Why the same profile can evolve without contradicting itself

That is the simplest case: new information, a personal experience, a change of professional or family situation have shifted a position. There is nothing more to explain, and nothing awkward about it: a conviction that moves under no circumstance resembles a belonging more than a judgement.

Work on the evolution of opinion distinguishes three sources of change it is useful not to conflate: the effect of age, the effect of the period lived through by everyone at the same time, and the effect proper to a generation. The same individual shift may belong to all three.

An item worded the same way may no longer denote the same situation a few years later, because the context it refers to has changed. Answering differently is then the mark of a constant position applied to a new case, and not of a reversal.

That is a reason to be wary of comparisons over time on unchanged wordings. The stability of the text does not guarantee the stability of what it measures, and it is a known problem of all long survey series.

Every measurement involves a variation from one sitting to another, linked to mood, attention and the context of reading. A gap of a few points on a dimension between two sittings therefore means nothing, and an honest model should indicate the order of magnitude below which a difference is not interpretable.

This variation is not a defect to be corrected but a property of the instrument. Acknowledging it avoids two opposite errors: concluding to a change that did not take place, or refusing to see a real change because it is modest.

A large shift, on several dimensions at once, and persistent across several sittings, designates a change of position. An isolated shift, on a single dimension, with no other modification, is more probably imprecision.

The change of meaning of the question is the hardest to spot from the inside, and the only reliable indication is rereading the items: if a statement seems to you to bear on something other than at your first sitting, that is probably the case.

Good practice consists in comparing the broad orientations and the gaps between dimensions, rather than the exact values. What matters is knowing whether the ranking of your priorities has changed, not whether a score went from 61 to 58.

A useful marker: note, at the time of the test, the question that made you hesitate most. At the next sitting, check whether it is still the same. The shift in your hesitations is a better indicator of evolution than the shift in your scores.