Michael's Happenings

Notes on "Computer-Mediated Communication and Deception" (Professor Joey George)

Joey George is a professor at Florida State University, and is here in NZ to spend time at the University of Canterbury. He is giving a talk on deception in computer-mediated communication. Joey has been studying this area for over a decade.

Deception is present in 20-33% of everyday communication interactions. Most deception is trivial (“How are you?”, “I’m well thanks” … when you really are unwell). Researchers in the communication discipline have studied deception for decades, but mostly in two-person face-to-face situations. Yet as computer-mediated communication has become more pervasive, the research has shifted focus to how things happen in this area. Eg, phishing, spear phishing (highly-targeted, socially engineered email message).

Two main research questions:
1. How can we detect deception in CMC?
2. How can we design software to overcome this problem?

With respect to question #1, the objectives were to (a) create an integrated deception detection model, and (b) to conduct experiments to confirm the model. There were four US universities involved. The research program ran for 5 years. At Florida State, over 2000 people were involved in the research. Today Joey will be talking about the resume study and the diary study.

Deception is … “messages and information knowingly transmitted to create a false impression or conclusion.” There has to be an intention to deceive. There are four types of deception:
– outright fabrication
– concealment
– equivocation (evasion, deflections, ambiguity)
– exaggerations

Our ability to detect deception … on average, we can detect deceptions 40-60% of the time. People are better at figuring out what’s true, and less better at figuring out when something is false. Why is this? In general, people have a tendency to believe what people tell us.

The theoretical basis for the basis:
– leakage theory (the deceiver has to control so many aspects of the communication when trying to deceive — facial expressions, body language, the words — but they can’t generally control everything … and thus cues to deception “leak” out).
– interpersonal deception theory (adds behavioural adaptation after the reception of a message)
– an integrated model ( … slide went by too fast to write down all of the boxes and arrows)

Key indicators of deception:
– liars’ stories are less compelling — they make less sense; they are told in a less engaging way; the stories are told in a less immediate way; and liars sound more uncertain.
– liars provide fewer details that truth tellers
– liars make more negative statements and complaints
– liars are more tense that truthtellers — more nervous, more vocally tense, speak in a higher pitch
– liars still too closely to the details, to the key elements of their story

An experiment … a Resume study, with student subjects.
– wanted to do a media comparison, between email, chat, chat with audio, and audio only
– students were asked to “enhance” their resume in order to get a scholarship.
– an other independent variable … warnings about lying or not.
– overall result: deception detection rate 8% of the time. If they were warned, they were much better at identifying the lies — 4x more likely to identify.
– … the media made no difference.

A second study … a Diary study.
– give your subjects a way to record their behaviour over a period of time. People were asked to record their communication and deception over a week.
– students were asked to fill out a questionnaire on a PDA over the course of a week.

Categories: Michael's Happenings