Article Figures & Data
Figures
- Figure 1.
Efficacy of two different drugs in increasing the time that mice can remain on a rotating rod. For each drug, the mean effect and the associated SEs are shown. The dashed line represents the null value of no difference. Example taken from Gelman and Stern (2006).
- Figure 2.
Same data as for Figure 1, except that SE bars have been replaced with 95% confidence intervals. The counternull is the value of the effect size that is equidistant from the mean (vs the null), but on the opposite side. See Rosenthal and Rubin (1994).
- Figure 3.
Each table shows the probabilities with which a random variable, X, can take on values of 1, 2, or 3 under two different hypotheses. A, The red box shows the Frequentist perspective in which only the H0 probabilities are considered and the most powerful test is to reject H0 for X = 2 or 3. B, A more Bayesian perspective is shown. Example is from Johnson (2013a), who apparently borrowed it from Berger and Wolpert (1988).
- Figure 4.
The importance of the prior probability of H0 when evaluating p values. Source: xkcd (https://xkcd.com/1132/).
Tables
BMI Category <18.5 Underweight 18.5–24.9 Normal or healthy weight 25.0–29.9 Overweight >30 Obese Source: Centers for Disease Control and Prevention.
Figure 1-1
PowerPoint slides used to teach these examples (and a few others). Download Figure 1-1, PPTX file.
In this issue
