This is a question that is asked hundreds of times per day, and most polygraph examiners simply quote the “party line” of “90 to 95 percent.” But this isn’t really correct because the answer is much more complex than that.
THERE IS NO SUCH THING AS A LIE DETECTOR
First, there is no such thing as a “lie reaction” that is identical for everyone. The physical indicators of deception very greatly from person to person, so there is no single indicator that tells us when a person is lying. However, people who lie will usually demonstrate at least some of the known indicators of deception, and it is these indicators that a polygraph examiner looks for. Simply, the more indicators that show up in the test the more likely that the person is lying.
KEEP THE QUESTIONS FOCUSED
The number of relevant questions used in a polygraph affects the accuracy of the test. Since polygraph accuracy is not a fixed number, there will always be a margin of error, and this margin is compounded as more questions are added to a test. Imagine rolling a pair of dice, and you are trying NOT to roll a “3.” The odds of rolling a “3” are about 5% (1 in 20), similar to the error rate of many polygraph exams. On your first roll, the odds of rolling a “3” are about 5%, but as you continue to roll the dice over and over, the chances that you will roll a “3” begin to increase. Adding more questions to a polygraph is like rolling the dice over and over, and the odds of making a mistake go up. The more questions asked, the more chances a decision error will be made.
DIFFERENT TYPES OF POLYGRAPH EXAMS
There are three different types of polygraph exams, each with it’s own considerations.
1. Specific issue polygraph – This type of exam focuses on one primary question (ie. Did you steal that missing watch?”) and produces the most accurate results possible.
2. Multiple issue or multi-facet polygraph – This type of exam includes up to 4 different relevant questions about different aspects of the same issue (ie. Did you steal that missing watch? Do you know who stole that watch? Do you know where that missing watch is right now?”)
3. Screening polygraph – This type of exam includes several different relevant questions about different topics, and is often used in pre-employment testing. This is the least accurate type of exam available and generally produces results in the 75 to 85% accuracy range.
Let’s focus on the Specific Issue polygraph exam, since the other types of exams are statistically less accurate and should not be used to resolve serious issues anyway. Next, we need to make sure that a validated polygraph technique is used. There are dozens of polygraph techniques that are taught, but only a few have been scientifically validated for accuracy. Two of the validated techniques, for example, are the Utah ZCT and the Air Force MGQT.
SCORING OF THE CHARTS
Assuming a validated polygraph technique is used, then an appropriate scoring system must be applied to the data. There are several scoring systems in use that convert the raw data (from the charts) into numbers. This conversion is done by hand, by the examiner using a set of rules established for that particular system. Depending on the system used, arbitrary “cutting scores” are used to make a decision of “deception indicated” or “no deception indicated.” The problem with most of these “cutting scores” is that they are arbitrary. Depending on which research study you look at, the decision accuracy ranges from 75 to 99 percent. This is because of the arbitrary cutting scores used and the inclusion of Screening polygraph tests in the results..
RECENT UPDATES TO SCORING METHODS
The latest of these scoring systems, called the Empirical Scoring System (ESS), actually converts these scores into probabilities based on recent scientific research. In other words, when a decision is made using ESS the examiner can give you an exact probability of error based on the 10,000 exams used in the study. For example, if someone scores a +3 in a specific issue exam, examiners can now report that person as truthful to a certainty of 93.3%. If someone else scores a -5, examiners can report that person as deceptive to a certainty of 96.8%.
COMPUTER SCORING IS UNRELIABLE
Computerized scoring algorithms have not yet been able to replicate the accuracy of an experienced polygraph examiner when scoring charts. There are simply too many anomalies and artifacts that can interfere with an exam that a computer would incorrectly try to interpret. A recent poll of examiners found that the computerized score disagreed with the official score 25% of the time, so it really has a long way to go.
SO HOW ACCURATE IS POLYGRAPH TESTING?
If the exam is done correctly by a qualified examiner, the accuracy of an exam is determined by the test score. Assuming the test is administered to current standards, the equipment is up-to-date and fully operational, and the scoring system is applied correctly, the final score of a particular exam actually determines the accuracy of that particular exam. It is unscientific to allow the judgment of an examiner to make the decision of what constitutes “truth” or “deception.” Science now allows us to establish the accuracy of each individual polygraph exam. As we have seen with other sciences, we are witnessing the evolution of lie detector testing which is moving us another step closer to widespread courtroom acceptance.