In-game willingness to solve hard levels with complex code is positively associated with posttest coding knowledge, controlling for pretest¶

Claim · All claims
Evidence · 1 study (2 entries) · associational r1 · q2 · i3 large
Currency · newest study 2016 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q2 i3 Hards Solved Complex correlated with posttest score (r = 0.76, p < .01) and explained 50.2% additional unique variance in posttest controlling for pretest. → Malkiewich 2016 q2 i3 The correlation survived the Benjamini-Hochberg correction and partial correlation controlling for pretest. → Malkiewich 2016

Evidence¶

Malkiewich 2016¶

Malkiewich, L. J., Lee, A., Slater, S., & Chase, C. C. (2016). Tenacious Assessments: Using In-Game Behaviors to Measure Student Persistence and Challenge Navigation. https://eric.ed.gov/?q=Tenacious+Assessments+Using+In-Game+Behaviors

q2 · i3 · associational · r1

Correlational analysis of gameplay log measures against paper posttest scores for students playing a block-based coding game (log-data subset N=18). "Hards Solved Complex (r = 0.76, p < .01)" led the challenge-approach measures correlated with learning.

"These included Hards Solved Complex (r = 0.76, p < .01), Hards Started Complex (r = 0.71, p < .01), and Levels Solved Complex (r = 0.70, p < .01). All three of these correlations maintained significance after the Benjamini-Hochberg correction."

Malkiewich 2016 (2)¶

Malkiewich, L. J., Lee, A., Slater, S., & Chase, C. C. (2016). Tenacious Assessments: Using In-Game Behaviors to Measure Student Persistence and Challenge Navigation. https://eric.ed.gov/?q=Tenacious+Assessments+Using+In-Game+Behaviors

q2 · i3 · associational · r1

Step-wise linear regression entering pretest first, then the most strongly correlated measures. Hards Solved Complex "explained 50.2% additional unique variance in posttest score, controlling for pretest"; Hards Started Complex added no further variance because the two were highly correlated (r = 0.90).

"Hards Solved Complex, when added to the model, explained 50.2% additional unique variance in posttest score, controlling for pretest, F(1,15) = 20.09, p < .01."

Discussion¶

Related Claims¶

• Downgrading to simpler code after failure on personally hard levels is negatively associated with persistence on an unsolvable task
• Relative-Hards Solved Complex, a log measure of using complex code on personally difficult levels, predicts time spent on an unsolvable challenge task
• Traditional challenge-navigation measures (level choice, replays, careful retries, hint reading, coding time) did not relate to learning outcomes — reports the opposite
• In gamers' interview reports, sustained gaming engagement appeared closely related to controlling an optimal level of challenge: games too hard were often abandoned and games too easy lost their interest — related
• Gaming the system in the self-explanation step predicts delayed posttest scores beyond pretest, adding 3 percent of variance — related