\ud83d\udea8<\/p>\n
4,000 interviews across the Niger Delta.
260 of them were conducted with nobody.<\/p>\n
A research team was running a corruption-reporting experiment across 106 communities in Nigeria.
The data was flowing in.
The numbers looked fine.<\/p>\n
But something nagged at them.
Some surveyors were finishing interviews far faster than anything seen in training.
Too fast to be real conversations.<\/p>\n
So the team added something quiet to the survey app.
Audio recordings of the surveyors reading the questions aloud.<\/p>\n
Then they listened.<\/p>\n
13.83% of the surveys failed the check.
Some recordings were completely silent.
Others captured surveyors chatting with family and friends \u2014 while entering fake responses into the tablet.<\/p>\n
Interviews with nobody.
Answers from no one.<\/p>\n
260 of 1,880 surveys were flagged and thrown out.
And here’s the part worth remembering: once surveyors knew the audio was reviewed nightly, fabrication dropped sharply.<\/p>\n
The lesson isn’t just about catching fraud.
It’s that a cheap audio snippet is one of the strongest deterrents against wholesale fabrication ever tested.
People behave differently when the tape is rolling.<\/p>\n
If your field data has never been listened to \u2014 literally listened to \u2014 how sure are you about what’s in it?<\/p>\n
Based on: Gomila, Littman, Blair & Paluck, “The Audio Check” (Social Psychological and Personality Science, 2017)
#FieldResearch #DataQuality #SurveyFraud #MEL #AudioAudit<\/p>\n
#FieldResearch #DataQuality #SurveyFraud #MEL #AudioAudit<\/p>\n<\/div>\n
submitted by \/u\/Significant-West-492 <\/a>
[link]<\/a><\/span> [comments]<\/a><\/span><\/p>
