Hi. - If you are using convenience sampling, that is considered a nonprobability (nonrandom) sample. The best way to make sense of that would be if you had regressor (auxiliary) data on the entire population, which could be related to the sample on an individual basis for the sample members, and used to estimate for the cases not in the sample. This is called 'prediction,' though the word sounds misleading. (It is not forecasting.) This is also referred to as model-based rather than design-based sampling. - Jim
'Convenience' and 'randomness' do not go together. First one is from non-probability, and the later goes with 'probability' sampling.
Not clear, why convenience sample is required/ needed!!. If say, one has fairly a large (original sample for) randomized controlled trials (like, what is studied in 'intervention-effect' research), and then there is some compulsion to selectively select (by convenience) a sub-sample (from the above sample), then it is a different case. And it is possible.
The sampling approach you are suggesting to use is a non-probabalistic sampling strategic and the research design seems to be probabilistic (randomized control). These are incompatible.
If you have regressor data for the population, then model-based estimation, prediction, can be used, basically regardless of collection method. However, one must collect from any separate strata required. For a related discussion, see the link below.
Article Efficacy of Quasi-Cutoff Sampling and Model-Based Estimation...
We run into this problem in ed. research a great deal. Sampling has to do with your ability to generalize, while random assignment to treatment has to do with your ability to draw causal conclusions. They are not the same thing, and you can in fact have one without the other. If you have random assignment to treatment, you can draw conclusions about cause but you cannot generalize the relationship past your sample construction. If you have random selection but not random assignment, you can generalize but you cannot draw conclusions about cause. Clearly you are in the strongest camp if you can do both.
Hi every one,
I know this topic was discussed while back and I was wondering if someone can help me figure out if the following research could be classified as an RCT study. I will include some de-identified information from this research:
" ....The subjects were assigned to two groups -experimental and control- using simple randomization. All participants provided consent. A convenience sampling was used for selecting participants from the patients who were referred to a health centre. The inclusion and exclusion criteria were explained to the referring health practitioners. The experimental group received X therapy and the control group received Y therapy by a registered and trained psychologist"
I would really appreciate your responses/feedback.
Sirous
Dear Sirous Momenzadeh Sir,
If you read again Julia B. Smith ma'am's answer, you will easily understand.
Yes, it can be considered an RCT if you randomly allocate the participants to test and control group, however you may not be able to generalise your causal findings/inference to the population.
random allocation and random sampling are two different things.
But there are different schools of thought and some may not agree because in an ideal scenario you should have used some probability sampling technique.
please read section 1 of this article: Article Understanding and misunderstanding randomized controlled trials
So you can write it as RCT but all may not agree.
Hope am able to clarify.
Thank you.
Thanks Sujaya Gupta for taking the time and responding to my query.
As you mentioned, the key is to differentiate between random allocation/selection and random sampling. Thanks again. Sirous
Convenience sampling may produce a sample that is not representative of the target population. It is likely to be difficult to define exactly who has and has not been included. Therefore, convenience sampling is often wildly biased, which can dramatically affect the results' reliability.
No, convenience and randomization can't go together. You may call it non-randomized controlled trial
- Badges
- Science topic
- Similar topics
- Biological Science
- Bioecology
- Systems Ecology
- Landscape Ecology
- Connectivity