FIGURE 1 TYPE I AND
TYPE II ERRORS Proper Course of Action Take No Action Take Action Take No I II Action Correct Type I Decision Error Decision Take III IV Action Type II Correct Error Decision FIGURE 2 THE BASIC DECISION MATRIX OF POLICYMAKERS Choice 1 Choice 2 Incremental Dramatic Action Action--Mild No Mitigation Action Proper Course of Action I II Correct Decision Type I Error III IV
Type II Error Correct Decision TABLE 1 FOUNDATIONS OF CONSTRUCTIVIST VS.
Cohen (1988) suggested that the
Type II error risk should be 4 times as great as the Type I error risk to ensure adequate analyses without having to use unrealistically high sample sizes for social science research.
Although previous statistical methods give data on quantity of type I error to researchers, the occurrence of
type II error mostly remains unknown (Stefano, 2001).
Courts, agencies, and academics have reacted to antitrust's unusual vulnerability to error by adopting a bias in favor of false negatives (
Type II errors).
For most groups,
Type II error is large relative to Type I error.
Empirical studies (e.g., Ferron, Foster-Johnson, & Kromrey, 2003; Ferron & Ware, 1995) concur with the latter statement showing that
Type II errors ate the main problem, while Type I errors ate usually controlled.
A
Type II error in this experiment occurs when R = 1 and P < 1.
When FDA commits a Type I error, "the status quo will be disrupted and people will die much sooner than they otherwise would." When the agency commits a
Type II error, people will be denied a treatment for their illness, but "the status quo will remain" (p.
[TrT.sub.A] =
Type II error Correct decision
Type II error [TrT.sub.B] 3.
The second class of errors, termed
Type II errors or false negatives, occur when the system fails to detect or identify that which the system has been designed to look for.
Given the possibility of a
type II error and the absence of any downside to involvement with Alcoholics Anonymous, an argument can be made for continuing to refer patients to this group.