Skip to main content
Cognitive Research: Principles and Implications

Table 2 SSM Effect Calculated Four Different Ways from the Same Data Set

From: From “satisfaction of search” to “subsequent search misses”: a review of multiple-target search errors across radiology and cognitive science

References

Different SSM error calculations on low-salience targets in Adamo et al. (2019) data

Results of t-test

Trials that are filtered

Fleck et al. (2010)

Single-low hit rate compared to low hit rate on dual-target trials where a high was detected first or second

t = 5.10

p < .001

Cohen’s d = 0.93

Single: None

Dual: Remove trials where high-salience target was not found

Cain et al. (2013)

Single-low hit rate compared to low hit rate on dual-target trials where a high was detected first

t = 8.95

p < .001

Cohen’s d = 1.63

Single: None

Dual: Remove trials where high-salience target was not found and trials where low-salience target was found first

Adamo et al. (2019)

Matched single-low compared to low hit rate on dual-target trials where a high was detected first

t = 8.22

p < .001

Cohen’s d = 1.50

Single: Remove trials that do not match the dual-target trials included in the analysis

Dual: Remove trials where high-salience target was not found and trials where low-salience target was found first

Becker et al. (2020)

Expected hit rate calculated from dual-target trials compared to low hit rate on dual-target trials where a high was detected first

t = 5.21

p < .001

Cohen’s d = 0.95

Single: Remove all single-target trials

Dual: Keep all dual-target trials to calculate expected hit rate. Remove trials where high-salience target was not found and trials where low-salience target was found first to compare to expected hit rate

  1. The data analyzed were used in Adamo et al. (2019) in a multiple-target search with high- and low-salience targets (see Fig. 1). A two-tailed, paired-samples t-test was used to analyze the data. Fleck et al. (2010) originally calculated SSM errors by comparing the hit rate for single-target trials of one target type (e.g., low-salience ‘T’) to the hit rate of the same type of target when a different type of target was found on dual-target trials (e.g., low-salience ‘T’ when a high-salience ‘T’ was detected; See Fig. 1 graph). Cain and Mitroff (2013) later restricted to instances in which the target of interest was not found first on dual-target trials (e.g., hit rate for a low-salience ‘T’ when a high-salience ‘T’ was detected first). Adamo et al. (2019) then suggested changing the methods to be more in-line with radiological methods and use matching displays—displays where single- and dual-target trials were identical in target and distractor identity and location with the exception that one of the two targets were removed in the single-target trials (see Fig. 1 search displays). They suggested restricting the SSM calculation to only include matched single- and dual-target trials (e.g., filter the dual-target trials where the high-salience ‘T’ was found first and compare the hit rate for the “second” low-salience 'T' to the hit rate on its matched single, low-salience display). Becker et al. (2020) recommended using only dual-target trials to calculate the expected hit rate for a specific target (e.g., low-salience ‘T’; See Eq. 1). This value is compared to the same type of target when a different type of target was found first on dual-target trials (e.g., hit rate for low-salience targets on dual-target trials when a high-salience target was detected first)
Back to article page