A molecular genetics learning progression web

Inspiration

Currently there is little understanding of how different ideas about molecular genetics connect together. We are familiar with using statistical significance to evaluate the effect of science instruction. However, we have little qualitative information about how students connect ideas that is also supported by quantitative evidence. Concept maps and student interviews are traditionally used in qualitative analyses. But with causal model search methods, we can also utilize learning progression-based assessments (where each response corresponds with a specific idea) make qualitative arguments quantitatively.

What it does

Gives a qualitative causal map for how college undergraduate students' ideas change between the beginning and end of an introductory biology course.

How we built it

We first imported the pre- and post-test data into TETRAD. We then manipulated the data by removing cases with missing data (<10% of the cases). We then added knowledge: we forbade directional paths from more advanced to simpler concepts since it is reasonable to expect that students will master a simpler concept before mastering more advanced concepts (based on conceptual change theory). After this, we implemented the Fast Greedy Search (FGS) model search procedure using penalties between 0.1 and 40. We found that a penalty of 4 made the strong connections between ideas in the same topic visible while also elucidating many of the connections in ideas between topics. This made for a model that was parsimonious, but also useful and intellectually satisfying.

Challenges we ran into

Certain edges had no directionality. Lowering the penalty such to obtain directionality would have yielded a model with unnecessary complexity which would reduce its tractability. However, other data provide information on directionality, such as analyses of item difficulty from prior research, our content understanding of biology, and qualitative literature on how students understand and connect molecular genetics ideas.

Accomplishments that we're proud of

We found that the FGS algorithm replicated our theoretical understanding of the progression despite adding only minimal knowledge (described above) to the data. The match between purely empirical results derived from FGS and theory derived from the multitude of other data from molecular genetics learning progressions research supports both the quality of the data as well as the generalizability and robustness of the FGS search algorithm.

What we learned

Less sensible connections between ideas are replaced by more sophisticated connections between ideas. Before instruction, all ideas were (directly or indirectly) connected to the central idea that there is phenotypic variance between species. However, after instruction, all ideas were (directly or indirectly) connected to the central idea that genes exist. The idea of phenotypic variation can come from popular science and observations of nature. However, the centrality of the idea of genes which was facilitated in this course is a more abstract and scientific notion that underlies all of molecular genetics.

What's next for A molecular genetics learning progression web

We are currently facing a trade-off between the need to obtain directionality information and the need to achieve model simplicity. Current empirical data don’t provide directionality information about some edges. We want a simpler model, but we also want information about directionality. To negotiate this, we need more data to detect more subtle trends. We can also use other empirical and theoretical information to deduce directionality such as item difficulty measures (students will master easier items before difficult concepts), current understandings of how students conceptualize molecular genetics, and understanding of biological sciences content. Once we make a case for directionality based on one or more of these methods, we can test our hypotheses within the framework of path analysis/structural equation modeling.
We also would like to use our findings as a basis for exploring more fully how students' ideas change over time. We recommend a dynamic temporal analysis across different majors and levels in college. We can also explore the questions: How do the connections among ideas change between majors/non-majors, grades kindergarten through college, and advanced placement (vs. non) experience? The effect of motivation on academic/cognitive outcomes is also of significant interest in current research on conceptual change. How does motivation for taking the course affect performance in course and how students generate connections between ideas? Finally, the theory of conceptual change suggests that prior knowledge is of key importance to students' acceptance and mastery of new concepts. This leads to the question: How do performance outcomes in basic biology courses influence outcomes in more advanced biology courses?

Built With

• cytoscape
• tetrad