Suppose a class tries reading strategy A on Monday, strategy B on Tuesday, and then announces that B is better because Tuesday’s scores are higher. There is a hidden rival explanation: Tuesday is also the second attempt. Children may understand the task more clearly, become less nervous, or learn how the marking works. A useful experiment must separate the method being tested from the opportunity to practise.
Counterbalancing means deliberately varying the order in which conditions are experienced. For a simple comparison of A and B, one group tries A then B, while another tries B then A. This does not guarantee a fair result; it helps distribute possible first-try and second-try advantages instead of giving them entirely to one method. The immediate question for a student is: If the order were reversed, would the apparent winner change?
Start with the problem, not the impressive-looking chart
An ordinary A-then-B class trial has two changes at once: the strategy and the order. If the second result rises, which change mattered? Without an order check, an attractive bar chart cannot answer that question. A short diagnostic is to ask learners three things: Which method went first? Did every student receive methods in the same order? What else becomes different on a second attempt?
This matters beyond reading. The same issue appears when comparing two memory games, two ways of taking notes, two keyboard layouts, two learning apps or two brief classroom explanations. Sometimes later scores improve through practice. Sometimes they decline through fatigue. Either way, position in the sequence can influence the result.
The mechanism: practice and fatigue travel with time
A first attempt teaches more than its answer. A child learns where to look on a worksheet, how the timer sounds, how a rubric awards marks and which mistakes to avoid. Even if a teacher gives a different passage next, that familiarity may transfer. If B always goes second, B gets this extra familiarity for free.
Order effects are not always beneficial. Students may be tired by the second task, especially if it is long. A method tested later might therefore appear worse even when it is equally useful. Counterbalancing shares the positional risk across the methods.
A crucial limitation follows: switching order does not remove all differences. If the first technique changes learning in a lasting way, the second technique may inherit that change. This is a carryover effect. Different passages may also differ in difficulty. Design and interpretation still matter.
Worked example: what happened to eight readers?
The figures below are a made-up teaching example, not a report of any real class or a promised effect. Eight students complete two short reading tasks, each marked out of 10. The teacher compares strategy A (summarise the paragraph) with strategy B (annotate key words), using passages intended to be similar in difficulty.
Four students receive A first and B second. Their A scores are 6, 7, 5 and 6: a total of 24, or a mean of 6/10. Their B scores are 8, 8, 7 and 9: a total of 32, or a mean of 8/10. Looking only at this group, B appears better by two marks.
The other four students receive B first and A second. Their B scores are 5, 7, 6 and 6: total 24, mean 6/10. Their A scores are 8, 7, 9 and 8: total 32, mean 8/10. In this group A appears better by two marks. The winning method switched when the order switched.
- A-first/B-second group: mean A = 6/10; mean B = 8/10.
- B-first/A-second group: mean B = 6/10; mean A = 8/10.
- Across all eight students, mean A = (24 + 32) ÷ 8 = 7/10.
- Across all eight students, mean B = (32 + 24) ÷ 8 = 7/10.
- Across all eight first attempts, mean = 48 ÷ 8 = 6/10.
- Across all eight second attempts, mean = 64 ÷ 8 = 8/10.
Notice what the numbers say, and what they do not. In this example, the position of the task matches the difference better than the method label. We can describe a second-attempt advantage in the invented data. We cannot prove that practice alone caused it: passages, fatigue patterns or other unrecorded factors could also matter. We certainly cannot announce that A and B are universally equal on the basis of eight imagined results.
How to hand-check the calculation
Every student supplied one A score and one B score. Therefore each method has eight scores. The combined mean is each method’s sum divided by eight, not by four, and no student should be silently counted twice within one method. A score of 8/10 represents eight marks out of ten, not eight minutes or eight percent. Keep the units and the denominator visible.
