bensoltoff
commented
7 years ago Very nice research question. I think the graphs you chose directly applied to the question at hand. In a lab it would be much easier to control environment and measure time elapsed more precisely, but for the assignment I think you did a good job controlling for this. With a larger sample size, I think the t-tests would have produced more significant results. An n of 20 is pretty small, even for an experiment. Could also split this up into a series of experiments so each participant doesn't have to evaluate every single channel. Might be more efficient that way.
Reflection: Getting familiar with ggplot2, it is much easier to draw grapics, and it is an interesting try to plot descriptive statistics. The difficult part for me is the research design. It is not easy to think of a principle and design an experiment that appropriately tests it. Moreover, it is important that the research design matches the pros and cons of the instrument in order to collect high quality data. MTurk and Qualtrics do not allow the experimenter to control the experimental environment. From my own data, it might not be a wise choice to include multiple choice questions and collect time information.