On a second viewing of a scene, you already know its broad course. Attention can therefore shift to details previously overshadowed by the action. Returning is especially useful with a question: why does an apology seem unconvincing, or why does one person appear to control the room? The question limits the investigation without deciding the answer in advance.
Recover the initial impression
First watch the chosen scene without interruption. Afterwards, note whether the earlier impression remains. Perhaps it was stronger when you did not know the outcome. That is an observation in itself. A second viewing need not confirm the first response to be useful.
Define the passage through its action rather than a convenient timestamp. With an apology, what happens immediately beforehand and how the other person behaves afterwards may matter. An extract that is too narrow can remove the very context that explains the effect.
Attend to one layer at a time
During one pass, notice positions and glances. Who moves, who remains seated and whose reaction appears on screen? On another pass, concentrate on voices and other sounds. You need not switch the sound off; directing attention separately is often sufficient. Muting can be an interesting experiment, but it changes the material and should be acknowledged as such.
The Hong Kong Education Bureau's scene-analysis material connects observations about camera and editing with their effects. For your notes, a useful distinction follows: describe first, interpret afterwards. The camera remains on the listener is an observation. This makes the listener's doubtful response more important is an interpretation that can be supported.
Test your explanation against alternatives
If you suspect that a long pause creates uncertainty, check whether posture or vocal delivery supports that reading. A pause might instead convey deliberate composure. Look for the detail that challenges your initial explanation. This makes the account more precise without forcing every effect into a single cause.
Finish with a short connection between choice and effect. For example: the reply remains off screen while we see the listener hesitate, directing attention towards their judgement. An individual observation needs no more than this. A small claim that can be checked is more substantial than a comprehensive theory resting only on memory. You can return to the whole scene afterwards to see whether the isolated finding still fits its rhythm and context.