Twenty mixed items that end in a link, not a score

Because settling into one is exactly what a real battery denies you, and because a set grouped by format would measure something different. Given five numerical items in a row, most people are quicker on the fifth than on the first — they have the arithmetic loaded and the format in their hands — so a grouped set flatters whichever format happened to come last. Interleaving costs everybody the same switching penalty on every item, which is what makes the four medians comparable with each other. It is also closer to what a graduate battery feels like, where sections arrive back to back with no warning about what is in the next one.

Nothing about how good you are, and something quite useful about where your time goes. Accuracy over five items is extremely unstable — one item either way is a 20-point swing, and a different set of five would routinely produce a different order. Median seconds per item is far steadier, because it is a property of how you work rather than of which five questions turned up, and it is the quantity that decides whether you reach the end of a real paper. That is why the routing weighs both and why the panel tells you to read the table across rather than down.

So that the strand measures the physics rather than the picture, and so that it works for anybody reading with a screen reader. A drawn pulley system asks two things at once: can you read the diagram, and do you know that the load is shared between the supporting rope sections. Separating them is genuinely informative — plenty of people who find the drawn version hard have no trouble at all once the arrangement is described in a sentence. It is a real departure from what an employer will send you, and the pictorial version lives on the mechanical page next door for exactly that reason.

That your pace is under two items a minute on this mix, which is a fact about pace and not about ability. Most published aptitude batteries are built so that finishing is difficult, because a test everybody completes cannot separate the people taking it — so running out of time is the designed outcome rather than the failure mode. What is worth looking at is where the minutes went: if one format's median is double another's, that difference is where the whole set was decided, and it is a much more actionable thing to know than a total.

On this drill, yes, and on most employer batteries, yes — but check, because the exception is real. Nothing here subtracts for a wrong answer, so a guess among four options is worth a quarter of a mark and a blank is worth nothing. A small number of assessments do apply a correction for guessing, and they say so in the instructions; where they do, a guess among four is worth nothing on average and the calculation changes. The instruction screen is the only place that answers this, and it is worth the thirty seconds of reading it takes.

It can, and the commonest way is a format the clock never reached, which is why those are excluded from the decision entirely rather than scored as zero. The other way is a run where all four came out level and the tie-break picked on a median separated by two seconds — a difference that means nothing. The table under the recommendation exists so you can overrule it: all four formats are linked, with their own numbers beside them, and if you already know which one you dread then that is better evidence than twenty items.

Yes, and the mechanical strand is the reason it works better here than a pictorial set would. Every option is a button, the items are text so nothing has to be pinched to read, and the answers you give by tapping are counted separately and reported. What a phone does cost you is a little time per item, because aiming a thumb is slower than moving a finger already resting on a number key — enough to matter if you compare a phone run against a keyboard one, and not enough to change which of the four formats is your slowest.