Model
Digital Document
Publisher
Florida Atlantic University
Description
Twenty-four naive human Ss were run in an auditory signal
detection task examining two hypotheses. 1) Naive Ss can
produce data consistent with the theoretical assumptions of
normality and equal variance of the underlying distributions. ROC curves on double-probability paper were found to be
generally supportive of these assumptions as indicated by their
linearity and slope. 2) The "trial-by-trial" method more
evenly distributes the effect of nonsensory variables, providing
less variable estimates of d' and S than does the "block"
method. No differences were found in mean d' values although
the "trial-by-trial" method produced less variable estimates.
The ''block" method produced somewhat better orderings of
individual, but not mean, beta values across probability levels. It is concluded that further use of naive Ss in signal detection paradigms permissible and that the "trial-by-trial"
method is preferable for obtaining stable estimates of d',
but not for obtaining appropriate orderings of beta.
detection task examining two hypotheses. 1) Naive Ss can
produce data consistent with the theoretical assumptions of
normality and equal variance of the underlying distributions. ROC curves on double-probability paper were found to be
generally supportive of these assumptions as indicated by their
linearity and slope. 2) The "trial-by-trial" method more
evenly distributes the effect of nonsensory variables, providing
less variable estimates of d' and S than does the "block"
method. No differences were found in mean d' values although
the "trial-by-trial" method produced less variable estimates.
The ''block" method produced somewhat better orderings of
individual, but not mean, beta values across probability levels. It is concluded that further use of naive Ss in signal detection paradigms permissible and that the "trial-by-trial"
method is preferable for obtaining stable estimates of d',
but not for obtaining appropriate orderings of beta.
Member of