Two shoes, two runs, with one feeling better and showing lower heart rate. Is the winner already clear? You may have observed a real difference in use, but order, weather or that day’s condition may also have entered the comparison. A fair A/B trial is not good because it is complicated. It is good because it states beforehand what to compare and handles circumstances consistently.
A research laboratory and your own shoe trial solve different tasks. One may measure small physiological differences; the other primarily informs fit and use experience. The following sequence makes both logics clear: question, order, repetition, separately recorded outcomes and proportionate conclusions. Our numerical examples are not ready-made performance tests; they illustrate how the method works.
Step one: choose an outcome suited to one question
Identifying where an upper presses requires a different procedure from measuring a percentage difference in economy. The runner evaluates comfort; oxygen or energy cost needs suitable instruments; a time trial is a performance task. If discussing all three during a test, record them separately. You need not infer one from another in its place.
A useful personal question might be: ‘In similar easy running, which shoe’s forefoot feels more comfortable?’ ‘Which is better for everything?’ is less clear. The first connects a specific experience with a use situation. The second would turn several different aims into one ranking. A question formulated beforehand also helps prevent selecting only a favourite result from many notes at the end.
In a laboratory, the same logic becomes a detailed protocol. Speed, sections, measurement window and outcome are specified. Bolliger and colleagues compared different shoe conditions at several speeds on three separate days, for example, treating oxygen cost, perceived effort and comfort separately. Results thus related explicitly to the questions measured.
Step two: order can have its own column too
If the new shoe is always second, shoe and second condition become linked. The runner may be better warmed up or already more tired, and measurement values can change over time. Handling order aims to reduce mistaken attribution of an ongoing process to the shoe. Random or counterbalanced order is therefore a meaningful methodological element.
In an example of our own, the shoes would actually give identical results, but a chosen time effect makes measured cost successively 100, 101, 102 and 103 units. The table assigns the same four values to two different sequences. These figures are entirely invented, and change is deliberately uniform to reveal order’s role.
| Position in time | Measured example value | ABAB order | ABBA order |
|---|---|---|---|
| 1. | 100 | A | A |
| 2. | 101 | B | B |
| 3. | 102 | A | B |
| 4. | 103 | B | A |
Under ABAB, A averages 101 and B 102, making B appear worse. Under ABBA, both average 101.5. In this simple example, the mirrored order balanced the linear time change. A real runner’s time effect need not be linear, and the previous shoe’s effect may carry over, so ABBA is not a trick eliminating every error. Deliberate handling of order is the important lesson.
Enlarge imageStep three: repetition has several levels
Repeating measurements on the same day reveals consistency over a short period. Repeating on another day also introduces variation in daily condition and circumstances. The two forms of repetition show different things. A laboratory may use elements of both to view uncertainty in an individual difference from several sides.
Barrons and colleagues’ methodological paper particularly emphasises repetition when assigning individual-response labels from a single economy measurement. It does not supply the same personal error margin for everyone. It reminds us that interpreting a small shoe difference requires knowledge of the measurement’s variability. A single best number is not necessarily the stable result.
In original research examining four shoe conditions, every condition appeared twice in a mirrored order. This is a concrete example combining order and repetition. Your everyday trial need not imitate an equally extensive laboratory protocol, but its lesson transfers: an experience recurring in a similar situation says more than one isolated occasion.
Step four: do not let the task change unnoticed
Trying one shoe at an easy pace and another faster also changes speed. A level, cool section differs environmentally from an uphill, sunny one. The experience can still be real, while its cause remains unclear. A research comparison therefore fixes the external task, while your log should at least identify differences.
Warm-up and familiarisation are also parts of the task. The first few steps in entirely new shoes may feel different from familiar use. Research can plan an adaptation period, but not every study uses the same procedure. Seek this part in the methods rather than assuming every participant knew the shoes equally well.
Longer running poses a particularly different question. In Madsen and colleagues’ eighty-minute comparison, researchers examined both conditions at the same external speed and conditions matched to the same relative physiological intensity. These comparisons ask different questions. An A/B label alone is therefore insufficient: state precisely what matched and what could change between the two situations.
