Is it Difficult to Measure Intelligence?

Many introductory psychology textbooks claim that measuring intelligence presents significant challenges (R. S. Feldman, 2015; Pastorino & Doyle-Portillo, 2016). This belief suggests that intelligence testing involves considerable uncertainty and that IQ score interpretations are questionable at best. Consequently, those who accept this premise often view intelligence research as unreliable.

In truth, these textbook authors are mistaken. Intelligence is remarkably straightforward to measure because any task requiring cognitive effort or judgment assesses intelligence (Jensen, 1998). Since all such tasks correlate positively and measure g, one simply needs to administer cognitive tasks to estimate an examinee's intelligence level.

Intelligence measurement is so straightforward that test creators have unintentionally developed intelligence tests while attempting to assess other traits. The Cognitive Assessment System and Cognitive Abilities Measurement battery were both designed to evaluate cognitive processes rather than g, yet they measure g regardless (Keith et al., 2001; Stauffer et al., 1996). Similarly, the Defining Issues Test, intended to assess moral reasoning, correlates with verbal intelligence measures (Sanders, Lubinski, & Benbow, 1995).

This phenomenon extends beyond psychology. The National Adult Literacy Survey, created to measure reading comprehension, produces scores that correlate with life outcomes identically to intelligence test scores (Gottfredson, 2004). The Test of Functional Health Literacy of Adults, which assesses comprehension of healthcare texts, correlates substantially with traditional intelligence tests (r = .53 to .74) and functions exactly like an intelligence scale (Apolinario et al., 2014; Gottfredson, 2004).

The reason diverse tests can measure intelligence relates to what determines test content. Mental abilities required by a test, rather than its surface content, determine what the test measures (Gottfredson & Saklofske, 2009). Since intelligence is a general ability, numerous tasks require examinees to utilize their intelligence, allowing varied question formats to assess intelligence even when tests appear dissimilar.

These findings support Spearman's concept of the indifference of the indicator: any task requiring cognitive work measures g, regardless of appearance (Spearman, 1927). Poor understanding of this principle leads to misinterpretations. For instance, law school bar exam passing rates largely reflect enrolled students' intelligence levels rather than exclusively indicating educational quality. Similarly, K-12 school rankings based on achievement tests primarily measure students' g levels rather than solely reflecting teaching quality.

Traditional intelligence tests require 90-120 minutes to administer, creating an impression that intelligence measurement demands extensive testing. However, this is unnecessary. A single test item produces reliability of approximately .25, but combining items increases reliability substantially. Seven items achieve .70 reliability (acceptable for research), twelve items reach .80, and twenty-seven items attain .90 reliability (Lubinski, 2004).

While any cognitive task measures g to some degree, different tasks vary in effectiveness. Matrix reasoning and vocabulary tasks excel at measuring intelligence, whereas maze tests prove inferior (Jensen, 1980b, 1985). Nonetheless, intelligence measurement remains relatively straightforward compared to other psychological traits. Many common tests—including school accountability assessments, licensing examinations, and college admissions tests—measure g, though not all equally well.



From Chapter 7 of "In the Know: Debunking 35 Myths About Human Intelligence" by Dr. Russell Warne (2020)