Trying to Apply Bayesian Approach to my Research
If I understand the Bayesian data analysis method correctly then, it’s an analytical method that is based on probability. This method differs from the traditional analytical methods (e.g. t-test, ANOVA) in that it takes into account assumptions made prior to data collection. That is, the likelihood of a particular hypothesis to be true. In addition, the Bayesian data analysis method can be updated as one collects data. Again, differing from traditional analytical methods.
As I was reading through this article, I was trying to figure out how to apply the Bayesian data analysis method to some of my own research. For instance, to the movie study I had discussed before. As you may recall, this study consisted of using a computer interface to a movie database to determine “Opening Gross Movie Earnings” of four pre-selected movies. Data consisted of Twitter tweets and predictive models. This data was be displayed in various formats such as graphs, charts, and text. Participants used these data to make their predictions. I hypothesized that teamwork would result in more accurate predictions than group work and individual work. My assumptions that teams would outperform the other two groups was based on prior work and a literature review.
Then, running the movie study using the Bayesian approach I would have not needed to compute a power analysis nor would I had to worry about setting the p-value to .05. Additionally, I would have not needed to stress about recruiting an equal amount of participants for each experimental condition (Individual/Group/Team). However, I would have tried to determine the probability of teams outperforming individuals and groups. Lastly, as I was collecting data I could have “continuously” analyzed the data. To me, this sounds pretty awesome. But, what happens if my initial hypothesis was wrong? In the case of the movie study, my initial assumptions were not correct because teams did not outperform individuals and groups. In fact, for some movies teams performed worse than individuals and groups. I am assuming that the Bayesian analysis approach can correct itself after acquiring some data, but how many data points does it take to do so? In particular, what happens if the sample size is small as in the movie study? Anyone know the answer?
