An operational definition states exactly how a concept is measured, in enough detail that a stranger following your protocol would produce the same numbers.
- "Plant health" → chlorophyll index from a SPAD-502 meter, mean of three readings on the third true leaf, taken between 09:00 and 10:00.
- "Reaction time" → milliseconds between stimulus onset and key press, median of 40 trials, excluding responses under 100 ms.
- "Water quality" → dissolved oxygen in mg/L by Winkler titration, sampled 10 cm below the surface.
The test: could someone else replicate your measurement from your write-up alone, without asking you a single question? If not, add the missing detail — usually the instrument, the timing, the number of repeats, or the exclusion rule.
Write these down before you collect data. Defining "healthier" after you have seen the results is how well-meaning students accidentally p-hack.