The pooled sample proportion is the combined proportion from two independent samples, used in Honors Statistics when you test whether two population proportions are different. It gives the shared estimate needed for a two-proportion z test under the null hypothesis.
In Honors Statistics, the pooled sample proportion is the single proportion you get when you combine the successes from two independent samples and divide by the total sample size. You usually write it as p-hat pooled, and it is used when the null hypothesis says the two population proportions are equal.
The formula is simple: p-hat pooled = (x1 + x2) / (n1 + n2), where x1 and x2 are the numbers of successes in each sample and n1 and n2 are the sample sizes. This is not just an average of the two sample proportions. It is a weighted estimate, so the larger sample has more influence on the result.
Why pool at all? Because under the null hypothesis, you are acting as if both groups come from the same population proportion. Instead of keeping the two sample proportions separate, you combine them to get one best estimate of that shared proportion. That pooled value is then used to calculate the standard error for the two-proportion z test.
A quick example makes the logic clearer. Suppose sample 1 has 18 successes out of 60, and sample 2 has 24 successes out of 80. The pooled sample proportion is (18 + 24) / (60 + 80) = 42 / 140 = 0.30. That 0.30 is the common proportion you plug into the test statistic when you are checking whether the groups are really different.
A common mistake is to use the pooled proportion for confidence intervals. In Honors Statistics, you usually do not pool for a confidence interval because you are estimating the difference between two proportions, not testing the null hypothesis that they are equal. Pooling belongs to the hypothesis test setup, not every comparison of proportions.
The pooled sample proportion shows up right at the point where Honors Statistics shifts from describing data to making an inference about two groups. If you are comparing the proportion of students who passed two different classes, the defect rate from two factories, or the success rate of two treatments, the pooled proportion gives you the shared baseline needed for the test.
It also connects the algebra to the logic of hypothesis testing. Under the null hypothesis, you assume there is no real difference in the population proportions, so the samples are treated as if they came from one common proportion. Pooling is the mathematical way to express that assumption before you compute a z score and p-value.
This term matters because it separates the two-proportion z test from other statistics you might use in class. If you mix up pooling with averaging or with confidence intervals, your work can look almost right but lead to the wrong standard error and the wrong conclusion. Once you know when the pooled estimate belongs, the rest of the test is much easier to follow.
It also sharpens your interpretation. A small p-value does not come from the pooled proportion by itself. It comes from comparing the observed sample difference to what you would expect if one common proportion really were true. That is the real job of pooling in this part of the course.
Keep studying Honors Statistics Unit 10
Visual cheatsheet
view galleryPopulation Proportion
The pooled sample proportion is built from sample data, but it is used to estimate a population proportion only under the null hypothesis. When you compare two groups, you are really asking whether their population proportions are the same or different, so this concept sits right on top of population proportion ideas.
Hypothesis Testing
Pooling happens inside a hypothesis test, not as a standalone calculation. You first set up null and alternative hypotheses, then use the pooled proportion to find the standard error and test statistic for the comparison.
Confidence Interval
Confidence intervals for the difference in two proportions usually use separate sample proportions, not a pooled one. That difference trips people up, because pooling is for testing a claim of equal proportions, while intervals estimate a plausible range for the difference.
Independence Assumption
Pooling only makes sense when the two samples are independent. If the data are paired or matched, the combined proportion does not reflect the right model, and you need a different method.
A quiz question or free-response problem will usually give you counts of successes and sample sizes from two groups, then ask you to test whether the proportions differ. Your first move is to check that the data fit a two-sample proportion setting, then calculate the pooled sample proportion from the total successes divided by the total sample size. After that, you use it in the standard error for the two-proportion z test.
You may also need to explain why pooling is appropriate. A strong answer says the null hypothesis assumes equal population proportions, so the samples are combined to estimate that shared proportion. If the problem asks for a confidence interval instead of a test, that is your clue not to pool.
These get mixed up because both use two proportions, but they do different jobs. The pooled sample proportion is used for a hypothesis test when the null says the populations are equal. A confidence interval usually uses the separate sample proportions instead, because you are estimating the size of the difference, not assuming it is zero.
The pooled sample proportion combines the successes from two independent samples into one proportion.
You use it in a two-proportion z test when the null hypothesis says the population proportions are equal.
It is calculated as total successes divided by total observations across both samples.
Pooling is for hypothesis tests, not the usual confidence interval for a difference in proportions.
If the samples are not independent, the pooled proportion is not the right tool.
It is the combined proportion of successes from two independent samples. In a two-proportion z test, you use it to estimate one shared proportion under the null hypothesis that the population proportions are equal.
Add the successes from both groups, then divide by the total number of observations from both groups. In symbols, p-hat pooled = (x1 + x2) / (n1 + n2). This gives the weighted combined proportion, not a simple average of the two sample proportions.
You pool because the null hypothesis says both groups come from the same population proportion. The pooled value gives one best estimate of that shared proportion, which is then used to calculate the standard error and test statistic.
Usually no. Confidence intervals for the difference in two proportions typically use the separate sample proportions, because you are estimating a range for the difference rather than testing equality under the null.