The PACT experiments

The PACT experiments represent a more commonplace – and more disturbing – problem with the treatment of forensic transcription than the crisis call experiment. Again they use audio from a real murder trial. If you have read the case study, you’ll recognise this story. Here we go into a bit more detail on the experimental results than in the case study itself.

The story

The case concerned three generations of a single family: a father (45), a son (21) and a grandfather (74). One night the son visited the grandfather, drank some beer with him, and then shot him. The grandfather died the next day of his wounds.

Some weeks after the murder, police arrested the son, who eventually confessed to the murder, claiming the grandfather had humiliated and abused him all his life, and that night had made comments about him that sent him over the edge.

The son was found guilty and is currently in jail. This trial concerns the father.

The father was arrested shortly after the son. A listening device planted in the family home – after the murder but before the son’s arrest – had captured a conversation between father and son.

Police alleged this conversation referred to a pact made before the murder, in which the son agreed to kill the grandfather at a time the father had a clear alibi, and the father agreed to share the proceeds of the grandfather’s will with him. If arrested, the son was to claim provocation, and serve a short sentence, getting his share of the money when he was released.

Existence of such a pact of course would make the father an accomplice to murder, at least as guilty as the son who pulled the trigger.

However, the father denied the allegation,  claiming the conversation shows him encouraging his son to turn himself in to police, and making a pact to stand by him no matter what.

The listening device recording was crucial evidence in deciding which side was telling the truth.

The audio

You will probably want to hear the audio for yourself before reading about the experiments. This is a 14-second excerpt from a 38-minute recording.

The crucial phrase (2.5 seconds) from within the above conversation is excerpted here:

The experiment

The experiments sought to investigate, not what the defendant actually said, but what participants might make of the recording if they listened to it carefully, using the transcript as an aid – as juries are instructed to do.


Briefly, the results show that no one ever hears anything remotely like the police transcript of this audio unless it is first suggested to them. When the transcript is suggested with no context, it has a significant priming effect, but one that is easily over-ridden by presentation of an alternative transcript.

main expt results However, when the transcript is suggested in the context of a story similar to the actual case in which the recording featured as evidence, it has a large priming effect, substantially less likely to be revised on presentation of a more plausible alternative.

graph showing large difference in perception depending on suggested words As in the crisis call experiment, the perception even of those who do not accept the suggested transcript is influenced by it in a range of subtle ways. Finally the interpretation of the audio, influenced by the transcript, has a significant effect on opinions about guilt.

graph showing large difference in verdict of guilt correlating with perception You can get a brief overview of the experiments, with preliminary results here. Full results and extended discussion appear as Fraser, H and Stevenson, B ‘The power and persistence of contextual priming: More risks in using police transcripts to aid jurors’ perception of poor quality covert recordings‘ International Journal of Evidence and Proof.

Sign up here for occasional news about Forensic Transcription (no spam!)