PSYO 2000 Lecture Notes - Lecture 17: Psychophysics, Calcite, Sensory Memory
Document Summary
Sensitivity index or detectability index (from noise or distractors) If no errors (misses or fa) d" will be abnormally high (uninformative and or misleading) B: position of the threshold on the x axis. When the data follows a normal distribution you use a parametric test (anova, t test) when it doesn"t follow a normal distribution, you use non parametric test (t test, kruskal wallis, counts) Non parametric tests have one advantage: they are less sensitive to the number of subjects. The previous slides state that id no errors (misses od. Fa0< d" will be abnormally high (uninformative and/or misleading) . It is difficult to get a d prime or an accurate d prime if you don"t have errors. This is partially true: we can apply a correction to approximate d" in this situations. Since we cant have fa= 0, or misses = 0. Solution 1: we can transform the proportions of 0 to .