The Argument-Based Approach
Reading a quantitative study as a question–answer contract
Contents
A finalized quantitative study can be read as a question–answer contract. The research question declares what answer is owed. The methods, results, supplementary materials, and stated reservations provide the record from which a reader decides whether the reported answer is warranted. This page summarizes the reading approach the chapters share. Part One Chapter 1, The Quantitative Research Journey, develops it in full with worked examples and is the place to learn it.
One question–answer pair, one argument, linked layers
One question–answer pair has one argument with linked layers. Pairs are counted by the substantive answers a study separately owes, not by its estimates, tests, or coefficients, so several results can contribute to one answer, and a study with several separable questions needs several arguments.
The statistical layer (STAT) concerns the recorded variables, the statistical support, the analysis, and the statistical conclusion. The substantive layer (SUB) connects that work to the people, practices, meanings, and setting named by the question. Substantive reasoning enters when a question and its measures are chosen, as well as when results are interpreted.
STAT-E evidence + STAT-J justification → STAT-I inference → STAT-C claim
↑ situates · augments and contextualizes · interprets · bounds ↑
SUB-E evidence + SUB-J justification → SUB-I inference → SUB-C answer
Evidence, justification, inference, and a bounded claim
In each layer, Evidence and Justification support an Inference, which supports a bounded Claim: E + J → I → C. Each layer contributes all four functions.
| Function | Statistical layer | Substantive layer |
|---|---|---|
| Evidence (E) | The observations, variables, and procedures through which the analysis data were produced or acquired | The people, setting, activities, and meaning those data represent |
| Justification (J) | The reasons and conditions that support the intended statistical comparison and inference | The domain reasons that support measurement meaning, implementation, explanation, and applicability |
| Inference (I) | The analysis connecting evidence and justification to a statistical conclusion: estimates, uncertainty, comparisons, and checks | The interpretation of that statistical inference in substantive terms, keeping its uncertainty and conditions |
| Bounded claim (C) | The statistical conclusion about the estimand, with its uncertainty, conditions, and scope | The question-facing answer, with its substantive scope, qualifications, and unresolved matters |
Across the layers, each substantive component works on its statistical counterpart: SUB-E situates STAT-E, SUB-J augments and contextualizes STAT-J, SUB-I interprets STAT-I, and SUB-C bounds the substantive reach of STAT-C. These contributions give statistical work its meaning for the question. A software result is therefore part of the inference, not the whole of it.
Six questions for evaluating an argument
Reading identifies how a study’s information supports its answer. Evaluating the argument’s quality is a separate step that asks whether that support is adequate for the question and the answer. Six evaluation questions guide it, in three groups.
| Group | Evaluation question | What it asks |
|---|---|---|
| Alignment | Question alignment | Whether the question, estimand, design, evidence, analysis, and answer stay directed at the same aim |
| Support | Evidence integrity | Whether the observations and contextual records, their production, and their representation are trustworthy, traceable, and appropriate |
| Support | Justification adequacy | Whether the statistical, design, measurement, and contextual reasons warrant the proposed moves |
| Support | Inferential adequacy | Whether the reasoning within both layers is adequate, including estimation, uncertainty, checks, sensitivity, and interpretation |
| Translation | Transition fidelity | Whether the links between the layers keep meaning, uncertainty, design status, and scope |
| Translation | Claim calibration | Whether the strength and breadth of the reported answer match its support |
Each question receives a reasoned judgment rather than a score, and the whole question–answer pair receives one conclusion: the answer is supported as stated, supported only after narrowing or revision, not supported, or not assessable from the accessible record. Chapter 1 works through these judgments on an example and also outlines how a researcher constructs one argument.
Where the approach comes from
The approach is Toulmin-informed: it keeps evidence, justified inferential steps, qualified claims, and explicit reservations from Toulmin’s account of argument (Toulmin, 2003) without treating its components as one-to-one replacements for Toulmin’s roles. LeMire’s link between statistical testing and subject-matter claims (LeMire, 2010) is a related statistical precedent, and Kane’s argument-based treatment of the interpretation and use of test scores (Kane, 2013) is a related precedent in measurement. The linked layers, the evaluation questions, and the construction workflow bring these influences together for the chapters rather than reproduce one source’s arrangement.
References
Argument chapters
Three argument chapters extend the approach beyond Chapter 1. They are optional enrichment: no numbered chapter in Part One or Part Two depends on them. A course can select a chapter with the statistical preparation it requires, without assuming completion of all three.
A. How to Read a Quantitative Study as an Argument
A complete worked reading of one study as an argument: the question–answer contract; one argument with linked statistical and substantive layers, traced across a report and its related materials; the reading of a reported regression argument; evaluating a question-level argument; and a six-stage construction cycle.
Read Argument Chapter A · Download PDF
B. Extending the Argument-Based Approach to Quantitative Studies
The approach carried from one question to several: separate questions and statistical strands; one complex argument with several strands; coordinated questions and their dependencies; evaluating answers and their coherence across questions; coordinated construction and revision; and transfer to a published experiment.
C. Arguments involving latent variables or an additional measurement layer
A reading of how observed indicators, a measurement model, and the interpretation of a latent quantity support a substantive answer. The chapter examines measurement assumptions, identification, model checks, interpretive validity, uncertainty, and sensitivity. It distinguishes model fit and reliability from evidence for the intended interpretation and carries unresolved measurement qualifications into the answer.
This chapter is not yet available.