Critical thinking test scored as five sub-scales, not one number

Because a single score assumes the five move together, and the best-replicated result in this literature is that they do not travel across content, let alone across each other. The same logical rule that almost nobody handles in the abstract is handled by most people the moment it is dressed as a social situation — so a number claiming to capture your reasoning in general is claiming something the evidence does not support. Five sub-scales are a weaker claim and a truer one: they say what you did on four items of each kind, and the shape is a place to look rather than a trait.

No — that is the test working, and most runs come back that way. Four items give a proportion with a very wide interval on it, and two intervals that overlap are two scores you cannot tell apart at 95%. The page could hide that and print a five-spoke chart with a confident shape, which is what almost every free version does; it prints the bands instead. If the shape matters to you, the fix is repetition rather than arithmetic: three runs on different orders will show whether one skill really does sit last.

Because the skills being measured are ones a clock actively damages. Four of the twenty items turn on stopping to work out a proportion — 90% of a 22% response rate, 78% of 62%, twelve defects in three hundred against five in a hundred — and a person who would have done that arithmetic under no pressure and skipped it under a clock has just been scored on their pressure tolerance. Timed reasoning is a real and separately useful measurement, and the version of these five under an exam clock is on the appraisal-practice page instead.

Two of the twenty could be argued and the rest could not, and the standard applied to all of them is stated in the explanation under each. The rule used throughout is that the right option is the one supported by the material given, not the one that is most likely true in the world — so an option can be correct about reality and wrong here. Where two options are both defensible, the item was rewritten rather than kept, because an item with two answers measures whether you share the author's taste. If you find one that still has two, the explanation will tell you which reading was used.

The parts this page measures respond to being named, which is why every item is explained afterwards whichever way you answered it. Reasoning that fails in the abstract routinely succeeds once the same structure appears in a familiar setting, and that gap is the strongest available evidence that what is missing is usually a way of seeing the problem rather than a capacity. The practical version of that: after a run, the useful move is not to retake it but to catch yourself doing the same thing on a real decision — accepting a rate without asking what the denominator was, or treating a nearby true fact as a reason.

Same five constructs, opposite jobs. The appraisal page reproduces an exam: its published section order, its five different response scales, a twelve-minute clock, and a report built around the five-point inference scale a candidate has to calibrate. This page throws all of that away and asks the same five questions as skills — four-option items, no clock, interleaved, and a profile with intervals. Take the exam version if you have an assessment day in the diary; take this one if you want to know which of the five is your weak one.

No, and the difference is easiest to see in what each one survives. Abstract reasoning scores are famously stable within a person and famously unstable across generations; performance on these five collapses and recovers depending on how the same problem is dressed, which is not how a general capacity behaves. Nothing on this page converts to an IQ point, a percentile or a band, and no browser test does — the ones claiming otherwise are pricing a number they cannot produce.