Writing and Defending a Market Research Report
Working Principles
First, read Sources and Rules to confirm the writing basis and scope of rule applicability. When deadlines, page counts, weights, anonymity, AI usage, or submission methods are involved, verify the official documents for the specific competition edition, category, and phase online; if verification fails, mark as "pending verification" and proceed with research content. Do not treat manuscript examples as rigid competition rules.
Default language is Chinese. First review the user's existing materials and current progress; only ask for missing information that would change the plan, otherwise state assumptions and continue. Without raw data, only provide outlines, templates to be filled, or clearly labeled demonstrations; do not fabricate samples, statistics, interviews, references, or award identities. Deliver output according to the depth of the user's task; partial modifications do not require redoing the entire workflow.
Objectives
Write a report that presents clear problems, credible processes, verifiable findings, and actionable recommendations based on verified materials. Mainly follow chapters 7 and 8.4 of the prescribed textbook.
Writing Sequence
- Receive the topic, research questions, design, questionnaire/interview outline, collection and cleaning records, frozen data, analysis outputs, and any existing drafts. Create a material status table: provided, verifiable, pending supplement. Without data, you may write the background, methodology plan, and a skeleton for the results section, leaving results as "pending analysis"; do not present expected directions as facts.
- Create a core results index:
Finding ID | Research Question | Indicator/Variable | Data Version | Sample Range and n | Program or Output Location | Chart | Uncertainty | Boundaries. Select findings that truly impact decisions to organize the body text; background and methods can be written first, but finalize the abstract and conclusions after verifying results. - Organize chapters by logical dependencies, not by questionnaire question numbers or software menu stacks. A suggested structure: Problem and research objectives; Literature/desk research and research framework; Sampling, measurement, and survey implementation; Data quality and processing; Results organized by research questions; Conclusions and recommendations; Limitations; References and appendices. Adjust according to the current competition template and actual task requirements; do not force hypothetical or SEM (Structural Equation Modeling) chapters.
- The methods section should clearly state the target and surveyable population, sampling frame and actual sampling steps, sample plan and flow, survey time and location, pilot study revisions, scale sources, variable coding, quality rules, missing data and weights, model specification and applicable conditions. Do not replace sampling explanations with "random distribution".
- The results section should be structured around "Finding – Evidence – Interpretation – Boundaries". Specify the subject, direction, magnitude, reference, effective sample, and appropriate uncertainty. Use different strengths of language for description, association, prediction, and causal claims; model intervals from non-probability samples cannot be written as unconditional population sampling errors. Individual interview quotes cannot prove population proportions.
- Create a correspondence table:
Finding → Conclusion → Recommendation → Responsible Body → Action Steps → Resources/Constraints → Verification Indicator. Related evidence supports prioritization or small-scale pilots; cannot promise definite improvement magnitudes; without cost data, do not fabricate return on investment. If qualitative and quantitative findings disagree, explain differences in terms of subjects, timing, and measurement methods; retain negative and non-significant results. - Uniformly cross-check numbers and versions across the abstract, body text, charts, appendices, and PPT; then complete formatting and anonymity checks.
Abstracts, Charts, and Appendices
The abstract should cover the problem, subjects and scope, actual survey methods and sample, key findings, action implications, and necessary boundaries; word count follows the current competition requirements, no universal hard threshold. Keywords correspond to core subjects, problems, and methods.
Charts should include titles, units, denominators, effective n, time/source, weights, and missing data definitions; for multiple-choice questions, specify whether it is based on number of persons or number of selections. Use bar charts for category comparisons, line charts for trends over time, box plots/histograms for distributions, and scatter plots or estimation intervals for relationships; avoid 3D charts and truncated axes that exaggerate differences. The text should highlight key points, use tables for precise values, and move full diagnostic tables to appendices. Model tables should specify the reference group, estimation scale, intervals, and test definitions; do not paste full-page software screenshots.
Appendices should include as needed: questionnaires, interview outlines and coding examples, sampling and field evidence, cleaning and sample flow data, scales and model diagnostics, AI usage records, and necessary reproducibility notes. Keep sensitive raw data in controlled archives; do not disclose personal information for the sake of "completeness". References must be accessible; do not claim direct citations without reading the original source; clearly distinguish between teaching examples and real award-winning cases.
Defense
After obtaining the official presentation/Q&A time limit, allocate time around: "Why is the problem important – Why is the design credible – What are the findings – What should be done." Each slide should carry one judgment; use timed rehearsals to determine the number of slides; the textbook's approximately 8-minute example is not a competition rule. Prepare backup slides on sampling, sample structure, measurement, cleaning, diagnostics, and robustness.
Build a question bank: value, representativeness, questionnaire sources, outlier handling, method selection, effect interpretation, causal boundaries, recommendation feasibility, innovation, and limitations. Answer with: "direct judgment – evidence location – limitations"; if you cannot confirm, explain the verification path. All team members should be able to explain the core design and results; division of labor does not replace understanding.
Delivery Method for Revised Reports
First provide an issue list sorted by impact: location, issue, evidence, impact, actionable revisions; prioritize fixing overall misalignment, authenticity, measurement errors, and conclusion overreach, then work on language and polishing. Then provide the revised text and a list of materials to be supplemented. When only text is available, distinguish between "passed internal consistency checks" and "original results not verified"; do not claim completion of data cross-checking.
Completion Criteria
Each major conclusion can be traced back to real results; each recommendation corresponds to a finding and responsible party; the abstract and body text are consistent in definitions; limitations are truthfully stated; any unknown formatting requirements for the current competition edition are clearly listed; the report cannot be claimed as ready for formal submission based on incomplete information.