OpenEDG PCAD: Statistical Methods: Descriptive and Inferential Statistics

What this domain tests

Descriptive and inferential statistics: summarizing distributions, quantifying uncertainty, and testing claims. The exam checks whether you can choose the right statistic and interpret it correctly.

Core rules

• Center and spread: the mean is sensitive to outliers; the median and IQR are robust. Report standard deviation with its ddof convention; pandas defaults to sample ddof=1, while NumPy defaults to population ddof=0.
• Distribution shape: examine skew and percentiles; a mean far above the median suggests right skew.
• Inference: the central limit theorem makes sample means approximately normal for large n; confidence intervals quantify uncertainty around an estimate.
• Hypothesis tests: state H0 and H1, choose a test such as a t-test for means or chi-square for categorical association, and interpret the p-value as the probability of data at least as extreme under H0, not the probability that H0 is true.
• The significance level alpha controls Type I error; power and sample size control Type II error.

Worked example

Two checkout variants have conversion counts. You run a two-proportion z-test and get p = 0.03 at alpha = 0.05. You reject H0 and say the difference is statistically significant, but you also report the effect size, for example 2.1 versus 1.8 percentage points, and a confidence interval. Reasoning: significance is not importance; a small p-value with a tiny effect may not justify a rollout, and the interval shows the plausible range.

Traps that cost marks

• Saying p = 0.03 means there is a 3% chance H0 is true.
• Confusing statistical significance with practical significance.
• Using a one-sample t-test for paired data instead of a paired test.
• Ignoring multiple comparisons when testing many groups.
• Mixing ddof=0 and ddof=1 and reporting inconsistent standard deviations.