Nifty Tricks

I like nifty tricks that I can use in data exploration and analysis. Here is one that I just came across today: Daniel Lakens used a simulation to illustrate why you should always report Welch's
Philosophy News image
I like nifty tricks that I can use in data exploration and analysis. Here is one that I just came across today:Daniel Lakens used a simulation to illustrate why you should always report Welch's t-test, which does not assume equal variance, and not Student's t-test -- even when using a two-step procedure that involves Levene's test. It's a simple trick, and it actually makes things simpler, and hey, you're doing the statistically right thing!Here is another one I came across a while back:Chris Said argued that bar graphs with error bars, for all their faults, can be useful for eyeballing comparison of means. However, instead of using the standard 95% confidence interval for means, the eyeballing trick involves using 83.4% CI for the error bars. Now, there are more limitations to this trick, but it's still useful.So, those are the two nifty tricks that I learned recently. What are yours?

Continue reading . . .

News source: Experimental Philosophy

blog comments powered by Disqus