Every interview edit is a construction. An hour becomes four minutes, which means ninety three percent of what someone said is gone and somebody chose what stayed.
That is not a problem in itself, it is the job. The question is where shaping ends and misrepresentation begins, and there are specific techniques that sit on the wrong side of it.
The line I use
Would the person, watching this, recognise it as what they said and meant?
Not whether they would like it. They may disagree with the framing, dislike the context, or wish they had said something else. Those are legitimate outcomes of being interviewed.
The test is whether the edit represents their position accurately. If they would say "I never said that" and be right, the edit has failed.
Removing qualifications is the most common failure
People speak with hedges. "In most cases", "generally", "with some exceptions", "based on what I've seen".
These are tedious and they slow a cut down, and removing them is the easiest tightening available.
They are also frequently the substance. "This generally works, though not for everyone" and "this works" are different claims, and the second is one the speaker did not make.
My rule is that a qualification can go if it is verbal padding and must stay if it limits the claim. Distinguishing those requires actually thinking about what the sentence means, which is more work than tightening usually involves.
Joining separated statements
Taking a phrase from minute six and a phrase from minute forty and cutting them together into one apparent sentence.
This is sometimes fine and sometimes not, and the distinction is whether the joined version says something neither original said.
Joining two consistent statements on the same subject to remove a digression is ordinary editing.
Joining an answer about one thing to a question about another produces a statement that was never made. This is the technique behind most genuinely deceptive interview edits and it is invisible to a viewer.
The practical guard is keeping a note of where each piece came from, so that when you look at a finished cut you can see that a sentence is assembled from three places and ask whether it should be.
Cutaways and what they conceal
The reason cutaways exist is to hide edits. That is their function and it is legitimate.
It is also why they deserve attention. A cut visible as a jump tells the viewer something was removed. A cut hidden behind a cutaway does not.
Some documentary practice has moved toward leaving jump cuts visible for exactly this reason — an honest signal that this is an edit rather than a continuous statement.
I do not think concealing edits is wrong. I think knowing that is what you are doing changes how carefully you do it.
Reaction shots that were not reactions
A technique from television that is straightforwardly deceptive and remains common.
The interviewer's reaction shots are frequently recorded after the interview, with the interviewer nodding or looking sceptical at nothing.
Placed against a particular answer, a sceptical expression tells the viewer how to read that answer, and the expression was not a response to it.
The same applies to reaction shots of an audience or of another participant, lifted from elsewhere.
This is the one I would not do. It attributes a response to someone who did not have it, and it directs the viewer's judgement using something manufactured.
Audio processing that changes meaning
Subtler and worth knowing about.
Pauses can be shortened or extended. A shortened pause makes someone sound certain. An extended one makes them sound hesitant or evasive.
This is done routinely for pacing and it changes how a person reads. Extending a pause before an answer is a way of implying discomfort that the speaker did not display.
Music does the same job more visibly. The same answer under warm strings and under a low drone reads completely differently, and neither was in the room.
What I do practically
Keep the full transcript with timecodes, and mark on it what was used and from where.
Watch the cut against the original for anything contentious, to check that the edited version says what the original said.
Never join answers to different questions unless the result is clearly consistent with both.
No manufactured reactions.
And where the material matters to someone's reputation or livelihood, show them the relevant section before publication. This is not universal practice and there are arguments against it. In my experience it has improved accuracy every time and has never cost me a story.
The part that is unavoidable
Selection is still an editorial act and there is no neutral version of it.
Choosing which four minutes of an hour to use is choosing what the person said, and no amount of care makes that choice invisible.
What care does is keep the result honest, which is a lower standard than objective and is the one actually available.