Project 2, Part 1 - Data Import & Variable Classification
Description & Justification Thus far in MR, we have focused heavily on the "why and how" of gathering data. Now, we are finishing the last components of the research process by taking data we collect (primary) / find (secondary) and preparing it for analysis and insight.
You cannot analyze what you do not understand. Before we can deploy advanced AI tools like Gemini to synthesize our findings, we must first deeply understand the structure and limitations of our data. Correctly identifying your variable types (Nominal, Ordinal, Interval, Ratio) now builds the conceptual foundation required for the data cleaning and advanced analysis you will execute in Part 2 of this project.
Deliverable Submit a single .ipynb notebook file. Using Google Colab is highly recommended for this assignment.
Instructions: The 5-Cell Notebook Structure
Your notebook must contain exactly 5 cells, ordered as follows:
Cell 1 (Markdown): Include your full name and a title for your notebook.
Cell 2 (Code): Import Dataset 1 (your pre-cleaned primary Qualtrics survey data) using the Pandas library, and print a clean preview of the data using the head() function.
Cell 3 (Markdown): Provide commentary identifying the specific variable types within Dataset 1. Bullet points are perfectly fine; full paragraphs are not required.
Cell 4 (Code): Import Dataset 2 (e.g. your secondary data) and print a clean preview using head().
Secondary Dataset Requirements: You may source this from anywhere (HuggingFace, GitHub, etc.), and it does not need to be business-related. However, it must contain a minimum of five columns. The dataset must feature a mix of categorical variables (nominal or ordinal) and continuous variables (interval or ratio).
Cell 5 (Markdown): Provide commentary identifying the specific variable types within Dataset 2.
Grading Rubric (500 Points Total)
Criteria Excellent (Full Credit) Needs Work (Partial Credit) Missing (0 Credit)
Notebook Setup
(50 Points)
Includes correct Name/Title info in a properly formatted Markdown cell.
Name/Title is missing details or placed incorrectly inside a code comment.
Entire setup cell is completely missing.
Data Imports
(200 Points)
Both primary and secondary datasets are imported without errors, and clean head() previews are printed for each.
Only one dataset is imported, or the data previews are missing/broken.
No data was successfully imported into the notebook.
Variable Commentary
(250 Points)
Accurate conceptual classification of all variables (nominal, ordinal, interval/ratio) in markdown formatting.
Classifications are provided but are partially incorrect, vague, or fail to demonstrate an understanding of measurement scales.
Commentary on variable types is entirely missing.
Research Topic & Question
My research topic is student involvement and academic success at Western Carolina University.
My research question is:
Does participation in campus clubs or organizations affect students’ academic performance and sense of belonging?
Target Sample
The target sample would be undergraduate students at Western Carolina University. I would try to collect responses from students across different majors and class levels (freshman, sophomore, junior, and senior). A good way to do this would be sending out an online survey through student email lists or campus social media pages.
Conceptual Model (Hypotheses)
I think that student involvement in clubs/organizations will positively influence both GPA and sense of belonging.
So basically:
Club participation → higher sense of belonging
Sense of belonging → better academic performance
Club participation → better academic performance
My expectation is that students who are more involved on campus feel more connected and motivated in school.
How I’ll Measure Each Variable
Club participation (independent variable):
“Are you currently involved in any campus clubs or organizations?” (Yes/No)
“How many clubs or organizations are you involved in?” (0, 1, 2, 3+)
“How often do you attend meetings or events?” (Never → Very often scale)
Sense of belonging (mediating variable):
Measured using agreement-scale questions (1 = Strongly disagree, 5 = Strongly agree):
“I feel like I belong at WCU.”
“I feel connected to other students.”
“I feel supported by the campus community.”
Academic performance (dependent variable):
Self-reported GPA range (Below 2.0, 2.0–2.9, 3.0–3.4, 3.5–4.0)
Survey Structure
1. Introduction + Informed Consent
Explain purpose of survey
Say participation is voluntary and anonymous
Ask if they agree to participate
2. Screening Question
“Are you currently an undergraduate student at WCU?” (Yes/No)
3. Club Participation Questions
Involvement
Number of clubs
Frequency of participation
4. Sense of Belonging Questions
Likert-scale statements about connection to campus
5. Academic Performance Question
GPA category
6. Demographics
Class year
Major
Age (optional)
Project 2 Part 2: Integrated Data Analysis & Synthesis
Total Points: 500
1. Data Export & Preparation
Now that you have your secondary data in Google Colab and your survey data cleaned, you must move them into your analytical environment.
- Export to CSV: In your Google Colab notebook, you can export your secondary dataset. You can ask Gemini to write a code cell for this, or use the following Python command:
df.to_csv('your_dataset_name.csv', index=False) - LLM Setup: Upload either your secondary dataset (CSV) and your cleaned survey dataset to an LLM of your choice (Gemini, ChatGPT, etc.). If you had at least 20 responses in your survey, you may use this. If you did not, you will need to use the secondary dataset.
2. Required Statistical Tests
You are required to conduct a total of (4) statistical tests. You must execute:
Two (2) Group Comparisons: Choose from a T-Test, ANOVA, or Chi-Square analysis to identify differences between segments. So you can do an anova and a chi-sq test, an anova and a t-test, etc. Just run it on different variables you find of interest.
Two (2) Regression Analyses: Perform a simple or multiple regression to identify relationships between continuous variables. You may run two simple regressions, two multiple regressions (if you have two dv's of interest), one+one, etc. But in either case, you need two separate regressions.
3. Apply what we learned
Once the AI has helped you run these tests, return to your Google Colab notebook. At the very bottom, create a new Markdown cell.
- Interpretation: Describe the results and business implications of your findings
- Integrity Rule: This must be written in your own words. Submitting raw, unedited AI output is considered plagiarism per the course policy. I am looking for your personal "Now What?"—what did you learn from this data that a marketer could actually use?. This means I expect brevity, and a concise description of 'what you did' for each dataset. E.g. "I ran a t-test, anova, and regression on this marketing dataset. The t-test showed that there was no difference in gym-visit frequency between men and women (t=2.5, p=.11). An ANOVA with follow-up tests showed that education level did significantly influence gym visit frequency (2.1 vs 2.5 vs 2.8, p<.05, f=113). Finally, a regression showed me that time spent at the gym significantly improved gpa, accounting for 40% of the variation in GPA (beta=1.5, rsq=.4, p<.05).
Part 2 Rubric (500 Points)
| Criteria | Description | Points |
| Statistical Breadth | Correct execution of 2 group comparisons and 2 regressions, meaning you tell the LLM which variables to use and which tests to run. | 200 |
| Photo evidence (free points) | Verify with a screenshot that you did submit a dataset to one of the LLM's (chat, gemini, grok, etc) and use it to perform above said tests | 100 |
| Synthesis & Implication | A clear, original Markdown summary in Colab explaining what tests you ran, their results of all 4 tests, and implications. | 200 |
| TOTAL | 500 |
A Note on the Reflection
I want to emphasize that for the synthesis section, clarity beats length. I am not looking for academic jargon; I want to see that you understand the relationship between the numbers the AI gave you and a real-world marketing decision. This means you must use your own words and interpretations. The only thing AI can help with is brainstorming, but pasting an LLM output is unacceptable and will result in the plagiarism policy in the course being applied.
Final Deliverable
- One screenshot verifying you used an LLM to successfully analyze a dataset. (100 points)
- .ipynb notebook covering part 1 and part 2, with markdown commentary described. This is your 'docx' file that I will be grading (900 points). Any other file type will not be considered.