College of Education research summaries

Let’s Code with GenAI: K-12 Teachers’ Self-Efficacy, Value Beliefs, and Coding Performance

A self-paced GenAI-supported professional development (PD) pilot

Kang, S., & Huang, W. (2026). Let’s code with GenAI: exploring K-12 teachers’ self-efficacy, value beliefs, and coding performance. AI, 7(8), 277. https://doi.org/10.3390/ai7080277

Purdue College of Education Cobrand

Purpose and Scope

  • Examines the feasibility and initial outcomes of a self-paced GenAI-supported PD module for K-12 teachers learning text-based coding and computational thinking
  • Focuses on self-efficacy, value beliefs, and coding/CT performance

Module and Pilot Design

Design

One Group Pre-Test/Post-Test

TBaCCT scale + Five-item coding/CT assessment

6
Coding/CT Concepts
3
Difficulty levels
10
Coding missions

Post-Test Scores Were Higher Across All Outcomes

Self-efficacy and Value Beliefs

Confidence rose across four belief measures

Coding Self-efficacy

3.55
Before
4.78
After

Large r=.85 | p<.0001

Teaching Efficacy

3.58
Before
4.60
After

Large d=1.31 | p<.001

CT Self-efficacy

4.27
Before
4.58
After

Moderate r=.42 | p=.015

Value Beliefs

4.56
Before
4.89
After

Medium d=.64 | p<.001


Coding/CT Performance

A modest but significant gain

3.88
Before
4.21
After

Out of 5 | p=.040 | moderate r=.35

11
Improved
19
Unchanged
4
Lower

Designing the next iteration

Pair on-demand AI support with sustained coding practice.

  1. Extend practice beyond one hour and add delayed follow-up.
  2. Use authentic classroom coding tasks and iterative feedback.
  3. Track prompt use, support intensity, and learner pathways.
  4. Compare against control conditions and test classroom transfer.