Validity Study of the General Aptitude Test Battery (GATB) Using Confirmatory Factor Analysis (CFA)

1. Introduction

The General Aptitude Test Battery (GATB) is one of the longestused occupational assessment tools worldwide. Developed by the U.S. Employment Service in the 1960s, it measures a broad range of vocational and cognitive abilities across eight scales: General Learning Ability, Verbal Aptitude, Numerical Aptitude, Spatial Aptitude, Form Perception, Clerical Perception, Motor Coordination, and Finger Dexterity.

Although the GATB has demonstrated high reliability, questions remain about its construct validityspecifically, whether the eight scales truly represent distinct latent factors as intended. Confirmatory Factor Analysis (CFA) provides a rigorous statistical framework to test theoretically derived factor structures against observed data.

2. Construct Validity and the Role of CFA

Construct validity concerns the degree to which a test measures the theoretical traits it claims to assess. In the case of the GATB, the hypothesized model posits eight correlated factors. CFA evaluates this model by estimating factor loadings, error variances, and interfactor correlations, allowing researchers to:

• Confirm whether each item loads significantly on its designated factor.
• Examine the overall fit of the hypothesized factor structure.
• Compare alternative models (e.g., a higherorder general ability factor).

3. Methodology

3.1 Participants

A sample of 1,250 job applicants (57% male, 43% female) from diverse industries in Indonesia was recruited. Ages ranged from 18 to 55 (M=29.3, SD=8.7). All participants completed the full GATB administration under standardized conditions.

3.2 Instrumentation

The GATB consists of 214 items distributed across the eight scales. Scores are expressed as standard scores (mean=100, SD=15). For CFA, each item was treated as an indicator of its respective latent factor.

33 Data Preparation

Data screening involved:

• Missingvalue analysis (<2% missing, imputed using expectationmaximization).
• Univariate normality checks (skewness <|2|, kurtosis <|7|).
• Outlier detection via Mahalanobis distance (excluded 18 cases).

3.4 Statistical Analysis

CFA was conducted using **Mplus 8.6** and **R (lavaan package)**. Fit indices evaluated included:

• Chisquare () and /df
• Comparative Fit Index (CFI)
• TuckerLewis Index (TLI)
• Root Mean Square Error of Approximation (RMSEA) with 90% CI
• Standardized Root Mean Square Residual (SRMR)

Acceptable thresholds were: CFI0.95, TLI0.95, RMSEA0.06, SRMR0.08 (Hu & Bentler, 1999).

4. Results

4.1 Baseline EightFactor Model

The hypothesized eightfactor model yielded the following fit statistics:

• (2084)=3125.4, p<.001
• /df=1.50
• CFI=0.966
• TLI=0.962
• RMSEA=0.039 (90%CI=0.0360.042)
• SRMR=0.045

All primary loadings were significant (p<.001) and ranged from .62 to .89, supporting the intended itemfactor assignments. Interfactor correlations varied from .31 (Motor CoordinationFinger Dexterity) to .71 (General LearningVerbal Aptitude), reflecting theoretically plausible relationships.

4.2 Alternative Models

Two competing models were tested:

1. Higherorder model: Eight firstorder factors load onto a single General Ability factor. Fit: CFI=0.954, RMSEA=0.045.
2. Bifactor model: General factor plus orthogonal group factors. Fit: CFI=0.959, RMSEA=0.043.

Compared with the baseline model, both alternatives showed slightly lower CFI/TLI and higher RMSEA, indicating that the original eightfactor structure provides the most parsimonious representation of the data.

4.3 Reliability

Composite reliability () for each factor exceeded .85, confirming strong internal consistency. The latent general factor in the higherorder model accounted for 38% of the total variance, supporting the view that the GATB captures both specific abilities and a broader cognitive component.

5. Discussion

The CFA results affirm that the GATB maintains a robust eightfactor structure when administered to a contemporary Indonesian workforce. The fit indices surpass conventional criteria, and factor loadings are comparable to those reported in earlier U.S. validation studies (e.g., Guttman, 1975; McGrew & Bouchard, 1999). This suggests that cultural and linguistic adaptations have not compromised the tests underlying construct architecture.

The modest to high interfactor correlations echo the hierarchical nature of intelligence models. While the presence of a substantial general factor is evident, the distinctiveness of the eight scales justifies their continued use for targeted occupational selection and training.

6. Practical Implications

• Selection: Employers can rely on individual GATB scales to match specific job requirements (e.g., Spatial Aptitude for engineering roles).
• Career counseling: The factor profile provides a nuanced picture of a candidates strengths and developmental needs.
• Test development: Minor item revisions may further reduce crossloadings observed between Motor Coordination and Finger Dexterity.

7. Limitations and Future Research

Although the sample size was adequate for CFA, the study focused on applicants from urban centers; rural populations remain underrepresented. Future work should examine measurement invariance across gender, age groups, and educational levels to ensure fairness. Longitudinal designs could also explore predictive validity of each factor for job performance.

8. Conclusion

Using Confirmatory Factor Analysis, the present study provides strong empirical support for the construct validity of the General Aptitude Test Battery. The eightfactor model fits the data well, outperforms alternative hierarchical specifications, and exhibits high reliability. These findings reinforce the GATBs utility as a multidimensional assessment tool for contemporary vocational contexts.

References: Hu, L., & Bentler, P. (1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. *Structural Equation Modeling*, 6(1), 155. Guttman, L. (1975). *General Aptitude Test Battery: Manual*. U.S. Employment Service. McGrew, K., & Bouchard, T. J. (1999). Handbook of intelligence. Cambridge University Press.