Slide 1: Question: What is a type I error and type II error?

Slide 2: Type I error: Falsely concluding there is a difference between groups, when in reality there is no difference. Type II error: Falsely concluding there is no difference between groups, when in reality there is a difference.

Graphic displaying types of error. Correct conclusion. Study conclusion shows no difference and reality shows no difference. Type II error. Beta. Study conclusion shows no difference and reality shows a difference. Type I error. Alpha. Study conclusion shows difference and reality shows no difference. Correct conclusion. Study conclusion shows difference and reality shows difference.

Slide 3: Example. A study is performed to assess whether a new drug significantly lowers blood pressure. The green boxes represent patients whose blood pressure lowered over the course of the study. Treatment group. n equals 6. Green boxes. 4 out of 6 sampled equals 67 percent. Control group. n equals 6. Green boxes. 4 out of 6 samples equals 67 percent. Does this actually mean no difference between groups?

The investigators conclude that there is no difference in mean blood pressure measurements between the treatment group and the control group. What type of error is present?

Slide 4: In reality, the mean blood pressure in the treatment group is significantly lower than the mean blood pressure in the control group. A type II error is present.

The probability of making a type II error (beta) is closely related to the concept of power. Power equals one minus beta. Power is the probability of detecting a true difference between group. Typically, before a study is conducted, a power analysis performed to determine adequate sample size. A larger sample size equals more statistical power.

Slide 5: In our example, increasing the sample size (n equals 6 versus n equals 18) would increase the power of the study and detect the true difference between groups. Treatment group. n equals 18. Green boxes. 11 out of 18 equals 61 percent. Control group. n equals 18. Green boxes. 6 out of 18 equals 33 percent. Sampling more of the population allowed us to see the true difference!

Slide 6: Type I error: false positive. Green boxes. n equals 6. 4 out of 6 equals 67 percent. Green boxes. n equals 6. 2 out of 6 equals 33 percent. Green boxes. n equals 18. 8 out of 18 equals 31 percent. Green boxes. n equals 18. 8 out of 18 equals 31 percent. Sampling more of the population allowed us to see there is truly no difference!

Slide 7: Type II error: false negative. Green boxes. n equals 6. 4 out of 6 equals 67 percent. Green boxes. n equals 6. 4 out of 6 equals 67 percent. Green boxes. n equals 18. 11 out of 18 equals 61 percent. Green boxes. n equals 18. 6 out of 18 equals 33 percent. Sampling more of the population allowed us to see the true difference!