Effectiveness of a teachable machine-based instructional program for developing electronic concepts among eighth-grade students

This study examined the effectiveness of a Teachable Machine-based instructional program for developing electronic concepts among eighth-grade students in the Palestinian technology curriculum. A quasi-experimental, nonequivalent pretest-posttest control-group design compared two pre-existing classes, since students were not individually randomized. The sample comprised 80 male students from one school, with 40 in each group. The experimental group completed 12 lessons over six weeks using a sequence that integrated Google Teachable Machine for supervised image classification with PictoBlox for linking classified images to concept names and explanations; the control group studied the same unit conventionally. Achievement was measured with a 25-item Electronic Concepts Test. The experimental group achieved a higher post-test mean (M = 24.20, SD = 0.88) than the control group (M = 7.75, SD = 2.56), t(78) = 38.44, p < .001, 95% CI [15.60, 17.31], Cohen's d = 8.60, and Black's modified gain coefficient was 1.57. A carefully structured visual-classification sequence can support immediate acquisition of electronic concepts, but the exceptionally large effect, the single-school sample, and the absence of delayed measurement require cautious interpretation and independent replication.