Types of Error

Math 140 „ There are two possible types of error in

significance testing:
Introductory Statistics Null hypothesis is actually

True False

Don’t Reject H0 Correct Type II error

Professor B. Abrego Your
decision
Reject H0 Type I error Correct
Lecture 16
Sections 8.2

1 2

Type I error Type II error

(When Test Statistic is Large) (When Test Statistic is Small)
„ If the test statistic is large in absolute value (like Miguel and „ If the test statistic is small in absolute value (like Jenny and
Kevin’s example), then the possible explanations for this are: Maya’s sample), then the possible explanations for this are:

„ 1. The null hypothesis is true and a rare event occurred. That

is, it was just bad luck that resulted in being so far from p0. „ 1. The null hypothesis is true, and you got just about what
you would expect in the sample.
„ 2. The null hypothesis isn’t true, and that’s why the sample „ 2. The null hypothesis isn’t true, and it was just by chance
proportion was so far from p0. that turned out to be close to p0.
„ 3. The sampling process was biased in some way, and so the „ 3. The sampling process was biased in some way, and so the
sample value is itself suspicious. sample value is itself suspicious.

„ If the last explanation is ruled out, then the usual decision is to

reject the null hypothesis H0. However, you may be making a „ If the last explanation is ruled out, then the usual decision is to
Type I error—rejecting H0 even though H0 is actually true. not reject the null hypothesis H0. However, you may be making a
Type II error—not rejecting H0 even though H0 is actually false.

3 4
Minimizing the Error P-Values
„ Type I Error. Null hypothesis is true but you reject it. „ Instead of just reporting that you either have or have
If the null hypothesis is true, then the probability of making not rejected the null hypothesis, it has become
a Type I error is equal to the significance level of the test.
To decrease the probability of a Type I error, decrease the common practice also to report a P-value.
significance level. Changing the sample size has no effect „ The P-value for a test is the probability of seeing a
on the probability of a Type I error.
result from a random sample that is as extreme as or
„ Type II Error. Null hypothesis is false and you fail to reject more extreme than the one you got from your random
it. sample if the null hypothesis is true.
To decrease the probability of making a Type II error, you (The P-value for a test is a conditional probability)
can take a larger sample n or you can increase the
significance level α. (But if you do the last option you will
increase the probability of a Type I error)

5 6

Example One-Tailed Tests of Significance

„ Suppose that 22 students out of a random sample of „ When testing the effectiveness of a new drug, the investigator
must establish that the new drug has a better cure rate than the
40 students carry a backpack to school. Follow steps older treatment (or that there are fewer side effects). He or she
a–d to test the claim that exactly half of the students isn’t interested in simply rejecting the null hypothesis that the
in the school carry backpacks to class. new drug has the same cure rate as the older treatment. He or
she needs to know if it is better. In such situations, the alternate
„ a. Name the test and check the conditions needed for hypothesis should state that the new drug cures a larger
it. proportion of people than does the older treatment.
„ b. State the hypotheses in words and symbols.
„ This is called a one-tailed test of significance. Tests of
„ c. Calculate the value of the test statistic. Calculate the
significance can be one-tailed if the investigator has an
P-value for the test. Use this P-value in a sentence that indication of which way any change from the standard should
explains what it represents. go. This must be decided before looking at the data.
„ d. What is your conclusion? Explain in the context of
this problem.

7 8
 Example: One-Sided Test
Alternate Hypothesis
of Significance
„ When testing a proportion, the
alternate hypothesis can take „ The editors of a magazine have noticed that
one of three forms. people seem to believe that a successful life
„ Ha : The percentage of depends on having good friends. They would
successes p in the
population from which the
Ha : p ≠ p0 like to have a story about this and use a
sample came is not equal to headline such as “Most adults believe friends
p0.
are important for success.” So they
„ Ha : The percentage of commissioned a survey to ask a random
successes p in the Ha : p > p0
population from which the sample of adults whether a successful life
sample came is greater than
p0. depends on having good friends. In a random
sample of 1027 adults, 53% said yes. Should
Ha : The percentage of
„
successes p in the Ha : p < p0 the editors go ahead and use their headline?
population from which the
sample came is less than
p0.

9 10

Example: One-Sided Test Example: One-Sided Test

of Significance of Significance
2. Give the name of the test and 2. State the hypotheses, defining 3. Compute the test statistic z and find the P-value
check the conditions for its use. any symbols.
Name: Significance test for a The test statistic is
proportion. „ H0 : The proportion of
people p who believe that a 0.53 − 0.5
n = 1027, p0 = 0.5 (the standard, if successful life depends on z= = 1.92
having a succesful life is not good friends is equal to p0 = 0.5(1 − 0.5)
affected by having good friends or 0.5. (p = p0 )
not) 1027
„ Ha : The proportion of
people p who believe that a
„ The sample is a simple random successful life depends on
sample (it says in the problem), good friends is greater than And then the P-value is the probability that we get a sample
from a binomial population (a p0 = 0.5 (p > p0 ) with proportion greater than this test statistic (1.92)
person either agrees or not that
a successful life depends on (Because we have a one-sided significance test).
good friends) .
„ np0 = (1027)(0.5) = 513.5 > 10 Thus P-value = normalcdf(1.92, 999999)
n(1 – p0) = (1027)(1– 0.5) = = 0.0274288
513.5 > 10
„ Total number of adults
> 10 (1027) = 10270

11 12
Example: One-Sided Test
Example: Halloween Treats
of Significance
4. Write a conclusion. „ Researchers at the Yale Center for Eating and Weight Disorders
wanted to see if children out trick-or-treating would be satisfied
„ Since the P-value equals 0.0274 and this is less with small toys instead of candy. In households in Connecticut
neighborhoods, children were offered two bowls: one contained
than α = 0.05 = 5%. Then we should reject the null candy and the other small, inexpensive toys like plastic bugs
hypothesis. that glow in the dark.
„ Of the 283 children, 148, or about 52.3%, chose the candy. The
„ If the percentage of all adults who believe a researchers report that the difference is not statistically
successful life depends on having good friends is significant. Is that correct? You may assume that the children
50% or less, then the probability of getting a sample are a random sample from all children trick-or-treating in
proportion of 53% or more is only .0274. Since this Connecticut neighborhoods that year.
value is too small this gives strong evidence that the [Source: “Trends: Halloween, for Skinnier Skeletons,” New York Times, October
21, 2003, report of a study published in The Journal of Nutrition Education and
true percentage is greater than 50%. The editors Behavior, (July–August 2003).]
should feel free to run the headline.

13 14