Correlation between scores on the test and confidence measure
Analysis based only on two questions: coding test score and self-reported confidence.
Sample used
11
valid responses (both score + confidence)
Total submissions: 22
Excluded entries with missing or "-" values
Score stats
62.0
Avg
60
Median
28–92
Range
Confidence (1–5)
2.64
Avg
2
Median
1–5
Scale
r
Correlation result
There is a very weak negative correlation between girls' test scores and how confident they felt in their skills.
Pearson r
≈ -0.03
Interpretation: essentially no linear relationship
Confidence phrases were mapped to a 1–5 scale: 1 Very low, 2 Low, 3 Moderate, 4 High, 5 Very high.
Scores vs. Confidence
Each dot = one response
5
4
3
2
1
28
40
50
60
70
80
92
Very low
Low
Moderate
High
Very high
Confidence distribution
Very low 9.1%
Low 45.5%
Moderate 27.3%
High 9.1%
Very high 9.1%
Key insights
- Correlation between score and confidence is effectively zero (r ≈ -0.03). Higher scores did not reliably align with higher confidence.
- High-score, low-confidence cases exist (e.g., scores 89 and 92 paired with low confidence), indicating underconfidence among some high performers.
- Mid-to-high confidence appears at mid scores as well (e.g., 68 very high confidence, 70 moderate), reinforcing the weak relationship.
- Average score is 62 with a median of 60, while median confidence is Low (2 on a 1–5 scale), suggesting many learners feel uncertain despite reasonable scores.
What this means
Confidence seems influenced by factors beyond test performance (e.g., prior exposure, debugging struggles, or anxiety). Support strategies should address mindset and experience, not just content difficulty.
Notes on method
- Only entries with a numeric score and a provided confidence statement were analyzed (N = 11).
- Confidence statements were categorized into a 1–5 scale based on wording: Very low, Low, Moderate, High, Very high.
- Pearson correlation was computed between numeric scores and the 1–5 confidence scale.