Structural Clarity
Making Quantitative Reasoning Visible
Quantitative reasoning requires students to coordinate many parts of an argument. A research question shapes the design that can address it, the design determines which comparisons the evidence can support, and the analytic model determines what is actually estimated. Interpretation then gives those estimates substantive meaning, and the resulting claim depends on the relationships established throughout that process.
Experienced researchers often see these elements as parts of an organized system. Students may initially encounter the same work as a series of separate tasks: identify the variables, select a procedure, use the software, read the output, and report a result. Research on expertise shows that novices are more likely to organize problems around surface features, while experts recognize deeper disciplinary structure (Chi et al., 1981). The number and interaction of relevant elements also contribute to the complexity of quantitative work (Chen et al., 2023).
One of the central aims of my quantitative teaching is to make those relationships visible. Structural clarity has an established lineage in mathematics education as an instructional quality concerned with the organization and visibility of relationships within disciplinary content (Drollinger-Vetter, 2011; Pauli et al., 2024). In my teaching, structural clarity means helping students see how the parts of a quantitative argument constrain and give meaning to one another.
From Correct Pieces to Coherent Arguments
Statistics education has long treated statistical reasoning as a form of sense-making that extends across an investigation. Garfield (2002) describes statistical reasoning in terms of making sense of statistical ideas and information, while Wild and Pfannkuch (1999) place statistical thinking within an investigative process involving the problem, plan, data, analysis, and conclusions. GAISE similarly emphasizes conceptual understanding, engagement with data, statistical thinking, and appropriate use of technology (GAISE College Report ASA Revision Committee, 2016).
A recurring problem in my teaching appears when students understand individual pieces of an analysis but lose the relationships among them. Consider a study measuring the same participants on two occasions. A student may understand the variables, distinguish paired and independent samples when asked directly, operate the statistical software correctly, and interpret the output of either analysis. If the two measurements are analyzed as independent groups, the model estimates a different comparison from the one established by the research question and design.
The problem lies in the coherence of the argument. Research on students’ evaluation of complete statistical investigations shows how demanding this coordination can be (Pfannkuch, 2005), and inferentialist work in statistics education similarly emphasizes relationships among reasons, actions, evidence, and claims (Bakker & Derry, 2011; Bakker et al., 2017). A question about change within individuals creates one set of constraints, while a question about differences between independent groups creates another. The design establishes the comparison available in the data, the model represents that comparison mathematically, and the interpretation must remain compatible with what the model estimates.
I want students to see a statistical procedure as part of this larger argument. They should be able to explain what problem a procedure addresses, which features of the design make it appropriate, what quantity it estimates, and what conclusions that quantity can support. Statistical inquiry also changes as new information becomes available: a measure may prove inadequate, a model assumption may fail, or a new pattern may motivate a revised question (Wild & Pfannkuch, 1999). Students therefore need to recognize how a change in one part of an investigation affects the parts that depend on it.
Questions help make those relationships explicit. What comparison am I actually making? What does this coefficient estimate? If the design changes, which later decisions also change? Does the conclusion still describe the evidence produced by the analysis? I want these questions to become part of how students inspect the coherence of their own quantitative reasoning.
Making Dependencies Visible
Students can encounter the same quantitative relationship through several representations. A verbal explanation can make a conceptual distinction explicit, a diagram can display dependencies, a hand calculation can reveal how a quantity is constructed, and a spreadsheet can expose intermediate transformations. Statistical software can then carry out the same operations efficiently at a scale that would be impractical by hand.
The value of using several representations comes from what each one makes available for examination. Ainsworth (2006) argues that multiple representations support learning when students understand their functions and can coordinate information across them. In my courses, I use representations to help students recognize an underlying relationship as its form changes. A conceptual comparison should remain recognizable when it becomes a statistical model, and the quantities produced by software should remain connected to the question and design that gave them meaning.
Hand calculation can be useful when it exposes structure. When students first study variability, calculating deviations from the mean makes visible the relationship between individual observations and the center of a distribution. Positive and negative deviations cancel, squaring changes how distance is represented, averaging the squared deviations produces variance, and taking the square root returns the measure to the original scale. Following those operations gives students a way to see how the final quantity is constructed.
Research on conceptual and procedural knowledge supports treating these forms of understanding as mutually developing (Rittle-Johnson et al., 2001). Procedural knowledge can also include understanding the goals, conditions, and relationships that govern a procedure (Star, 2005). I therefore use calculations when they help students examine the quantitative structure being represented, especially when software would otherwise compress several consequential decisions into a single command.
Worked examples serve a similar purpose. Their value depends partly on directing attention toward principles and relationships underlying a solution (Atkinson et al., 2000). A useful quantitative example makes visible why a comparison is relevant, what information it requires, how the model represents it, and what the resulting evidence supports. Experienced researchers compress many of these decisions through practice; teaching often requires expanding them long enough for students to examine how they work.
Using Error Diagnostically
Structural clarity also shapes how I use student errors. A wrong final answer can indicate a problem somewhere in the reasoning, while the form and location of that problem provide more useful instructional information. The break may occur in the statistical concept, the calculation, the coding of a variable, the choice of model, the reading of output, or the interpretation of an estimate. A technically correct analysis can also produce an answer to a question different from the one the investigation was designed to address.
