Knowledge base from "Business Research Methods" (Thirteenth Edition) by Pamela S. Schindler. Use when applying academic and business research frameworks for research questions, design, sampling, measurement, data analysis, hypothesis testing, association, reporting, or integrated research projects.
Installer avec Codex ou Claude Copiez ce prompt, collez-le dans Codex, Claude ou un autre assistant, puis laissez-le vérifier la page du skill et l'installer pour vous.
Une commande directe contourne le prompt de vérification. Examinez la source avant de l'exécuter.
La commande reste sur une seule ligne. Faites défiler horizontalement pour la vérifier avant de la copier.
Vous préférez une copie locale ? Téléchargez les fichiers actuellement disponibles dans SkillsMP.
Explorateur de fichiers
21 fichiers
Affichage de SKILL.md
SKILL.md
Instructions source · Aperçu en lecture seule
name
RES500-Academic-Writing-and-Research-Skills
description
Knowledge base from "Business Research Methods" (Thirteenth Edition) by Pamela S. Schindler. Use when applying academic and business research frameworks for research questions, design, sampling, measurement, data analysis, hypothesis testing, association, reporting, or integrated research projects.
Without arguments: Load the core research workflow and decision rules below.
With a topic: Ask about sampling, Likert, missing data, ANOVA, or another indexed topic; load the relevant chapter before answering.
With a chapter: Ask for ch05 or a chapter title for the detailed, worked treatment.
Browse: Ask what chapters or supporting files are available.
When a question is not covered in the Core Frameworks, use the Topic Index to find and read the relevant chapter file. Treat this skill as a research-methods guide, not as a substitute for project-specific protocols, current statistical software documentation, institutional review, or professional judgment.
Core Frameworks & Mental Models
Start with the decision, not the method
Treat a management dilemma as a decision symptom or opportunity, not yet as a research question. Move through the Management-Research Question Hierarchy:
Use the hierarchy to keep every question traceable to an action. If a proposed measurement cannot be traced upward, it is probably scope creep. Use exploration to test the dilemma against internal records, published and digital sources, knowledgeable people, and process evidence before choosing a method.
Apply the value-of-information test before approving a study: define at least two action options, the decision variable that separates them, an unbiased decision rule, and the research cost. If the manager would choose the same option with or without new information, the study has negative value after cost and opportunity cost.
Run the five-stage research process as a connected loop
Clarify the research question: define the dilemma, explore, value the study, and budget it.
Design the research project: specify sampling, data collection, and the measurement instrument.
Collect and prepare data: field the instrument, code, edit, examine, and document the file.
Analyze and interpret data: distinguish findings from insights and test the evidence.
Report insights and recommendations: select evidence for the audience and decision.
Use the stages iteratively when new evidence changes an assumption, but do not skip a decision-critical activity. The output should narrow from data to information to findings to insight to an actionable recommendation.
Set design dimensions deliberately
Research design is a time-based blueprint, not a method label. Classify the objective, ability to manipulate variables, topical scope, measurement emphasis, complexity, collection method, environment, time dimension, and participant awareness. Use reporting or descriptive designs for what/who/where/how much; causal-explanatory designs for why; and experiments for controlled causal-predictive claims. Use qualitative methods for meaning, motive, process, and why; quantitative methods for magnitude, frequency, prediction, and hypothesis testing. Increase design complexity when the cost and risk of the decision increase.
Match the sample to the inference
Define the target population, case, population parameter, sample frame, number of cases, selection method, and recruitment protocol before sampling. Probability sampling gives known selection chances and supports precision and population generalization when the frame and procedures are sound. Nonprobability sampling is defensible for exploration, access, hidden populations, atypical cases, or nonrepresentative objectives, but it does not support unqualified confidence intervals. Accuracy comes before precision: a large sample cannot repair a biased frame, selection process, or instrument. Set sample size from variance, confidence, interval width, population size, and subgroup needs; take the largest requirement among critical variables.
Select collection methods by the information needed
Use qualitative research when the question is how or why; recruit for relevance, range, articulation, and interaction, then continue until data saturation.
Use observation for overt behavior, events, records, physical conditions, and natural context; separate factual coding from interpretation and choose event or time sampling.
Use surveys for articulated attitudes, motivations, expectations, knowledge, intentions, and remembered behavior; choose self-administered, telephone, personal, or mixed mode by access, privacy, speed, probing, cost, and error.
Use experiments when manipulating an independent variable is feasible. A causal claim needs covariation, temporal order, and elimination of plausible alternatives; random assignment and control improve internal validity, while replication and realistic settings support external validity.
Build measurement from the construct down
Use the chain construct -> operational definition -> indicant -> mapping rule -> variable. Select the weakest scale that answers the question without discarding needed information: nominal classifies, ordinal orders, interval adds equal distances, and ratio adds a meaningful zero. Scan four error sources: participant, situation, measurer, and instrument. Evaluate validity, reliability, and practicality together. A reliable measure can still be invalid, and a scientifically strong measure can fail if it is not economical, convenient, or interpretable.
Treat the instrument and data file as engineered systems
Build instruments in three phases: refine measurement questions; assemble non-question elements; organize, physically design, and pretest the complete instrument. Start with a preliminary analysis plan and dummy tables. Use screens, filters, transitions, buffers, and tested skip logic. Pretest every relevant mode and report the longest realistic completion time.
Stage 3 ends with a clean, auditable file, not merely returned responses. Preserve a CaseID and codebook. For open text, define context, sampling, and recording units and establish coding reliability. Edit for completeness, accuracy, and appropriate coding. Diagnose MCAR, MAR, and NMAR before deletion or replacement. Use visual exploratory data analysis before confirmatory analysis; every percentage needs a base and direction, and every outlier needs source tracing before correction or deletion.
Separate evidence types in analysis
For hypothesis testing, state H0 and HA, choose the test from sample count, independence, measurement scale, assumptions, and power, set alpha before examining results, compute the statistic, obtain the critical value or p value, and interpret both statistical and practical significance. Say fail to reject H0, not accept or prove it. Parametric tests are efficient when assumptions hold; nonparametric tests protect against unsuitable distributions or measurement levels.
For association, plot or cross-tabulate first. Correlation is symmetric association; regression assigns predictor and dependent roles and predicts with uncertainty. Neither establishes causation alone. Choose nominal measures such as phi or Cramer's V, ordinal measures such as gamma or Spearman's rho, and regression only after checking table shape, ties, linearity, outliers, residuals, subgroup structure, and practical usefulness.
Report for the audience effect
Use Audience-Centric Planning: analyze who receives the report, define what they should think, feel, and do, choose oral/written structure, then organize, visualize, compile, practice, and deliver. Convert data to information, insights, and actionable insights. Select only evidence needed for the decision; use Ethos, Logos, and Pathos without replacing evidence with emotion. Report limitations, protect confidentiality, disclose conflicting findings, keep recommendations within scope, and distinguish the researcher's fact-finding role from the sponsor's decision-making role.
patterns.md - reusable research techniques and workflows
cheatsheet.md - decision rules, selection tables, and fast diagnostics
Scope & Limits
This skill covers the content and methods extracted from Pamela S. Schindler's Business Research Methods, Thirteenth Edition. It does not replace current statistical documentation, institutional review requirements, project-specific sampling frames, or professional and legal advice. Apply the framework with the study's actual population, data, assumptions, risks, and decision authority.