Plan a better classroom comparison
Write down the purpose before the activity begins. Are we evaluating speed, accuracy, comprehension or perceived usefulness? Do not combine those outcomes into one number without an explicit rationale. Decide how long each attempt will last, how work will be scored, and how order will be assigned. If conditions permit, assign students to the orders fairly rather than placing the strongest readers in one sequence.
Use comparable but distinct tasks. If students answer the exact same questions twice, recognition of the answers may dominate the result. If tasks differ greatly, passage difficulty becomes another source of variation. A teacher can cross-check difficulty and switch which passages are paired with which methods. For a low-stakes demonstration, explain these limitations openly instead of presenting the activity as a definitive clinical experiment.
Students should record a simple table with student code, first method, first task, first score, second method, second task and second score. The code preserves pairing without publishing names. Include information on breaks and unexpected interruptions. If one student misses the second activity, note the missing outcome rather than filling it in with their first score.
Guided practice: three steps toward an independent answer
Step 1: spot the confound
A teacher gives every learner method X before method Y. Y has the higher average. Ask: “What two things changed?” A correct response names both method and test order; it does not simply repeat that Y had higher marks.
Step 2: reverse the order
Plan a second sequence, Y then X, for another comparable group. The learner must draw both sequences and identify which method appears first in each. They should also suggest one way to keep passage difficulty or time limits reasonably comparable.
Step 3: evaluate an ambiguous result
The second attempt produces higher scores in both sequences. Write a cautious conclusion with the words “observed,” “possible” and “not proven.” A good response might say: “We observed higher scores on the second attempt regardless of the method order; familiarity with the task may have contributed, so this trial does not isolate an advantage of either method.”
A follow-up challenge is to imagine the opposite pattern, where the second attempt always scores lower. What explanation involving tiredness would then deserve consideration? That question tests understanding rather than memorisation of the worked example.
Parent and teacher uses
Parents can apply the idea gently when trying two revision activities at home. Avoid judging the second activity after one evening just because a child has become familiar with the material. A teacher can ask pupils to explain why an investigation needs comparable tasks, clear scoring and transparent sequence assignment. The observable progress measure is not “my child preferred the new app.” It is that the learner can recognise an order effect in a new situation and design a fairer small comparison.
For more formal education research, counterbalancing is only one tool among others. Individual differences, sample size, non-random assignment, carryover and test difficulty remain relevant. A small school exercise is best presented as a lesson in careful comparison, not proof that a learning method raises grades.
Did you know? Equal averages need not mean identical experiences
In the invented class results above, both strategies average 7/10 overall. But that overall equality hides a clear pattern: both score higher in second position. Students who look only at two summary bars miss a central feature of the data. This is why checking how data were collected is as important as calculating them correctly.
Frequently asked questions
Must every student try both methods?
Not always. A different design assigns each student to only one method. That avoids some carryover problems but has other issues, including differences between groups. Choose the design to suit the question and school constraints.
Does ABBA order solve every practice effect?
No. More elaborate sequences can help with some systematic order patterns, but uneven or long-lasting carryover may remain. A simple A–B/B–A arrangement is an accessible introduction, not a guarantee of unbiased causation.
Why not simply retest everyone after a week?
A delay can affect memory and recovery, but it does not by itself balance method order. It may also introduce new experiences. Record the timing and explain what is being compared.
Does this change how pupils should study?
It encourages fair self-testing. Compare methods on different but suitably similar tasks, note which was tried first, and keep the conclusion modest. The point is better judgement, not turning ordinary homework into a laboratory.
Useful next reading
The University of Groningen research-skills chapter on experimental design explains order effects, practice and counterbalancing. Read alongside eduKate’s guide to what an open-book quiz can actually show and the English learning guides. The Education Hub supplies the wider learning context, while the Clementi Mathematics reference illustrates the value of diagnosing a reasoning step before rushing into more practice.
The habit to keep is short: when a method appears to win, ask whether it also happened to go second. A small question about order can prevent a big mistake about learning.