When an interpretation does not fit an analysis, I want students to trace the argument backward. The interpretation can be checked against the output, the output against the model specification, the model against the variables and design, and the design against the original research question. Wild and Pfannkuch (1999) describe statistical thinking as deeply interrogative, and that orientation gives error a productive role in quantitative instruction.
Suppose a student reports a difference between two groups even though the study measured change within the same individuals. The important issue extends beyond the name of the statistical procedure. The analysis has disconnected the comparison represented by the model from the comparison established by the design. Locating that break focuses attention on the relationship the student needs to understand.
The same process applies to interpretation. A student may calculate an estimate correctly and assign it a meaning that exceeds what the model supports. Tracing the reasoning back through the analysis helps identify where the evidentiary relationship changed. Learning to locate and repair these breaks gives students practice examining the organization of a quantitative argument instead of treating an error as an isolated incorrect answer.
Learning to See the Structure
The longer-term purpose of structural clarity is for students to recognize these relationships in unfamiliar quantitative work. I want them to determine what comparison a question requires, identify which features of the design matter, understand what an analytic model represents, and evaluate whether an interpretation follows from the evidence. Those capabilities depend on seeing the argument as a connected system whose parts can be examined in relation to one another.
Students also need to inspect their reasoning when results are unexpected. A surprising estimate can prompt examination of the model that produced it, the model can be checked against the design, and the interpretation can be checked against the quantity being interpreted. When those relationships are visible, students have something concrete to question, evaluate, and repair.
This becomes especially important when software produces technically polished output. A table of coefficients or a significance test can look authoritative even when the underlying analysis has become disconnected from the research question. Quantitative instruction therefore has to keep the statistical argument visible beneath the output so that students can evaluate how the result was produced and what the evidence permits them to conclude.
For me, structural clarity supports rigorous quantitative reasoning by making its internal organization available for inspection. Questions, designs, models, representations, estimates, interpretations, and claims form an argument whose relationships students should be able to explain. The goal is for students to understand how that argument holds together and to recognize when one of its connections needs to be reconsidered.
References
Ainsworth, S. (2006). DeFT: A conceptual framework for considering learning with multiple representations. Learning and Instruction, 16(3), 183–198. https://doi.org/10.1016/j.learninstruc.2006.03.001
Atkinson, R. K., Derry, S. J., Renkl, A., & Wortham, D. (2000). Learning from examples: Instructional principles from the worked examples research. Review of Educational Research, 70(2), 181–214. https://doi.org/10.3102/00346543070002181
Bakker, A., Ben-Zvi, D., & Makar, K. (2017). An inferentialist perspective on the coordination of actions and reasons involved in making a statistical inference. Mathematics Education Research Journal, 29, 455–470. https://doi.org/10.1007/s13394-016-0187-x
Bakker, A., & Derry, J. (2011). Lessons from inferentialism for statistics education. Mathematical Thinking and Learning, 13(1–2), 5–26. https://doi.org/10.1080/10986065.2011.538293
Chen, O., Paas, F., & Sweller, J. (2023). A cognitive load theory approach to defining and measuring task complexity through element interactivity. Educational Psychology Review, 35, Article 63. https://doi.org/10.1007/s10648-023-09782-w
Chi, M. T. H., Feltovich, P. J., & Glaser, R. (1981). Categorization and representation of physics problems by experts and novices. Cognitive Science, 5(2), 121–152. https://doi.org/10.1207/s15516709cog0502_2
Drollinger-Vetter, B. (2011). Verstehenselemente und strukturelle Klarheit: Fachdidaktische Qualität der Anleitung von mathematischen Verstehensprozessen im Unterricht. Waxmann.
GAISE College Report ASA Revision Committee. (2016). Guidelines for Assessment and Instruction in Statistics Education: College report 2016. American Statistical Association.
Garfield, J. (2002). The challenge of developing statistical reasoning. Journal of Statistics Education, 10(3). https://doi.org/10.1080/10691898.2002.11910676
Pauli, C., Lipowsky, F., & Reusser, K. (2024). Capturing the subject-specific quality of mathematics instruction: How do expert judgments relate to students’ assessments of the quality of their own learning and understanding? ZDM: Mathematics Education, 56, 893–905. https://doi.org/10.1007/s11858-024-01561-3
Pfannkuch, M. (2005). Characterizing Year 11 students’ evaluation of a statistical process. Statistics Education Research Journal, 4(2), 5–26. https://doi.org/10.52041/serj.v4i2.512
Rittle-Johnson, B., Siegler, R. S., & Alibali, M. W. (2001). Developing conceptual understanding and procedural skill in mathematics: An iterative process. Journal of Educational Psychology, 93(2), 346–362. https://doi.org/10.1037/0022-0663.93.2.346
Star, J. R. (2005). Reconceptualizing procedural knowledge. Journal for Research in Mathematics Education, 36(5), 404–411. https://doi.org/10.2307/30034943
Wild, C. J., & Pfannkuch, M. (1999). Statistical thinking in empirical enquiry. International Statistical Review, 67(3), 223–248. https://doi.org/10.1111/j.1751-5823.1999.tb00442.x