Meta-informational Cue Inconsistency and Judgment of Information Accuracy: Spotlight on Intelligence Analysis

Meta-information is information about information that can be used as cues to guide judgments and decisions. Three types of meta-information that are routinely used in intelligence analysis are source reliability, information credibility and classification level. The first two cues are intended to speak to information quality (in particular, the probability that the information is accurate), but classification level is not (it is intended as information about information sensitivity). Two experiments involving professional intelligence analysts (N = 25 and 27, respectively) manipulated meta-information in a 6 (source reliability) x 6 (information credibility) x 2 (classification) repeated-measures design. Ten additional items were retested to measure interindividual reliability (82 trials in total). Analysts judged the probability of information accuracy based on its meta-informational profile. In both experiments, the judged probability of information accuracy was sensitive to ordinal position on the scale and directionality. That is, the directionality of linguistic terms used to anchor the levels of the two scales led analysts to group the first three levels of each scale in a positive group and the fourth and fifth levels in a negative group, with the directionally neutral “cannot be judged” options falling between these groups. Critically, as reliability and credibility cue inconsistency increased, there was a corresponding decrease in intra-analyst reliability, inter-analyst agreement, and effective cue utilization in judgment. In Experiment 2, there was also evidence of a secrecy bias among analysts in which “TOP SECRET” intelligence was judged more likely to be accurate than identical intelligence marked “OFFICIAL.”