Your starting point
Prerequisites: Use the ideas and vocabulary from lesson 26; revisit its worked examples if needed.
Definition: A group comparison examines a summary across categories. An absolute difference subtracts summaries; a relative difference scales that change by a specified baseline. A fair interpretation requires comparable definitions and attention to how records entered each group.
Compare like with like
A team handling simple requests cannot be judged directly against a team assigned complex cases using mean duration alone. Inspect task mix, time period, exposure and measurement rules. A difference in outcomes may reflect these factors rather than skill or effort. Start descriptively and identify what additional evidence would support a stronger interpretation.
Show counts and spread
A mean from three observations and a mean from three hundred do not carry the same amount of information. Report valid counts, missing values and a spread measure. A group with a higher mean can still overlap substantially with another group. Do not infer that every member of the first group has a higher value than every member of the second.
Check aggregation choices
Total output reflects both group size and per-unit performance. Rates require appropriate denominators. Combined summaries can change when group proportions change, even if within-group performance stays constant. Compare relevant subgroups when justified, but avoid searching through many arbitrary splits until a preferred story appears. Explain the reason for each grouping and retain uncertainty about unmeasured differences.
Two worked examples
Totals versus per-person output
Team A completes 100 tasks with ten people; Team B completes 80 with four people. Totals favor A, while tasks per person are 10 and 20. Neither measure alone accounts for task complexity, hours worked or quality. State the question before calling one team more productive.
Small and large groups
Channel A has durations 4,5,6; Channel B has many records centered near 6. A’s mean is 5 from three observations. The one-minute difference may matter operationally, but the small count and possible case-mix differences limit a broad claim. Report the evidence instead of translating the difference into certainty.
Where this becomes useful
A school compares completion across morning and evening cohorts. Work schedules, prior experience and attendance opportunities may differ. The comparison can reveal a support need without proving that the time slot caused the difference.
Make an attempt
Use Dataset Comparison Studio with Course engagement. Compare cohort summaries and their valid counts. Then inspect week as a possible grouping factor and explain what information is lost when all weeks are pooled.
Misconception check
Claim: A fair formula makes any group comparison fair.
Inspect the correction
Comparable measurement and group composition matter as much as arithmetic. A formula cannot remove unmeasured differences.
Independent application
Compare two synthetic groups using counts, means, medians and ranges. Write an absolute difference and a carefully scoped interpretation. Identify a potential confounder and explain why the evidence does not justify ranking every individual by group membership.
Completion records practice and self-review. It is not certification of statistical competence.
