Evidence-Based Practice for Nurses

levels are usually either .05 or .01. In the figure, the placement of the alpha level

shows how type I and type II errors can be adjusted. In this example of a one-

tailed test, the area under the curve to the left of the alpha level represents the

amount of type II error. The area under the curve to the right of the line is the

amount of type I error. Notice that when the alpha level is larger, there is more

space to the right of the line than when the alpha level is smaller.

What are the implications of these alpha levels? When the alpha level is set at

.05, it is likely that 5 times out of 100, the researcher would make a type I error

by wrongly rejecting the null hypothesis. When .01 is used for the alpha level, a

researcher would make a type I error only 1 time out of 100. Thus, alpha levels

of .05 increase type I errors while reducing type II errors. In general, although

alpha levels of .01 reduce type I errors, the likelihood of making a type II error

increases. In nursing, .05 is used more commonly than .01 is.

Although the mathematical difference between these alphas is miniscule,

the implications for decision making are great. Figure 13-13 provides an

illustration of how the acceptance or rejection of the null hypothesis is af-

fected by the alpha level selected. When conducting statistical tests to test

hypotheses, suppose there is a chance for three possible critical values. Note

in Figure  13-13 that critical value “A” falls to the left of both alpha levels.

Regardless of the alpha level chosen, the results will be statistically nonsig-

nificant, and the null hypothesis will be accepted. Critical value “B” falls

to the right of both alpha levels. The results will be statistically significant

regardless of the alpha level selected. Now consider the position of critical

FIGURE 13-13

Relationship of Alpha Levels

and Critical Values

Alpha = 0.05 Alpha = 0.01

A

C

B

360 CHAPTER 13 What Do the Quantitative Data Mean?