Page 706 - ISC PROCEEDINGS 21.4

Table 1. Summary of the refined measurement scales
Retained Composite mean used in
Construct Code
items SPSS

Perceived ease of use PEOU 5 PEOU_Mean

Perceived usefulness PU 5 PU_Mean

Social influence SI 4 SI_Mean

AI self-efficacy AISE 3 AISE_Mean

Trust in AI TR 3 TR_Mean

Intention to use AI tools for IU 5 IU_Mean
learning

Source: Compiled from the research model and the refined SPSS dataset.
4. Results
4.1. Sample description
The study analyzes 210 valid responses. This sample size is appropriate for the
refined model after the exclusion of observed variables that did not perform adequately
during the scale-review stage. All results reported below are based on the second-round
SPSS outputs obtained after removing SI3, AISE4, AISE5, TR3, and TR4.
4.2. Cronbach’s Alpha reliability analysis
The reliability results show that all refined scales reach acceptable levels of internal
consistency. Cronbach’s Alpha is 0.829 for PEOU, 0.817 for PU, 0.807 for SI after the
removal of SI3, 0.746 for AISE after excluding AISE4 and AISE5, 0.733 for TR after
excluding TR3 and TR4, and 0.851 for IU. All corrected item-total correlations exceed
0.500, indicating that the retained scales are sufficiently consistent to proceed to EFA and
regression analysis.
Table 2. Reliability results after scale refinement

No. of Cronbach's
Scale Retained items Conclusion
items Alpha

PEOU PEOU1–PEOU5 5 0.829 Accepted

PU PU1–PU5 5 0.817 Accepted

SI SI1, SI2, SI4, SI5 4 0.807 Accepted

AISE AISE1–AISE3 3 0.746 Accepted

TR TR1, TR2, TR5 3 0.733 Accepted

IU IU1–IU5 5 0.851 Accepted

Source: Compiled from SPSS outputs.
4.3. Exploratory factor analysis
After scale refinement, EFA was conducted on the 20 observed variables associated
with the five independent constructs. The results show that KMO equals 0.805, exceeding
the minimum acceptable threshold of 0.500, while Bartlett’s Test is significant at 0.000.

705