Psychometric Properties and Measurement Invariance of the Generalized Tracking Questionnaire‑Children (GTQ‑C)

Context and objectives

The Generalized Tracking Questionnaire-Children (GTQ-C) is an adaptation of the adult GTQ designed to assess generalized tracking—a theoretically relevant construct—in child and adolescent populations. Although the original GTQ has demonstrated solid psychometric properties in adults, its adaptation for younger populations required rigorous psychometric validation. The authors aimed to examine the psychometric properties of the GTQ-C and establish its measurement invariance across gender and age groups in a sample of Colombian schoolchildren.

Method

Participants: A total of 730 Colombian children and adolescents (52.6% female) aged 7–17 years (M = 12.29, SD = 2.29) were recruited. The sample included students from 2nd to 11th grade, predominantly from private educational institutions (75.1%) with a smaller proportion from public schools (24.9%). Participants were randomly divided into two approximately equal subsamples (n ≈ 357 for EFA; n ≈ 373 for CFA) for cross-validation purposes.

Instrument: The GTQ-C consists of 12 items assessed on a 5-point Likert scale (1 = never true to 5 = always true). It was adapted from the adult GTQ (11 original items) through vocabulary modifications to suit children's comprehension levels and the addition of one extra item to enhance construct coverage.

Analysis:

For exploratory factor analysis (EFA), robust diagonally weighted least squares (Robust DWLS) with Direct Oblimin rotation and polychoric correlations were employed (Factor 10.9.02). Assumptions regarding sample adequacy were evaluated using Bartlett's test and the KMO index. Parallel analysis (PA) was used to determine the optimal number of factors.

For confirmatory factor analysis (CFA), LISREL 8.71 software was used with Robust DWLS estimation and polychoric correlations. Multiple fit indices were evaluated: Satorra-Bentler χ² (S-Bχ²), RMSEA with 90% confidence intervals, CFI, NNFI, and SRMR. Measurement invariance was examined through multigroup comparisons, assessing metric and scalar invariance across gender (male vs. female) and age groups (7–12 years vs. 13–17 years).

Internal consistency was evaluated using Cronbach's α and McDonald's ω, with 95% confidence intervals. Additionally, corrected item-total correlations were calculated. Finally, a two-way analysis of variance (ANOVA) was conducted to examine main effects and interactions as a function of gender and age group.

Results

Exploratory Factor Analysis: Bartlett's test was significant (χ²(66) = 605.4, p < .001) and KMO was .79, indicating acceptable sample adequacy. Parallel analysis recommended a single-factor solution explaining 35.7% of variance. Factor loadings ranged from .38 (item 5) to .59 (item 2). Unidimensionality indices were favorable: UniCo = .96 and MIREAL = .21, with ECV = .80, confirming the essential nature of the unifactorial structure.

Confirmatory Factor Analysis: The initial one-factor model showed acceptable fit (S-Bχ²(54) = 130.03, p < .05; RMSEA = 0.062, 90% CI [0.048, 0.075]; CFI = 0.95; NNFI = 0.94; SRMR = 0.063). After allowing error terms for items 4–7 to correlate, fit improved substantially (RMSEA = 0.053; NNFI = 0.96; CFI = 0.97; SRMR = 0.057), meeting stringent fit criteria.

Measurement Invariance:

Across gender: The model demonstrated metric invariance (ΔRMSEA = -.0003, ΔCFI = -.005, ΔNNFI = .000) and scalar invariance (ΔRMSEA = -.0002, ΔCFI = -.004, ΔNNFI = .001), indicating that the factor structure and item magnitudes are equivalent between boys and girls.

Across age groups: Metric invariance (ΔRMSEA = .0014, ΔCFI = -.002, ΔNNFI = .002) and scalar invariance (ΔRMSEA = .0009, ΔCFI = -.002, ΔNNFI = .002) were supported, confirming that the instrument functions equivalently in the 7–12 year age range and the 13–17 year age range.

