First page Back Continue Last
page Summary Graphics
Abstract:The results of all psychological experiments are noise degraded. In psychophysical experiments this noise degradation can be modeled as coming from two types of noise sources: unique noise sources and common noise sources. One source of common noise is caused by the size of the stimulus set. If the stimulus set is too small then the variability in measures of sensitivity may be so great as to render any inference useless. This seminar will look at the results of a computer simulation of common noise sampling variability. The simulation estimates the amount of variability in measures of sensitivity as a function of the number of stimuli. This allows estimation of the number of stimuli for a real stimulus set.
Today I am talking about the effect of common noise sampling variability on psychophysical experiments. I became interested in this problem when I was considering the trade-offs involved in determining the size of my signal sets for my thesis experimentation. I chose to talk about it today because it is a problem we all come up against when doing psychological experimentation. But before I get into sampling variability itself I will put you into context by talking about noise in general and how it is modeled in psychophysics.