Common Statistical Bloopers, How to Overcome Them?

Statistics is a field of study that cannot be kept away from the research scholars. Some of the research scholars may find themselves less competent than the others when it comes to understanding, and applying statistical tools and techniques in their research. If doing a Ph.D. is your calling then you must be very clear that you cannot stay away from being thorough with statistical skills. However, there are some common problem with statistical communication that are easy to see in research fraternity, with people who are seasoned with the tools and techniques of statistics. It is surprising, but true that often statistics is seen to be used in research in an improper way. Few of the commonly found statistical bloopers are:

1. Sometimes, researchers take the liberty to adjust the scales or neglect the outliers in their data because they want it to be more coherent.
2. Quite often, researchers are confused between correlation and causation, and at times they tend to generalise the results more than the facts.
3. Academicians, Ph.D. scholars, and researchers are always under the pressure to publish more and more. Because of this, when the statistical tests show their results as insignificant, they keep increasing the sample size till the time the results become significant.  It is again a big mistake that is made in statistics. In technical jargon, it is referred to as, “P-hacking”
4. One more big concern in the application of statistics is that the results cannot be reproduced or replicated in an identical study. Statistics show that 97 per cent of the published studies have significant results, but upon the replication of the study only 36 per cent show statistically significant results. This is a big question on the authenticity of the research.

There is alarming concern among the research fraternity to address these issues and more, so that the statistical education among the researchers can be improvised. Fore mostly, journals have to ensure that they offer complete and clear guidelines in context to data analysis and results. They must also encourage the researchers to give the specification of the methods used for data analysis and most importantly, open access of raw data should be facilitated for researchers. In addition to that, statistical education can be improved by perpetually stimulating the research department in universities and organisations to adopt to statistical training and other mechanism to enrich the statistical knowledge. A focussed approach on statistical education is surely a hot need that needs prompt attention and action.