The quantitative descriptive study is supposed to cover 60-100 target participants however due to the pandemic the research locale had been provincewide instead of regionwide. It only yielded 20 respondents. Would you recommend that the study be continued as statistical treatment had been used as planned?
You might be able to use a t-test with only 20 respondents - please refer to a previous discussion on ResearchGate in this regard:
https://www.researchgate.net/post/Minimum-sample-size-for-t-test
Thank you Rodney Duffet for your response. It is well appreciated.
Dear Researcher, Number of respondents should be 5 times higher than the number of variables to maintain the reliability of the study. Otherwise, your R square will not be accepted.
It's difficult to give feedback without more details such as the population you're targeting and the method you used to reach the respondents. At first glance I would suggest trying to reach your respondents by phone or via the internet, but I realize I'm making a number of assumptions about their access to electricity or internet devices that may not be appropriate. BL
You can do a power analysis using the g*Power program to determine your likelihood for finding an effect with such a small sample.
Simple. Whatever the situation, if you are going to research with a limited number of responses, you may not be able to recover significant relationships using any hypothesis testing, unless there is a large deviation between testing and sample values. For example, if you are testing whether the mean of some variable is 57 or not with a sample mean of 52 with 20 observations, you may not be able to reject your null hypothesis. However, if you have 100 observations with you and the sample mean is 52, you will have enough evidence to reject the null hypothesis.
Sample size plays a big role in hypothesis testing. For large samples, even smaller differences are significant. In contrast, for smaller samples, even relatively large differences are not significant.
Sorry but it is difficult to answer your question without more information for example, the type of data you are collecting, the size of the target population. Also you have indicated the study is descriptive, which leads me to believe you are not testing a hypothesis so it would be helpful if you indicated how you initially intended to analyse your data. Thanks.
20 is a very small number for statistical analysis and most standard tests are not accurate with such a small number as one or two outliers will skew the results. You also have to use your common sense - how much trust can you really place into the results obtained by only 20 people? If these 20 are the best of experts on the topic, there may be some validity here but if that is not the case your results are not meaningful.
You might be able to change your research design to widen your number of respondents, or to change to a qualitative approach where 20 respondents may be more acceptable.
Due to pandemic situations unable to reach to respondents,at least we can use technology as tools for gets informations and used in research...
Unless this is for a class assignment, where your prof may introduce some specific rules, why would a pandemic (or anything) causing you to have a lot fewer people affect how you would address scientific questions with data. I understand how things can affect aspects of the sampling, but I am not sure what you asking. Is this for an assignment?
If you have only 20 observations with you, you can proceed with Non-parametric tests.
Small Ns are just as serious a problem for Non-parametric tests as for interval level data.
It should not be limited number of respondents for qualitative research
En una investigación cuantitativa, considerar mayor cantidad en la muestra puede ser toda la provincia, tratar de ver contactos y los resultados serían buenos para generalizar.
- Badges
- Science topic