Bayesian methods used in COVID-19 research
During the times of uncertainty in statistical methods used in COVID-19 research. Were Bayesian methods employed much in these contexts? Yes they were here and there but probably once again, just like in regular research, the methods were not used very often or were put as secondary analyses, possibly out of fear of the public not understanding the methods. They also did not seem to be used in the comparisons of vaccinated and control groups in clinical trials of the major vaccines that have been in the news and even administered.
The question is then why might Bayesian methods have been useful for these analyses? Perhaps in studies where sample size was an issue and there was not enough power for frequentist methods, Bayesian methods could have been useful. Perhaps also where p-values were an issue, than reporting Bayes factors could have been useful. Finally, in convergence issues of modeling, Bayesian methods using Monte Carlo Markov Chain (MCMC) could have been useful from these contexts. There are so many reasons why Bayesian methods could have been useful in the COVID-19 research. However, it appears from a simple search of “Bayesian” and “COVID-19” that these methods were not highly utilized in major research.
Bayesian analyses certainly did not take center stage during research done during the major part of the COVID-19 pandemic, but certainly these methods could have been used more. As time goes on, Bayesian methods are being used more often due to their flexibility and utility. Perhaps reaching for classical frequentist methods was more convenient.
Keywords: Bayesian, MCMC, Bayes factor, statistical analysis, COVID-19, Usha Govindarajulu
June 23, 2021