A Cognitive Framework for Lie Detection

Summary

The costs to businesses annually due to undetected employee lies are outstanding. The costs of employee misconduct to the company range from somewhere between $6 billion to $200 billion annually in the United States (Berry & Lilly, 2003; Lipman & McGraw, 1988). Around 1/3 of businesses fail each year due to employee theft and personnel crimes according the United States Chamber of Commerce. Additionally, a study conducted in 2002 by Avert, Inc. found that 24% of the 1.8 million job applications whose credentials were checked made falsifications in regards to their qualifications.

The purpose of this study is to develop a framework to determine if a potential employee is indeed lying during a job interview, thus cutting down on the billions of dollars lost annually by businesses as described above.

For this study, two experiments were conducted. The first experiment was designed to find out if there was a difference in response time between liars and truth tellers and the roles social skills play (if any) in these response times. The second experiment was designed to determine response time within the subjects (vs. the first which was between the subjects). For example, if the subject will take longer answering falsely than when answering comparable questions truthfully. Another purpose of this experiment was to determine if making the actual decision to lie will add more response time to yes/no questions.

In the first study, it was hypothesized that answering deceptively will take longer than truth telling due to the construction component required when lying. It was also hypothesized that the higher the participants’ social skills, the less time it will take to respond to open-ended questions. This experiment used an experimental design as the participants were randomly selected. The participants consisted of 87 undergraduates that were enrolled in psychology courses in a university located in the southern U.S. Of these, 54 were women and 12 were African American. For the first hypothesis, the independent variable in this experiment was answering deceptively or truthfully (depending on the group that the participant was randomly assigned to) and the dependent variable was response time. For the second hypothesis, the independent variable was the level of social skills and the dependent variable was response time.

Foremost, the participants were to take the SSI test-which is a self report measure of basic social skills. The experimenters would use these results later to correlate social skills and lie response times. The participants were then given instructions as to whether they should tell the truth or lie plausibly. The questions were generated by a computer in the manner that only one word of the question appeared at a time and timing started when the last word of the question appeared. The subject then responded aloud via a microphone and the response was recorded. The questions consisted of both yes/no questions and open-ended questions that were broken down into the following categories: general questions, questions regarding test scores, questions designed to recall recent episodic events and questions designed to recall remotely episodic events.

Their results were consistent with the experimenters’ hypothesizes. Lying did, in fact, take longer than truth telling. On average, the researchers found that it took 230ms longer. The researchers also found out that social skills were unrelated to yes/no questions but responses to open ended questions yielded a faster response time with socially skilled liars than less socially skilled liars.

For the second experiment, the experimenters made three hypothesizes. They hypothesized that when answering deceptively, participants will take longer than when answering comparable questions truthfully. They also hypothesized that making the actual decision to lie will add time to response time to yes/no questions. For the third hypothesis, the experimenters hypothesized that when participants must decide to lie and then lie, their within-subject, within-scenario standard deviations will be larger than their within-subject, within-scenario standard deviations when they are truthful. The participants in this study consisted of 58 undergraduate students from the same subject pool of experiment 1. Of these 58 subjects, 47 were European American; 10 were African American and 1 was Chinese American. As in the first experiment, the independent variable in this experiment was answering deceptively or truthfully (depending on the group that the participant was randomly assigned to) and the dependent variable was response time.

The same procedures were used in this experiment as those that were used in Experiment 1 with two exceptions. First, the SSI test