Training Professionals to Detect Deception

1
Training Professionals to Detect Deception

• Joey F. George, David P. Biros, Judee Burgoon
• J.F. Nunamaker, Jr.
• First NSF/NIJ Symposium on Intelligence and
  Security Informatics, June 2003

2
Overview

• Research Question Motivation
• Study Logistics
• Research Design
• Experimental Procedures
• Development of Materials
• Results
• Implications Conclusions

3
Research Question Motivation

• People are not very good at detecting deception
  in 50 to 60 range
• RQ Whether training people to identify cues to
  deception detection helps improve their accuracy
  in detecting deception
• Past training effort results have been mixed
• What role might computerized training system play?

4
Training Study Logistics

• Duration
• 4 sessions 10 Sept, 17 Sept, 1 Oct, 15 Oct
• 1 session for administering questionnaires, 3 for
  instruction
• Personnel
• FSU Joey George Kent Marett
• AFIT Monti Knode, Monica Dziubinski, Mark
  Lankowski
• UofA Judee Burgoon, Jay Nunamaker, Janna Crews,
  Jinwei Cao, Ming Lin, Mark Adkins
• Subjects 168 officers in basic communications
  training program

5
On Location
Study at USAF facility
Our Team
Study pilot at AFIT
6
Research Design
7
Experimental Procedures

• Subjects reported for training
• Began with battery of instruments including
  knowledge judgment pre-tests
• Training
• Powerpoint component
• Video, audio only, or text examples
• End with additional instruments including
  knowledge judgment post-tests

8
Development of Materials

• Curriculum materials
• Pre-tests post-tests
• Knowledge
• Judgment (deception detection)
• Other measurement instruments
• Agent99
• Pilot at AZ
• Usability
• Recording of cues lecture by same people giving
  it at USAF facility
• Content exactly the same across delivery modes

9
Screen Shot - View Examples
10
Results Knowledge Tests
Table 1 Means and standard deviations (in
parentheses) for knowledge pre-tests and
post-tests
11
Results Judgment Tests
Table 2 Means and standard deviations (in
parentheses) for judgment pre-tests and
post-tests
12
Results

• Trained groups beat untrained groups in knowledge
  tests
• No differences between trained and untrained
  groups on judgment tests
• No differences among treatment groups on
  knowledge tests
• No differences among treatment groups on judgment
  tests
• Overall trend of improvement on judgment test
  scores across all groups over duration of
  experiment

13
(No Transcript)
14
Implications Conclusions

• Training worked for immediate increase in
  knowledge about deception detection
• Overall training process associated with improved
  detection accuracy for all subjects
• Computerized training no worse than traditional
  lecture only training
• Additional research called for