Designing a Research Question

Research begins with a question — not a vague curiosity, but a specific, testable question that can be answered through data analysis. Strong research questions have particular characteristics: they are answerable with available data, narrow enough to be manageable, and important enough to justify the effort.

Rather than asking "Is sea level rising?" (which is already well-established), advanced students ask more specific questions: "Which regions are experiencing accelerating sea level rise, and what ocean conditions correlate with acceleration?" Or: "How well can machine learning models predict sea level anomalies six months in advance?"

To design your research question, start by reviewing what is already known. What aspects of sea level change have been extensively studied? Where are knowledge gaps? Understanding scope helps ensure your question is answerable with available data while still providing valuable new insights.

Strong vs Weak Questions

STRONG

"How does the rate of sea level rise in the Gulf of Mexico compare to the global average over the past 30 years?"

WEAK

"Is sea level rising?"

STRONG

"Can AI models detect acceleration in regional sea level trends that traditional linear regression misses?"

WEAK

"What does AI say about sea level?"

Selecting and Evaluating Data Sources

public

NOAA/NASA Satellite Data

Official government datasets from Jason-3 and other altimetry missions. Processed and quality-controlled.

+ Global coverage since 1992
+ Standardized processing
- Requires understanding of data format

anchor

PSMSL Tide Gauge Network

Permanent Service for Mean Sea Level — global tide gauge records going back over 150 years.

+ Long historical records
+ Local-scale detail
- Requires land movement corrections

science

Research Publications

Peer-reviewed papers that have already analyzed and interpreted sea level data.

+ Expert analysis included
+ Methodology documented
- May not include raw data

Data Quality Checklist

check_box_outline_blankWhat is the source authority and reputation?

check_box_outline_blankWhat time period does the data cover?

check_box_outline_blankWhat processing or corrections have been applied?

check_box_outline_blankAre there known gaps, errors, or limitations?

check_box_outline_blankHas this data been used in peer-reviewed research?

check_box_outline_blankIs the methodology documentation available?

AI-Assisted Research Methodology

Modern sea level research increasingly uses AI not just for analysis, but throughout the research process. Machine learning can help identify relevant datasets, detect quality issues, suggest analysis approaches, and even assist with literature review.

However, AI assistance requires critical oversight. AI tools can make mistakes, hallucinate non-existent data, or apply inappropriate methods. Researchers must understand what the AI is doing and verify its outputs against domain knowledge.

AI Can Help With:

- Processing large datasets quickly
- Identifying patterns and anomalies
- Suggesting relevant research papers
- Generating initial visualizations

Humans Must:

- Define research questions and scope
- Verify AI outputs for accuracy
- Apply domain expertise to interpretation
- Take responsibility for conclusions

Critical Evaluation: Is the AI Right?

AI systems can produce confident-sounding results that are completely wrong. Critical evaluation — checking whether AI outputs make sense — is essential for responsible research.

This means comparing AI findings with established science, checking for physically impossible results, and understanding the limitations of the specific AI methods used.

Red Flags to Watch For

warningResults contradict well-established physics without explanation
warningPredictions are far outside historical ranges without clear reason
warningAI cannot explain how it reached its conclusions
warningResults change dramatically with small input changes
warningAI was trained on data from different regions or time periods

Communicating Data Findings

For Scientific Audiences

checkInclude detailed methodology and data sources
checkReport uncertainty ranges and confidence levels
checkAcknowledge limitations and potential errors
checkCompare findings with existing literature

For General Audiences

checkLead with key findings in plain language
checkUse clear visualizations that don't require expertise
checkExplain why the findings matter for real life
checkAvoid false precision — round appropriately

Capstone Project Guidelines

Your capstone project will demonstrate mastery of the research skills developed throughout this module. You will design and execute an original analysis of sea level data, critically evaluate your results, and communicate your findings.

Requirements

1. Formulate a specific, answerable research question
2. Identify and justify your data sources
3. Describe your analysis methodology (AI and/or traditional)
4. Present findings with appropriate visualizations
5. Discuss uncertainty and limitations
6. Connect findings to broader scientific context
7. Prepare both technical and general-audience summaries

Level 5 — Advanced Research