I know this feedback is coming really late and I don’t know whether it will be of any importance this semester, but it will definitely help you for the next semester. The feedback I’m including is general and is meant for everyone. If I manage to write feedback to each and every one of you I will, but till then this feedback should clear misconceptions.
First a general misconception needs to be cleared. Many of you have thought of purposive sampling and convenience sampling to be synonymous which they’re not. They are two different types of sampling methods under the non-probability sampling method as discussed in an earlier post.
It should be noted that convenience sampling is purely on the basis of the convenience of the researcher. Because he/she finds it convenient to access the sample, or any other reason of convenience the sample is selected.
Whereas purposive sampling would be done based on pre-decided purpose of the researcher. E.g. If the researcher decides to carry out a study on work patterns of night shift workers, then s/he will select night shift workers because the study itself demands that the sample be a night shift working population. Hence the name purposive.
Most of you have answered almost all true or false questions correctly except for one:
Cluster sampling guarantees generalizability to the wider population – answer = false
Even though Cluster sampling is a type of probability sampling it does not guarantee generalizability to the wider population because of its nature.
Since only clusters of a certain kind are chosen the results of the study would be genralisable only to individuals having similar characteristics. Also as different clusters are chosen it becomes imperative that the characteristics of all these clusters remain similar. Otherwise it might give rise to sampling error (difference in expected and calculated values of standard deviation). If this sampling error is not decreased then the results yielded may become less and less genralisable to the population
Coming to the last question in the post test, I needed your suggestions on the situation with respect to the criteria of sampling design.
Considering the example in the post test; the suggestions could have been:
- She should first decide what the goal of her study is going to be. Whether it is going to be a study on part time working professionals from the whole of Mumbai or just a selected area like her building. She should try to align her goal and with what she is actually doing. Maybe types of sampling could be suggested here which she could proceed with, which some of you have done.
- She should try to make her sample measurable. Depending on which type she chooses, the researcher must pay attention to achieving representative ness in her sample at every step. This again depends on the type of sampling chosen by the researcher. For this reason she should try and stick to the probability types of sampling as the non-probability types would not give her a representative sample. Simple random sampling would be the best choice to bring in representative ness but choosing this type may interfere with the practicality of the study or the economy of it. Thus she needs to decide the priority of her study considering the limitations she may face.
- She should also think about the practicality issues of this study. Since the sample participants are working as well as studying there might be a problem with accessing them or communicating with them. The researcher should take note of it and as one of you has said negotiate the time that both the researcher as well as the participants could be available for communication. Thus problems like these should be foreseen by the researcher and ways to tackle them should be devised.
- The researcher should also look at the economy aspect of her study. Making telephonic interviews with participants may require the researcher to invest more time on her part. It may also prove to be costly as opposed to printed questionnaires. If the sample that the researcher has chosen is spread over a large area, then she could take help of assistants to collect data for her, but again the need for these assistants should be justified.
The above answer is only my interpretation; I did not expect your answers to be exactly like this but somewhat along these lines. Considering the different criteria of sampling design, weighing one against another and thus making suggestions to the researcher is what I had expected.
The general performance is satisfactory, but I know you girls are capable of doing much better than this. If anyone has any extra information which would contradict something that I have given above you are free to discuss it with me, u know what to do for that. Just post a comment on my doubts section or better still write a post on your own.
Let’s not allow our blogs to stop because of stagnation. I am sure all of you have a lot to offer. Posting on your blogs does not mean it has to be related only to sampling it can be related to anything else but make sure it makes sense ;)
Cya then, I hope I have been able to make the interaction over blogs as helpful as I could (though I expected more of myself). Still I tried my best.
Regards,
Hemangi :-)