As a personal check, begin your note with: ‘In this trial, I kept these things similar and could not control these differences.’ This locates the experience rather than weakening it. It prevents later reading laboratory precision into a run mainly intended to explore use.
Step five: distinguish expectations from construction
A shoe’s name, price, story or appearance may influence the experience we expect. This does not make the runner’s report worthless. It means construction and information supplied about it can be separate variables. If comparison specifically seeks the mechanical product effect, expectations deserve consideration too.
In Hébert-Losier and colleagues’ original randomised experiment, twenty-four recreational female runners tried two pairs of the same shoe model with different descriptions; one pair’s appearance was altered. Researchers found no significant differences in physiological or the biomechanical outcomes examined, but the more favourably presented condition was judged more comfortable and enjoyable. This examined description’s relationship with perception, rather than calling every shoe-technology advantage a placebo effect.
Your own trial need not deceive others or alter shoes. A simpler, honest approach is writing expectations beforehand, then separately recording experiences after running. ‘I expected more from the expensive pair’ and ‘the forefoot pressed’ are different information in a log. Giving both room makes the decision easier to understand later.
Confidence in a more striking shoe may also be a genuine experience, but is not direct oxygen measurement. Equally, a negative preconception does not establish poor mechanical function. Research is strong when it measures the appropriate outcome; a personal use choice is clear when you know which part is feeling, data and inference.
Step six: do not rename heart rate as an economy measurement
At a given pace, heart rate is useful personal data for many runners. Alone, however, it is not equivalent to respiratory-gas analysis for measuring small metabolic differences between shoes. Daily condition, temperature, fluid status, duration and sensor operation may all matter. A precise-looking graph does not automatically isolate a shoe effect.
Our own example: observed heart rate is 145 in shoe A and 142 in B, but the second occasion occurred on a cooler, better-rested day. Record the difference without confidently assigning it to the shoe. Even on the same day, warm-up or time effects may occur. A heart-rate trial is therefore at most personal observation alongside circumstances, rather than verification of a small physiological percentage.
Comfort comparisons remain useful. Where does the shoe hold the heel, how much forefoot room remains, what is the upper like and for which task is it pleasant? Runners answer these specific questions first-hand. They need not be elevated through uncertain metabolic inference. Usability is an important endpoint in itself.
Enlarge imageStep seven: retain the whole result
At the end of A/B comparison, keep more than the single best row. If two of three trials gave similar experiences and one the opposite, the whole pattern matters. Mark differing circumstances, but do not retrospectively fit explanations only to the desired result. Selection is clearer with rules chosen beforehand.
- Question: exactly which use or measurement difference were you seeking?
- Order: which condition occurred when?
- Situation: pace, surface, duration and important circumstances.
- Outcome: comfort, effort and instrument data on separate lines.
- Repetition: how consistent is the difference?
The individual stiffness-response study involving 96 runners also reminds us that differing individual results can underlie a group mean. Interpreting a personal response nevertheless requires measurement uncertainty and repetition too. A sentence tied to the observed situation is therefore better in a log than assigning a lasting runner type from one occasion.
A fair conclusion can be simple yet specific: ‘On two similar easy runs, A’s forefoot felt more comfortable; I did not decide economy from heart-rate data.’ This is a useful result. A fair test need not produce a spectacular winner; it clarifies what you learned, under which circumstances and which question remains open.
When closing the trial, retain original notes rather than only the attractive run selected later. If one shoe required stopping twice for laces, describe this as a separate usability result. It is not fair to combine that interrupted section with a continuous run’s instrument average, but it gives real information about handling the shoe. Decide beforehand which question will use it.
Sources
- Bolliger and colleagues: repeated shoe comparisons over several days.
- Perry and colleagues: a four-condition shoe experiment with mirrored order.
- Hébert-Losier and colleagues: a randomised experiment on expectations and shoe perception.
- Barrons and colleagues: a methodological paper on repeated measurements.
- Madsen and colleagues: longer-running shoe comparisons at external and relative intensities.
- Chollet and colleagues: individual stiffness responses.
How Futószív articles are prepared: our editorial guidelines.


