Skip to content
AP Statistics · Inference for Categorical Data: Proportions

Null hypothesis

Null hypothesis appears in 6 reviewed AP Statistics questions and sits in Inference for Categorical Data: Proportions. Below are worked examples showing exactly how the exam tests it, with the answer marked and every distractor explained.

How Null hypothesis is tested

Example 1Easier
In a study examining whether a new fertilizer increases crop yield, the null hypothesis is H0: the new fertilizer has no effect on yield. A Type I error in this context would be:
  1. AUsing too small a sample size to detect a real effect.
  2. BApplying the fertilizer to the wrong type of crop.
  3. CConcluding the fertilizer has no effect when it actually does.
  4. DConcluding the fertilizer increases yield when it actually has no effect.Correct

Why D is right

A Type I error (false positive) is rejecting the null hypothesis when it is actually true. Here, the null hypothesis is that the fertilizer has no effect. A Type I error means concluding that the fertilizer works (rejecting H0) when in reality it has no effect on yield.
Why the other 3 choices miss
A
Using too small a sample size to detect a real effect relates to low statistical power, which increases the probability of a Type II error, not a Type I error.
B
Applying the fertilizer to the wrong crop is an experimental design error, not a Type I error in hypothesis testing.
C
Concluding the fertilizer has no effect when it actually does is a Type II error (false negative) — failing to reject H0 when H0 is false.
Example 2Easier
A quality control engineer wants to test whether the defect rate of a manufacturing process has decreased from the historical rate of 8%. She inspects 250 items and finds 12 defective. Which are the appropriate null and alternative hypotheses?
  1. AH0: p = 0.08, Ha: p ≠ 0.08
  2. BH0: p < 0.08, Ha: p = 0.08
  3. CH0: p = 0.08, Ha: p < 0.08Correct
  4. DH0: p = 0.048, Ha: p < 0.048

Why C is right

The null hypothesis should state the parameter equals the historical value: H0: p = 0.08. Because the engineer suspects the rate has decreased, the alternative hypothesis is one-sided: Ha: p < 0.08. The sample proportion of 12/250 = 0.048 is the observed statistic, not a hypothesis value.
Why the other 3 choices miss
A
Incorrect because the engineer specifically wants to test whether the rate has decreased, making this a one-sided test (Ha: p < 0.08), not a two-sided test (Ha: p ≠ 0.08).
B
Incorrect because the null hypothesis must contain the equality (H0: p = 0.08), and the alternative contains the inequality. The roles are reversed here.
D
Incorrect because the hypotheses should reference the historical rate of 0.08, not the sample proportion of 0.048. Hypotheses are statements about the population parameter, not the sample statistic.
Example 3Harder
A study examines whether a new tutoring program improves student test scores. Students are randomly assigned to tutoring or control. After the program, the tutoring group shows a 5-point improvement compared to the control group (p = 0.002). Which statement BEST interprets the p-value?
  1. AThe probability that tutoring improves test scores is 99.8%, given the observed 5-point difference between groups.
  2. BThere is a 0.2% probability that the null hypothesis of no tutoring effect is true, given the observed data.
  3. CThe tutoring program improves scores by 5 points in 99.8% of cases.
  4. DIf tutoring had no effect, there would be a 0.2% chance of seeing a difference of 5 points or more.Correct

Why D is right

The p-value is the probability of observing results as extreme as (or more extreme than) what was observed, ASSUMING the null hypothesis is true. P = 0.002 means that if tutoring truly had no effect, there would be only a 0.2% chance of seeing a 5-point or larger difference between groups by chance alone.
Why the other 3 choices miss
A
The p-value does not give the probability that the treatment works. It gives the probability of observed data under H0. These are fundamentally different.
B
P-values are not probabilities that H0 is true. In frequentist statistics, H0 is either true or not; the p-value describes data probability under H0, not the probability of H0 itself.
C
The p-value says nothing about how consistently the treatment produces the same improvement across individuals. It describes the probability of the observed aggregate outcome under the null.

Studied alongside

Other concepts from Inference for Categorical Data: Proportions.