AP Statistics inference problems require students to choose the correct test or interval before doing any calculations. This cheat sheet organizes the major decision points in an inference flowchart so students can match a situation to the correct procedure. It is especially useful for free response questions, where naming the procedure, checking conditions, and writing conclusions all matter.
Students need this reference to avoid mixing up proportions, means, paired data, and categorical tests.
The core idea is to identify the parameter, decide whether the task is a confidence interval or significance test, and then choose the correct formula. Most inference procedures use a test statistic of the form . Conditions such as randomness, independence, approximate normality, and expected counts determine whether the method is valid.
Strong conclusions must interpret the result in context using the parameter and the evidence from the -value or confidence interval.
Key Facts
- Use a one-proportion procedure when the parameter is and the data come from one categorical sample with success and failure counts.
- Use a two-proportion procedure when comparing from two independent categorical samples or treatments.
- Use a one-sample procedure when the parameter is and the data come from one quantitative sample with unknown population standard deviation.
- Use a two-sample procedure when comparing from two independent quantitative samples or treatments.
- Use a paired procedure when data are matched pairs or before-and-after measurements, and analyze the differences with .
- A common test statistic structure is .
- A confidence interval has the general form .
- For chi-square procedures, use and check that expected counts are usually at least .
Vocabulary
- Parameter
- A parameter is a numerical value that describes a population, such as , , , or .
- Statistic
- A statistic is a numerical value calculated from sample data, such as , , , or .
- Null Hypothesis
- The null hypothesis states the default claim, usually that there is no difference, no association, or a parameter equals a specific value.
- Alternative Hypothesis
- The alternative hypothesis states the claim being tested, such as , , or .
- P-value
- A -value is the probability, assuming is true, of getting a result as extreme or more extreme than the observed statistic.
- Standard Error
- Standard error estimates the typical sampling variability of a statistic, such as for one proportion.
Common Mistakes to Avoid
- Choosing a two-sample test for paired data is wrong because matched observations are not independent. For paired designs, subtract within each pair and use a one-sample procedure on the differences.
- Using instead of in a one-proportion significance test standard error is wrong because the test assumes the null hypothesis is true. Use for the test statistic.
- Forgetting to check conditions is wrong because inference formulas are only valid when sampling, independence, and distribution conditions are met. Always state conditions in context before calculating.
- Interpreting a confidence interval as a probability about the fixed parameter is wrong because the parameter is not random in frequentist inference. Say that the method captures the true parameter in about the stated percent of repeated samples.
- Writing a conclusion without context is wrong because AP Statistics scoring requires a decision tied to the population and problem setting. Include the parameter, comparison to , and whether there is convincing evidence.
Practice Questions
- 1 A random sample of students finds that support a later school start time. Which inference procedure should be used to estimate the true proportion of all students who support the change, and what is the sample proportion ?
- 2 A company tests whether a training program changes employee productivity by measuring the same employees before and after training. Which inference procedure should be used, and what variable should be analyzed?
- 3 In a two-way table with rows and columns, what are the degrees of freedom for a chi-square test of independence?
- 4 Explain how you would decide between a one-sample test, a two-sample test, and a paired test when reading an AP Statistics inference problem.
Understanding AP Stats Inference Flowchart
A flowchart is useful because the words in a problem can hide the data structure. Start by marking each observational unit. It may be a student, a household, a bottle, or a day of production.
Then identify what was recorded for each unit. A yes or no result is categorical. A measured value such as sleep time, temperature, or exam score is quantitative.
Next, look for connections between observations. If the same person gives a score before and after a treatment, those two values belong together.
Their difference is the real measurement to analyze. Treating them as independent loses useful information and can give the wrong standard error.
The study design controls what a conclusion can claim. A random sample supports a conclusion about the population from which it was sampled. Random assignment to treatments supports a cause and effect conclusion for the subjects in the experiment.
Neither feature means that the result applies to every person everywhere. For example, a school survey selected randomly can estimate opinions at that school, but it cannot prove that an activity caused those opinions.
A randomized experiment on volunteers can show an effect of the treatment among the volunteers, though the volunteer group may not represent all teenagers. AP Statistics questions often test this distinction through small details in the description.
Conditions are not a ritual list. Each one protects a part of the calculation. Randomness helps prevent systematic bias.
Independence keeps one observation from supplying hidden information about another. When sampling without replacement, the sample should be no more than ten percent of the population so the observations are close enough to independent. For means, graphs and descriptions matter.
Strong skewness or unusual outliers can make a small sample unreliable. Larger samples reduce this concern, but they do not repair bias.
For proportions, the expected success and failure counts must be large enough for the sampling distribution to behave predictably. Write the evidence for each condition using numbers or facts from the problem, rather than simply writing that conditions are met.
A significance test begins with a claim about no difference, no effect, or a stated population value. The p-value measures how surprising the sample result would be if that claim were true. A small p-value is evidence against the null claim, not proof that the alternative claim is true.
A large p-value does not show that the null claim is correct. It only shows that the data did not provide strong enough evidence against it. Confidence intervals answer a related but different task.
They give a range of plausible values for a population parameter. If an interval for a difference includes zero, the matching two-sided test would not find significance at the corresponding level. For chi-square work, compare observed counts with counts expected under the model.
Large mismatches increase the statistic. The final conclusion should name the population, the variables or treatments, and the direction of the evidence. It should never claim that a sample statistic itself is the population truth.