A worksheet contains measurements before and after an activity. It is tempting to compare the two columns immediately. First ask a quieter question: does each row represent the same subject at both times?
Make the pairs explicit
Consider an illustrative table with three labelled samples. Sample A changes from 4 to 7, B from 9 to 8, and C from 6 to 6. The individual changes are +3, −1 and 0. Those differences have meaning because the label ties each later measurement to the correct earlier one.
Watch what happens when a value is missing
Suppose B has no later measurement. You can describe the observed changes for A and C, but you cannot insert zero for B’s missing result. The group with complete pairs is now different from the original group. Report how many pairs remain and what you know about the missing observation.
Do not shuffle one column into numerical order while leaving the identifiers behind. A row is carrying a relationship, not merely providing a convenient place for two numbers.
Keep change separate from explanation
Even correctly paired measurements do not by themselves establish why a change occurred. Other influences, measurement conditions and the study design still matter. The worksheet exercise establishes who changed by how much within the available records; it does not establish an intervention’s effect.
Use the distinction between observations and subjects to describe the dataset accurately. A helpful methods note names the pairing identifier, the two measurement occasions and the treatment of incomplete pairs. That is more informative than presenting two neat columns with their relationship left implicit.