Independent Samples — Definition, Formula & Example
Independent samples are two groups of data collected so that picking the individuals in one group has nothing to do with picking the individuals in the other. No individual in one sample is matched or paired with an individual in the other.
Samples from population 1 and from population 2 are independent if every is statistically independent of every . Under this condition, . That additivity is what justifies the two-sample and procedures.
Key Formula
Where:
- = sample standard deviations of groups 1 and 2
- = sample sizes of groups 1 and 2
- = standard error of the difference in sample means (valid only for independent samples)
Worked Example
Problem: A random sample of 25 students from School A has mean test score 78 with standard deviation 10. A separate random sample of 36 students from School B has mean 72 with standard deviation 12. Find the standard error of the difference in means and the two-sample statistic.
Step 1: Check independence. The two schools were sampled separately and no student is paired with another, so the samples are independent of each other. Each sample is also less than 10% of its school's population, so observations within each sample are approximately independent.
Step 2: Because the samples are independent, the variances add. Compute the standard error.
Step 3: Divide the difference in sample means by the standard error. Using the conservative choice, degrees of freedom are .
Answer: The standard error is points, and the test statistic is with 24 degrees of freedom (conservative).
Why It Matters
On the AP Statistics exam, you must state the independence condition before running any two-sample -test or two-proportion -test. Choosing between a two-sample procedure and a paired procedure is a common free-response question, and the choice depends entirely on whether the samples are independent.
Common Mistakes
Mistake: Treating before-and-after measurements on the same people as independent samples.
Correction: Measurements taken twice on the same subjects, or on matched pairs such as twins, are paired data. Compute the difference for each pair and run a one-sample procedure on those differences.
