Fellowship Interview Analysis

This project analyzes structured JSON data from oncology fellowship interviews. The dataset includes fields such as:

• question
• answer
• answer_quality
• interviewer_name
• interview_location
• question_category
• answer_word_count

The Python script produces summary tables and visualizations to help identify strengths, weaknesses, and preparation strategies for future interviews.

---

Requirements

• Python 3.9 or later

• Packages:

  pip install pandas numpy matplotlib scipy statsmodels streamlit

  Or install from requirements.txt:

  pip install -r requirements.txt

---

How to Run

1. Prepare Your Files

• Save the analysis script as fellowship_interviews.py
• Place your interview dataset JSON file (e.g., interviews.json) in the same directory

2. Open PowerShell

Navigate to the folder where your files are stored:

cd "C:\Users\Jake\InterviewAnalysis"

3. Run the Script

Run the script with your JSON file:

python fellowship_interviews.py ".\interviews.json" --outdir ".\analysis"

• Replace interviews.json with your actual dataset filename
• The --outdir flag sets the output folder (default: analysis/)

4. Open Results

To quickly open the results folder after running:

Start-Process ".\analysis"

Interactive Streamlit App

For an interactive experience with visualizations and exploration:

1. Install Streamlit (if not already installed):

   pip install streamlit

2. Run the Streamlit app:

   streamlit run streamlit_app.py

3. Use the app:

   • Upload your JSON file through the sidebar
   • Explore different tabs: Overview, Plots, Statistics, Details
   • View interactive plots and statistical summaries
   • No files are saved - all analysis happens in your browser

---

Outputs

1. CSV Tables (in analysis/):

   • overall_summary.csv — overall stats
   • freq_by_question_category.csv — frequency of categories
   • quality_by_question_category.csv — mean + CI for each category
   • wordcount_by_question_category.csv — word count descriptives
   • quality_by_interviewer.csv, quality_by_location.csv
   • correlation_wordcount_quality_overall.csv, _by_category.csv
   • anova_quality_by_category.csv, kruskal_quality_by_category.csv
   • wordcount_sweetspot_by_category.csv — optimal word count ranges

2. Figures (PNG):

   • Heatmap: mean quality by category × location
   • Scatterplot: word count vs quality with LOWESS fit
   • Violin/Boxplot: quality distributions by category
   • Histogram: distribution of word counts

3. Console Summary (printed in PowerShell):

   • Overall stats
   • Top categories by mean quality
   • Correlation results
   • ANOVA/Kruskal results
   • Word count “sweet spot” ranges per category

---

Example Run

cd "C:\Users\Jake\Documents\InterviewAnalysis"
python fellowship_interviews.py ".\interviews.json" --outdir ".\analysis"
Start-Process ".\analysis"

---

Next Steps

• Use the sweet spot word count ranges to pace your answers.
• Focus practice on low-scoring categories identified in the CSVs.
• Review institution-specific profiles to anticipate themes.
• Track improvements by re-running this script after more interviews.