Internal Consistency: The total sample exhibited adequate internal consistency (α = .76, 95% CI [.73, .78]; ω = .76, 95% CI [.73, .78]). This pattern was replicated in both cross-validation subsamples: sample 1 (α = .75, ω = .75) and sample 2 (α = .76, ω = .76).

Corrected Item-Total Correlations: Ranged from .33 (item 3) to .45 (item 7), indicating consistent item contributions to the overall construct.

Analysis of Variance: A significant main effect of age group was detected (F(1) = 11.88, p = .001, η² = .016), with younger children (7–12 years) obtaining higher scores than adolescents (13–17 years). Gender produced no significant main effect (F(1) = 0.58, p = .44), and no significant interaction between gender and age group was observed (F(1) = 1.27, p = .26).

Descriptive Statistics: Mean scores by group: Boys 7–12 years (M = 44.39, SD = 7.33); Boys 13–17 years (M = 41.95, SD = 6.62); Girls 7–12 years (M = 43.39, SD = 7.86); Girls 13–17 years (M = 42.15, SD = 6.94). The overall mean was 42.91 (SD = 7.24).

Discussion and conclusions

The findings provide strong evidence that the GTQ-C possesses adequate psychometric properties for use in Colombian child and adolescent populations. The unidimensional structure emerging from both exploratory and confirmatory analyses is consistent with the theoretical conceptualization of generalized tracking as a unidimensional construct. The ECV value (.80) reinforces this interpretation, indicating that most of the common variance is captured by the general factor.

The demonstration of measurement invariance across gender and age is particularly important because it validates the assumption that the GTQ-C assesses the same construct equivalently regardless of these demographic characteristics. This allows for valid comparisons between groups without confounding score differences with measurement artifacts. The pattern of polychoric correlations and improved fit when allowing error correlations among certain items reflects the complex structure of responses on ordinal Likert scales, justifying the use of robust estimation methods.

The significant age-related difference—with younger children showing higher levels of generalized tracking than adolescents—is conceptually coherent and suggests that this construct may naturally vary with development. This pattern could reflect changes in tracking behavior linked to cognitive and socioemotional maturation processes.

Significance and contribution

This study contributes meaningfully to the psychometric literature in several ways:

1. Transcultural adaptation and validation: Represents the first systematic validation of the GTQ-C in a Spanish-speaking context, specifically in Colombia, extending the instrument's applicability beyond English-speaking contexts.

2. Evidence of invariance: Demonstration of metric and scalar invariance establishes the comparative validity of the instrument across relevant demographic groups, facilitating its use in clinical and epidemiological research.

3. Rigorous methodology: The use of cross-validation, robust statistical methods (DWLS, polychoric correlations), and multiple fit indices exemplifies contemporary standards in psychometric evaluation.

4. Measurement of tracking in minors: Provides a valid and reliable tool for assessing a construct relevant to child clinical psychology, with implications for research and clinical practice.

Verification Checklist

Element	Present	Notes
Adequate sample size	✓	N = 730; divided into two approximately equal subsamples
Detailed participant description	✓	Age, gender, educational background specified
Clear instrument description	✓	12 items, 1–5 scale, adaptation process explained
Explicit statistical method	✓	EFA (Factor 10.9.02, Robust DWLS), CFA (LISREL 8.71)
Multiple fit indices reported	✓	χ², RMSEA, CFI, NNFI, SRMR reported
Cross-validation	✓	Random division into two samples
Measurement invariance analysis	✓	Gender and age; metric and scalar
Internal consistency reported	✓	α and ω with 95% CI
Item-total correlations	✓	Range .33 to .45
Construct validity results	✓	ECV, UniCo, MIREAL, convergent EFA/CFA
Limitations discussed	Partially	Geographic coverage limitations implicit
Theoretical interpretation	✓	Connection to development and psychological processes