The geoscience education research (GER) enterprise faces a challenge in moving instructional resources and ideas from the well-populated domain of "practitioners' wisdom" into the research-tested domains of St. John and McNeal's pyramid of evidence (this volume). We suggest that the process could be accelerated by seeking out clusters of instructional materials that share a common design pattern and then researching the affordances, pitfalls, and mechanism for the cluster. As a proof of concept, we sought design patterns that would support the learning goal that students be able to use authentic geoscience data to make inferences about Earth processes and decisions about Earth-human interactions. Analyzing intro-level modules from the InTeGrate project revealed six such patterns, each of which was used in at least three modules. For each pattern, we describe the instructional sequence, provide an illustrative example, describe the variability observed within the pattern, hypothesize mechanisms by which the instructional sequence might lead to improved learning, and pose potential research questions. In order from most to least abundant, the observed design patterns are "Data Puzzles," "Pooling Data to See the Big Picture," "Make a Decision or Recommendation," "Predict-Observe-Explain," "Nested Data Sets," and "Deriving a New Data Type." We conclude that a research program based on design patterns for teaching and learning with authentic geoscience data is viable, and we offer recommendations for moving forward.