If a researcher has a hypothesis in his study, i.e., "There is no significant difference in perception of Ph.D., Masters, & bachelors students of higher education towards internet usage", and the researcher developed a Likert scale and collected data by using Google form and sending it to various WhatsApp groups through friend and colleagues circle and got one hundred fifty responses. Can these procedures, which the researcher has used in his study, meet the assumption of ANOVA to use for testing the above null hypothesis?
Another thing is, nowadays, online data collection has been very popular due to the advancement of technology, and the Google form is very general and useful also. So what kind of sampling procedure will it follow when we share it person to person and collect data? Is it scientific to use inferential statistics in the analysis of online collected data, as it may violate the assumption of random sampling?
I request the experts and experienced people and also the scholars to share their experiences related to this query.
Thank you
The use of ANOVA in analyzing data depends on whether the assumptions of ANOVA have been met. The assumption of random sampling is one of the critical assumptions in the use of ANOVA. The researcher in this case used a Likert scale, sent the form to various WhatsApp groups through friends and colleagues, and collected one hundred fifty responses. The sampling method used in this online data collection procedure is a convenience sampling method because participants were selected based on their accessibility and willingness to participate in the study. Convenience sampling is not a probability sampling method, and it violates the assumption of random sampling, which is a requirement for using inferential statistics.
Thus, it may not be scientific to use inferential statistics, like ANOVA, for analyzing this online collected data due to the violation of the assumption of random sampling. However, other types of descriptive statistical analyses, such as mean, median, mode, frequency, and percentage, can be used to analyze the data. It is essential to understand the limitations of online data collection and the methods used to collect the data to determine the type of statistical analysis that suits the research purpose.
I think that what matters is whether this is a contribution to knowledge in your field. If so, you could report the results of the ANOVA along with a warning about the non-random sample, and also mention this as a limitation in your Discussion section.
However, if I were a reviewer of this design, I would think that it would be relatively easy to obtain a better sample, so I would be doubtful about this data.
What is often done with convenience samples is that the sample are randomly allocated into conditions, so that E(X1) = ... E(Xk) for the k groups. While all sampling procedures in in the social sciences have issues, using a snowball procedure from friends and colleagues and their contacts has more issues than some others. Ultimately it depends on the design and research purpose.
Daniel Wright's comment about random allocation reminded me of this slide from one of my lectures. HTH.
https://www.wolterskluwer.com/en/solutions/ovid/pdq-epidemiology-8144
Thank you all for sharing valuable knowledge and resources for understanding this concept. Respected experts @Ajit Singh, David L Morgan,Daniel Wright & Bruce Weaver.
- Badges
- Science topic
- Similar topics
- Psychology
- Mental Processes
- Learning