Hello,
I have a paper where I am getting R squares at 23%. The reviewer has asked if this can be considered as sound!
I remember reading that in social sciences, since there are several factors which affect the predictability, low R squares can be considered as sound. Can someone help me find the reference please
Thanks in advance
Anamika
For better or worse, 23% is not uncommon in the social sciences. An option might be to describe "future research" that could address this issue.
One way to increase the R-sq is with better measures, especially if some of your current ones do not have high reliability (i.e., you are using a single item when a multi-item scale might explain more variance).
Another source of low R-sq is omitting important theoretical variables that might have additional explanatory power. so you could hypothesize about what to include in future research.
I've seen a lot pf papers with R2 between 02-0.3! There is certainly the common perception that high R2 equaltes to 'successful' research, and vice versa. This tends to be too dogmatic. At the end, whether a study with low R2 can be considered as valuable research depends in my view on other aspects: Whether your model(s) are theoretically sound, add to fill a research gap by providing answers to important questions etc.
Perhaps you might want to read through the following related discussion, which provides more arguments and references. https://stats.stackexchange.com/questions/114919/low-r2-value-in-social-science-or-education-research
Anamika Sinha I agree with Andreas Kopf regarding misperceptions about the quality of a study based on R-square. What more important is that how do we understand and interpret R-square and other output of a statistical analysis.
Secondly, 23% explanation of the total variance (i.e., R-square or Coefficient of the determination) is not at all small but just below Large (i.e., Large is 0.2592 and above), according to Cohen (1988). Accordingly Small is 0.0196, and Medium is 0.1304. Consult the references given below please. Regards
Cohen, J. (1988). Statistical power analysis for the behavioral sciences. (Second ed.). Hillsdale, NJ: Lawrence Erlbaum Associates.
Cohen, J. (1992). Statistical power analysis. Current Directions in Psychological Science, 1(3), 98-101. doi:10.1111/1467-8721.ep10768783
- Badges
- Science topic
- Similar topics
- Social Science