Why the last question of an interview decides the first impression

Fifty minutes of solid answers, then a stumble on the closing question, and the panel writes up a candidate who lacked confidence. Reverse the order and the same stumble becomes early nerves that were overcome. The material is identical. Its position is not, and position is doing the summarising.

What everyone sees

Panels believe they are averaging performance across the session, which is what the scoring sheet asks for. Retrospective evaluation does not average. It samples a few moments, weights the end heavily, and then constructs a coherent account in which the earlier material supports the ending.

What is actually happening

Miller and Campbell varied the timing of opposing speeches and of the measurement, and found that whether primacy or recency dominated depended on those intervals rather than on argument strength — so the same content wins or loses according to where it sits in the sequence. Fredrickson and Kahneman showed that retrospective evaluations of episodes track peak and end moments while largely ignoring duration, meaning that a long stretch of adequate performance contributes far less to the remembered whole than its length would suggest. In an interview these two mechanisms compound. The closing exchange occupies the end position, the panel writes up minutes later, and the note becomes the record that all subsequent discussion refers to.

Why it stays hidden

The reordering hides because the story feels like recall. Panellists do not experience themselves as reweighting; they experience a clear memory of a candidate who was nervous throughout. The narrative is generated from the ending and then read as evidence, and the earlier evidence is retrieved through it.