Statistical Reasoning¶

Learner Variable · All learner variables
Evidence · 17 claims (13 show a difference, 4 mixed) · 7 studies (4 causal, 1 associational, 1 qualitative, 1 design), q2–q3 · 1 of 7 report an effect size · 17 claims rest on one study

Description¶

Statistical reasoning is thinking about uncertainty and building mental models of real-world phenomena from data. The Learner Variability Navigator describes it as a cycle: learners pose a question about data, decide what data they need, gather it, organise, analyse and display it, and then describe what it shows and draw inferences to answer the question. It is assessed with questionnaires on understanding of variation (in chance settings such as dice, and in data and graphs) that pair multiple-choice items with written explanations, and with the data-analysis and probability subtests of diagnostic batteries.

LVN follows it from grade 3 to grade 10. It builds on measurement and proportional reasoning and connects to algebraic thinking through functions that describe data. Understanding of ideas like the average can grow when learners meet data that contradicts what they expected.

Implications¶

Context¶

Investigations in which learners pose their own questions and collect their own data follow the full statistical cycle.
Data that contradicts learners' expectations can prompt them to revise their understanding of ideas like the average.
Stereotype cues about who is good at mathematics can depress performance.

Target Learners¶

Upper-primary and secondary learners working with data, graphs and chance.
Learners with weaker listening or reading comprehension, for language-heavy data tasks.

Target Learning Objectives¶

Understanding of variation, averages and probability.
Reading, making and drawing inferences from graphs and data displays.

Claims¶

A bias-free experimental-probability subcategory (o**) emerged only in the die-rolling task [~M] — learners who differ on it differ in outcomes
Ability to Reason (SMP 2) benchmark levels improved linearly across all three time points (d ≈ 0.78) with no significant quadratic trend [+M] — instruction changes it
Attributing outcomes to Allah's will among some PSMTs functioned as probabilistic reasoning, not mere personalist belief [+M] — learners who differ on it differ in outcomes
Mathematically oriented PSMTs exhibit equiprobability and insensitivity-to-prior-probability biases [+M] — learners who differ on it differ in outcomes
Some outcome-oriented PSMTs (types o and o) predict single-trial outcomes, and type o thinkers conflate causality with conditionality [+M] — learners who differ on it differ in outcomes
Post-test mean differences favored the ChatGPT group on knowing, applying, and reasoning subscales and overall at α=0.05 [+M] — instruction changes it
Repeated Reasoning (SMP 8) benchmark levels improved linearly across all three time points (d ≈ 0.77) with no significant quadratic trend [+M] — instruction changes it
Risk status tends to persist: 75 percent of students significantly below benchmark at BOY remained so at EOY, and earlier grades showed lower year-to-year persistence [+M] — learners who differ on it differ in outcomes
Students showed a large, statistically significant pre-to-post improvement in ability to reason (SMP 2) in the Mindsets program (d ≈ 1.20) [+M] — instruction changes it
Students showed a large, statistically significant pre-to-post improvement in repeated reasoning (SMP 8) in the Mindsets program (d ≈ 1.19) [+M] — instruction changes it
Subjectively oriented reasoning emerged in the giving-birth context but disappeared in the die-rolling task [~M] — learners who differ on it differ in outcomes
Visualisation literacy positively predicts task accuracy in both conditions, and data comics show no equalising effect for lower-literacy students [+M] — learners who differ on it differ in outcomes
On a delayed assessment 1 to 3 months later, many students transferred their appreciation of variation to another context and could explain the meaning of variation [+M] — instruction changes it
Student groups' TinkerPlots text boxes for one student's data were mostly literal reports of plot aspects, and comparison text boxes for the two scenarios mostly made basic rather than reasoned comparisons [~M] — instruction changes it
Students produced a wide range of self-chosen hand-drawn representations, mostly frequency-based, but most summary statements only described the graph rather than its message [~M] — instruction changes it
After training, prospective teachers caught up with current faculty in reasoning but still had lower noticing scores [+M] — instruction changes it
Reasoning scores of both prospective and current faculty teachers increased significantly during the short pedagogical training, but noticing did not improve significantly in either group [+M] — instruction changes it

Measures¶

Statistical Variation Measure (Watson et al., 2003): Questionnaire examining student understanding of statistical variation, including understanding of basic chance measurement and graph reading, variation in a chance setting (e.g., results of throwing a die), and variation in data and graphs. Students respond to multiple choice questions but also have space to explain their responses which allows for a more comprehensive measure of their understanding.
KeyMath-3 Diagnostic Assessment (Connolly, 2007): In the Data Analysis/Probability subtest, students interpret tables and tally charts, and estimate probability.

Executive Function: noticing a gap between an expected and an actual value, a cognitive-flexibility demand, can deepen understanding of the average.
Algebraic Thinking: functions such as a linear relationship are used to describe data.
Stereotype Threat: stereotype cues can lower early-adolescent girls' performance on statistical reasoning, especially among girls who value mathematics.
Communication Skills: listening comprehension predicts performance on data-analysis and probability tasks.

Examples¶

Key Sources¶

Digital Promise. Learner Variability Navigator: Statistical Reasoning (Building Math 3-6, Building Math 7-10). lvp.digitalpromiseglobal.org. Licensed CC BY-NC 4.0.

Page metadata

Field	Value
`type`	`learner-variable`
`id`	`statistical-reasoning`
`title`	`Statistical Reasoning`
`description`	`A learner's ability to reason about data and uncertainty: posing questions, collecting and organising data, and describing and drawing inferences from it.`
`status`	`review`
`canonical`	`True`
`generated`	by `claude/unspecified` on `2026-10-16`
`trust`	unverified — no one has confirmed this page's content

Frontmatter is the page's machine-readable record — see the Schema & Guide.