We divide a quiz topic into dimensions. In a rope quiz, for example, we might score aesthetic interest separately from interest in restraint. An answer can add points to one dimension, several dimensions or none. It can also subtract points where the wording points away from a dimension.

On a single-choice question, only your chosen answer counts. On a multiple-choice question, we add the points for every answer you select, up to the stated selection limit.

These points only have meaning within that quiz. Three points in one quiz are not three times as much interest as one point in another quiz. They are part of the answer map we wrote for a particular set of questions.

A raw total is hard to read on its own, so we place it between a lower and an upper endpoint. For each answered question that can affect the dimension, we set these endpoints from its answer weights and a zero baseline. We add the endpoints across your answered questions.

We then calculate (your raw total minus the lower endpoint) divided by (the upper endpoint minus the lower endpoint). This uses the usual min-max transformation, but our endpoints come from answer weights and the zero baseline on your path, not from other visitors' scores.

We keep zero as a possible endpoint. This matters when every answer adds points or every answer subtracts points. On a multiple-choice question, the endpoint also respects the maximum number of answers you may select.

If an answered dimension has no distance between its endpoints, we do not report a comparable ratio for it. Otherwise, the result runs from 0 to 1 and is shown as a percentage. A score of 0 is the lower calculated endpoint for your answered path, and 100 per cent is the upper one.

Imagine that a dimension appears in two single-choice questions. The first question has answer weights of 0, 1 and 3, and you choose 3. The second has weights of -1, 0 and 2, and you choose 0. Your raw total is 3, the combined lower endpoint is -1, and the combined upper endpoint is 5.

The calculation is (3 - (-1)) / (5 - (-1)), which is 4 / 6, or about 0.67. We display that as 67 per cent relative strength. It describes where your answers sit within the calculated range for those two questions.

It does not mean that 67 per cent of your identity fits the dimension, that you have a 67 per cent chance of enjoying it, or that you scored above 67 per cent of visitors. We do not use population rankings in this calculation.

When you skip a question, we leave both its points and its possible range out of the calculation. A skip therefore does not lower your raw total, but it does leave the score resting on fewer answers.

We also keep a skip separate from choosing 'unsure', 'none' or the least-interested answer. AAPOR's survey guidance makes the same distinction between no response and a substantive answer. We do not guess what you would have chosen.

If a quiz changes which questions it asks based on earlier answers, only the questions you were shown can affect your result. Questions on a path you never saw contribute nothing. This keeps the numerator and the possible range tied to the same answered path.

Because skipped questions leave the possible range as well as the points, answering a skipped question can move a percentage up or down. Two people can also reach the same percentage from different numbers of answers. We show answered and skipped counts so that difference is visible.

We rank comparable dimension scores from highest to lowest. Equal ratios receive the same rank. If two dimensions tie for first, both are first; their display order does not turn one into the winner.

The strongest dimension is not always the whole result. A quiz can choose a written result from several scores together, so your result may describe a pattern even when one dimension sits at the top.

If more than one written result remains eligible, we use the priority and order set for that quiz. This makes the outcome repeatable without suggesting that one tied score was higher.

Each quiz has a set of results written for its topic. Depending on the quiz, a result may be attached to a score range, a leading dimension or a particular shape across several dimensions.

For a result based on several dimensions, we compare your score pattern with the pattern written for that result. We use the cosine-similarity formula documented by NIST: the dot product of the two score patterns divided by the product of their lengths. This compares their shape. Proportional patterns have the same fit even when one set of scores is higher overall.

A qualifying pattern across several dimensions takes precedence over a result based only on a threshold or leading dimension. We choose the closest qualifying pattern. If two patterns are equally close, we use their priority, then the order set for that quiz.

Some quizzes have a general result for answer patterns that do not match a more specific one. We use it only after checking the specific results. If you answer none of the questions you were shown, we do not assign a result.

Your result page starts with the written result that matched your answers. It then lists every comparable dimension from strongest to weakest. We show the rank, the rounded relative-strength percentage and how many of the relevant questions you answered or skipped.

A separate 'Not scored' list covers dimensions that had no comparable score. This can happen because the relevant question was skipped, was not part of your path or did not produce a usable range.

The percentages compare your answers with the possible range on that quiz path. They are not percentiles, probabilities or a common scale shared by every quiz. We cover broader questions about reliability and validity on our accuracy and limitations page.