Comprehensive Adult Intelligence Test

Welcome to the Comprehensive Adult Intelligence Test (CAIT) by u/EqusB.

The aim of project CAIT is to provide a free resource for estimating a full scale IQ (FSIQ) score for adults interested in cognitive testing. The CAIT is designed for individuals over the age of 16.

The CAIT is inspired by and formatted to resemble the Wechsler Adult Intelligence Scale (WAIS). Overall, the goal is to provide a similar but streamlined experience that can estimate FSIQ in a timely manner.

Estimating FSIQ across a diverse range of test batteries is the best way to estimate the g factor, or general factor of intelligence. Most online IQ tests concentrate on measuring a single factor. The CAIT is designed to provide a more comprehensive view of general intelligence by testing multiple factors.

The CAIT is an open-ended project that will be updated over time to make it more accurate and more comprehensive. If you have taken the WAIS professionally and decide to take the CAIT, please submit your scores to help improve this test.

No outside resources are allowed at any point during this test. Pencils, paper, calculator, Google, etc., are NOT ALLOWED. The use of any outside resources invalidates your test result.

You can view your subtest and index scores in the dashboard after taking the relevant tests. You can also access the score calculator at the following link: CAIT Score Calculator.

Note: The CAIT is not a substitute for a professional IQ test. Scores obtained using the CAIT, if taken correctly, are designed to give an accurate estimation of FSIQ. However, the CAIT is not a diagnostic tool and cannot be used in any capacity other than as an informative tool. Individuals seeking a diagnosis or comprehensive psychological report should be tested by a professional.

The CAIT is designed for native English speakers and adults over the age of 16. Non-native English speakers cannot receive an accurate Verbal Comprehension Index. Individuals under the age of 16 cannot receive an accurate IQ score.

Verbal Comprehension Index

Perceptual Reasoning Index

Visual Spatial Index

Block Design

Cognitive Proficiency Index

Validity

Sample

Out of the total 1692 attempts, the sample had to be filtered according to various criteria to ensure that the influence of invalid factors would be minimized. Only the following attempts were considered: first attempts, both VCI and PRI attempted, non-floor attempts, attempts from native English-speaking countries (US, CA, UK, IE, AU, NZ). After narrowing down this sample, we are left with 449 valid attempts.

Intercorrelations

	V	GK	VP	FW	BD
V	1.000	0.672	0.305	0.283	0.212
GK	0.672	1.000	0.320	0.393	0.212
VP	0.305	0.320	1.000	0.649	0.623
FW	0.283	0.393	0.649	1.000	0.501
BD	0.212	0.212	0.623	0.501	1.000

CAIT Bifactor Model

          lavaan 0.6.15 ended normally after 51 iterations
          
            Estimator                                         ML
            Optimization method                           NLMINB
            Number of model parameters                        17
          
            Number of observations                           449
          
          Model Test User Model:
            Test statistic                                30.331
            Degrees of freedom                                 3
            P-value (Chi-square)                           0.000
          
          Model Test Baseline Model:
            Test statistic                               836.403
            Degrees of freedom                                10
            P-value                                        0.000
          
          User Model versus Baseline Model:
            Comparative Fit Index (CFI)                    0.967
            Tucker-Lewis Index (TLI)                       0.890
          
          Loglikelihood and Information Criteria:
            Loglikelihood user model (H0)              -5507.965
            Loglikelihood unrestricted model (H1)      -5492.800
          
            Akaike (AIC)                               11049.931
            Bayesian (BIC)                             11119.750
            Sample-size adjusted Bayesian (SABIC)      11065.799
          
          Root Mean Square Error of Approximation:
            RMSEA                                          0.142
            90 Percent confidence interval - lower         0.099
            90 Percent confidence interval - upper         0.190
            P-value H_0: RMSEA <= 0.050                    0.000
            P-value H_0: RMSEA >= 0.080                    0.990
          
          Standardized Root Mean Square Residual:
            SRMR                                           0.047
          
          Parameter Estimates:
            Standard errors                             Standard
            Information                                 Expected
            Information saturated (h1) model          Structured

        

g-Loadings

Measure	Mean	SD	Reliability	g-Loading *
GAI	124.79	15.98	0.923	0.852
VCI	125.06	15.63	0.904	0.804
PRI	119.76	17.54	0.890	0.689
VSI	121.66	17.04	0.879	0.636
V (SS)	14.14	2.66 †	0.795	0.825
GK (SS)	15.12	3.65	0.870	0.704
VP (SS)	13.93	3.46	0.826	0.648
FW (SS)	13.38	3.68	0.816	0.620
BD (SS)	14.07	3.52	0.835	0.504

* This sample has a mean of 124.79, much higher than the average person. In order to ensure an accurate measure of this test's g-loading, it must be adjusted for SLODR (Spearman's law of diminishing returns). For example, while the GAI g-loading was calculated at 0.716 for this sample, the corrected g-loading returns 0.852.
† Due to the standard deviation of Vocabulary being below 3, it was corrected for range restriction.

Conclusion

Looking at the g-loadings of various subtests, some things stand out. Vocabulary being the highest subtest makes sense, being based on the already well-established SAT-V.

Let's compare the rest of the subtests to the WAIS-IV and WISC-V:

	CAIT	WAIS	WISC
IN (GK)	0.704	0.648	0.721
VP	0.648	0.679	0.648
FW	0.620	0.715	0.530
BD	0.504	0.687	0.639
Average	0.619	0.682	0.635

As shown, the CAIT seems to stand with the professional counterparts it was designed to estimate. Why CAIT's Block Design is so low is up to speculation, but it may be due to format differences. The CAIT BD format is based on the multiple-choice version of WISC BD for the physically-impaired that does not require blocks. However, the WISC and WAIS both make use of physical blocks.

Disclaimer

The sample that was used to calculate the g-loadings is of inferior quality compared to the WISC and WAIS. Unfortunately, due to the nature of online testing, it is difficult to control for all external factors that may have affected this sample, such as cheating, distractions, interruptions, etc. Nonetheless, this doesn't invalidate the g-loadings calculated above.

Test Structure

Verbal Comprehension Index

Core Subtests:

Vocabulary (10 min)
General Knowledge (10 min)

Perceptual Reasoning Index

Core Subtests:

Visual Puzzles (13 min)
Figure Weights (12 min)

Visual Spatial Index

Supplemental Subtests:

Visual Puzzles (13 min)
Block Design (9 min)

Cognitive Proficiency Index

Core Subtests:

Digit Span (5 min)
Symbol Search (2 min)