HOW A GEOLOGIST CAN GET LEAD ASTRAY: A VIDEO LOG STUDY EXAMINING HOW ERRORS IN OBSERVATIONS AND INTERPRETATIONS YIELD ERRORS IN GEOLOGIC MAPS

In this study, we integrate data capturing the physical actions, spoken thoughts, and navigation paths of geologists as they make a geologic map. Eight geologists, from novice (undergraduate) to expert (professional), wore a head-mounted video camera with an attached microphone to record their visible actions and their spoken thoughts, creating “video logs” while in the field. Two of the eight geologists, one novice and one expert, made similar errors on their final maps. Here we focus on their two video logs as sources of insight into the origin and evolution of those errors; we also contrast their spoken thoughts and actions with those recorded in two other video logs by participants (also one novice and one expert) who produced more accurate maps.

The participants mapped a field area for which there exists a consensus understanding of the underlying geology (i.e., an “answer key”). In addition to a head-mounted camera, participants also wore a GPS unit to record their position throughout the day. Both the GPS data and video logs are time-stamped, enabling the data sets to be synchronized. The videos were coded both for instances of visible actions (e.g., measuring strike and dip, breaking a rock sample for a fresh surface, or testing a rock sample with HCl) and for themes in spoken thoughts (e.g., procedural and declarative knowledge, reasoning, or metacognition).

From analyses of the video logs, we find that for the two novices, data collection drives model development. The differences in the accuracy of their maps are determined more by their interpretations of the data. Both experts use an aerial photograph of the field area to propose an initial interpretation of the underlying geology; their initial interpretations guide later data collection and model refinement. The difference in the accuracy of their maps reflects the differences in their initial interpretations and subsequent model development.