10 participants or they say that the sampling was theoretical and describe the a priori established socio-demographic characteristics of their subjects.
I think that the sampling procedure is one of the most important elements of a research. I think that the research should be evaluated according to this procedure and I think that many pieces of research should be rejected because of sampling not being appropriate.
Please help me to clarify this and correct me if I am wrong:
The sampling procedure should state the sampling criteria and should justify the number of participants.
There are two kinds of sampling in social research: statistical – probability or non-probability (when the sampling criteria and the number of participants are established before entering the field according to some rigid sampling rules) and theoretical (when the sampling criteria and the number of participants are flexible, decided in the research process according to relevancy and saturation rules).
Dear Florentina Scârneci-Domnișoru ,
I think you are basically correct. There are of course several approaches to random sampling and different sampling conventions in different disciplines. It is therefore important to understand to audience of any article as well as its content.
Agree with you and Peter Samuels,
sampling strategies and conventions depend on what the aim of your research is. If you want to understand (in german: "verstehen") in depth some events, actions, etc, then you will rather use small n and theoretical sampling based on the phenomenon you want to understand. Theoretical saturation is reached if analyzing new relevant cases does not add to your findings any more. This is typically done in grounded theory analysis or hermeneutical analysis.
If you want to make general arguments (i.e. explain something) for a whole population you will have to sample according to the relevant attributes of this population. The sample by and large represents the population, if basic attributes are distributed similarly (like age, sex, income for populations of countries).
There is another discussion on case selection for cases studies, you have not mentioned. For example in political science you will often choose a case for case studies based on theoretical considerations. You might have a look at John Gerrings Chapter on
Chapter Case Study Research: Principles and Practices
for this tradition.
The theoretical implications for sampling typically enter study design-logic during the sampling frame or sampling stage phase of study participant recruitment. IN this case, sampling frames may be purposive (targeted) for a theoretically relevant group, yet there may be accidental or purposive sampling conducted within a target frame (purported generalization popualtion of interset).
Hope this helps,
Matt
Basically in social research, the researcher selects an appropriate sampling method according to whether or not each and every member of the population has equal chance of being drawn into the sample. if every population member is given an equal chance of sample selection, a random sampling method is being used; otherwise, a nonrandom type is employed. Examples of nonrandom sampling methods are; accidental sampling, quota sampling and judgement or purposive sampling. While random sampling includes; simple random sampling, systematic sampling and stratification sampling methods.
Thank you all very much for your answers!
Rolf Frankenberger:
Isn't the sampling theoretical in case studies?
Abubakar Mohammed Idris:
What are the sampling rules in accidental and in purposive sampling? Do you know the subjects' number before entering the field? If you don't then what is the difference between purposive and theoretical sampling?
Thank you again!
Essentially, one samples according to research purpose. If one's research results are intended to be generalised from sample to population then random sampling in which each member of the sample has an equal chance of being chosen, is appropriate; this is 'probability sampling' which seeks to include variations within a population in a sample, therefore enabling it's generalisation to a population, as previously stated.
Alternatively, in more qualitatively oriented research---usually--one uses 'non-probability' or theoretical sampling of identified 'key informants'.
This is probably. the norm in research; so generally Florentina, you are correct.
Note, however, that in-principle, one could use a random sample in qualitative research, just as one could use purposive sampling in quantitative research.
Thank you, Philip Adams!
So you say that the theoretical sampling is a kind of non-random statistical sampling?
In my opinion statistical sampling (be it random or non-random) must specify before entering the field the exact number and some demographics of the subjects.
In that case, the theoretical sampling is not a statistical one.
I have always liked this thoughtful paper by Clyde Mitchell from some time ago......
Article Case and Situation Analysis
As I see it, one can actually call any kind of sample a 'theoretical' one----based on your theory of how you are going to get respondents say.
In qualitative research, one might seek 'key informants'---a 'theoretical' sample; while a random sample is also theoretical---based on the theory of long-run graduation to the mean (typicality).
It is usually advisable that the research sample should be large enough to ensure a reasonable research findings. if an adequate number of subjects is not included in the sample size, the one's confidence in the research findings will be weak. Remember that sample error is the function of the sample size, with the error being largest when the sample is small.
- Badges
- Science topic
- Similar topics
- Social Science