Children’s self-estimates of IQ become more accurate with age—but only to a point

Children under 10 are not very accurate at judging their own intelligence, according to a large-scale study published in Intelligence. Researchers in Estonia found that younger children’s self-assessments of how smart they are often have little to do with their actual performance on standardized IQ tests. But around the age of 10, this begins to change. At that point, children’s self-perceived intelligence starts to reflect their measured cognitive abilities more reliably—though the two never fully align.

Intelligence refers to a general ability to think abstractly, solve problems, learn quickly, and adapt to new situations. Psychologists often measure intelligence using standardized tests, such as the Raven Standard Progressive Matrices, which assess pattern recognition and reasoning ability without relying on language. While these tests are widely accepted among researchers, people’s everyday ideas about intelligence tend to be much broader. Many individuals—especially children—may associate being “smart” not only with academic performance but also with traits like being well-behaved, popular, or confident.

The mismatch between scientific and lay definitions of intelligence has raised questions about whether simply asking people how intelligent they think they are can serve as a useful substitute for formal testing. Prior research has found that people’s self-reported intelligence usually has only a weak relationship with their actual IQ scores. This gap has limited the usefulness of self-reported intelligence in psychological research and education, but little was known about how this relationship develops over the course of childhood.

The goal of the current study was to pinpoint the age at which children begin to form a meaningful understanding of their own intelligence. The researchers, Jüri Allik and Helle Pullmann, wanted to know when self-reported intelligence starts to reflect measured cognitive ability and how stable these self-assessments are over time. The work was part of a broader effort to understand how intelligence, personality, and academic performance are linked during childhood and adolescence.

The study also builds on a long tradition of intelligence research in Estonia. In the early 20th century, Estonian school principal Juhan Tork conducted pioneering work on children’s cognitive ability that was later suppressed during Soviet occupation. After Estonia regained independence, researchers sought to revisit and expand upon these early studies using modern methods.

Free daily newsletter

“In 1939, Tork defended his doctoral thesis at the University of Tartu, in which he studied the intelligence of Estonian schoolchildren. After the occupation of Estonia by the Soviet Union, this work, which was far ahead of its time, was one of the first to be banned and physically burned by the new authorities,” explained Allik, a professor of experimental psychology at the University of Tartu.

“Tork’s work was commissioned by the Estonian Ministry of Education, as it was believed that the average intelligence level of a country with a very large number of peasant children was lower than that of developed countries. Tork’s work showed that the intelligence of Estonian children was no lower than that of the United States or Britain. When Estonia regained independence in 1991, there was suspicion that the deportation of the smartest part of the population to Siberia and emigration abroad had lowered the average mental level of the Estonian population.

“Helle Pullman’s doctoral thesis, which she defended in 2005, showed that the intelligence of Estonian children is basically at the same level as, for example, children in Iceland and the United Kingdom. In addition to the intelligence of Estonian schoolchildren, we also studied personality and the children’s self-perceived intelligence, as there are frequent cases in psychology where belief in one’s abilities is as important as the abilities themselves.”

The study used data from thousands of Estonian schoolchildren between the ages of 7 and 18, drawn from two related samples. The first group included 2,712 adolescents in Grades 6, 8, 10, and 12 who were assessed in 2001, with a follow-up conducted in 2003 that included 1,681 students from the same schools—some of whom had participated previously, while others were new. A second group consisted of 1,832 younger children in Grades 2, 3, and 4 from across Estonia, with ages ranging from 7 to 11.

Participants completed the Raven Standard Progressive Matrices, a nonverbal test of fluid intelligence that requires identifying patterns in visual matrices. Children also rated their own intelligence using age-appropriate self-report items. For students in Grades 1 through 4, a simplified three-point scale was used in response to the statement “I am very smart, and I understand everything immediately.” Older students (Grade 6 and above) used a 10-point scale to assess how their cognitive abilities compared to others, ranging from “Others are significantly more capable than me” to “I am significantly more capable than others.”

To enable comparison across different ages and formats, all responses were standardized within classrooms, and students were grouped into low, average, or high self-reported intelligence categories based on z-scores. In addition to the single-item question, younger children also answered a six-item scale assessing academic self-perception, which improved in reliability with age.

Self-esteem was also measured using the 10-item Rosenberg Self-Esteem Scale, which was administered to students in Grades 6 through 12. This allowed the researchers to examine the relationship between self-perceived intelligence, actual cognitive performance, and broader self-concept over time.

The results indicate that young children are not particularly good at evaluating their own intelligence. Among children aged 7 to 9, those who rated themselves as very smart actually scored lower on IQ tests than their peers with more modest self-assessments. This suggests that self-perceived intelligence during early childhood may reflect wishful thinking, confidence, or self-esteem rather than actual cognitive ability.

But starting around age 10, this pattern begins to shift. From that point onward, children’s self-reported intelligence begins to track more closely with their measured IQ scores. The correlation between the two measures becomes increasingly consistent with age, peaking at a correlation of about 0.41 in 11-year-olds. This pattern suggests that by age 10, children begin to develop the cognitive and social maturity needed to make more accurate judgments about their own abilities.

Importantly, this shift happens before the self-report format changes from the simpler version used in early grades to the more nuanced scale used in older students. This rules out the possibility that the change in response format is driving the improvement in accuracy.

“It is surprising that although we collected the data over 20 years ago, the question of at what age a child’s self-rated intelligence begins to align with their psychometrically measured intelligence had not been explored until our research,” Allik told PsyPost. “We have now established that starting from the age of 10, a correlation begins to emerge between subjective and objective measures of intelligence.”

However, the researchers also found that self-reported intelligence never aligns perfectly with measured IQ. Even among older students, the correlation remains modest. And somewhat surprisingly, by the final year of high school, the relationship between self-perceived and measured intelligence starts to weaken. Among 12th graders, the correlation dropped to just 0.12. The researchers suggest that as adolescents grow older, their self-assessments may become more influenced by broader self-concepts, including their social standing and emotional self-worth.

Supporting this idea, the researchers found that self-reported intelligence was strongly correlated with self-esteem at all ages. This link became especially pronounced in high school, where self-esteem was a much better predictor of perceived intelligence than actual IQ scores. By 12th grade, how smart students believed they were had more to do with how they felt about themselves in general than how they performed on cognitive tests.

“The main point is that a simple question about a person’s self-assessed intelligence cannot replace a verified and reliable IQ test that measures mental abilities,” Allik said. “In addition to the fact that one question is a very unreliable measure, the folk concept of intelligence is different from that of professional psychologists.”

The study provides a rich look at how children’s understanding of intelligence develops, but it has some limitations. The researchers focused on Estonian students, and while previous comparisons suggest their developmental trajectory is similar to children in other countries, cultural differences could still affect how children think about intelligence. Future research might explore how these patterns play out in other cultural settings or in children with different educational experiences.

Another open question is how children’s concepts of intelligence evolve beyond adolescence. While this study tracked students into late high school, it’s possible that their self-assessments change again in early adulthood, particularly as they gain more life experience and academic or occupational feedback.

“One of the goals of this project was to investigate how grades given in school depend on children’s mental abilities and personality traits,” Allik explained. “Since this is an important topic, ourarticle on it has been noticed and cited quite a few times. According to Google Scholar, 986 times to date. After matching our data with the birth register, we discovered that within the normal range of birth weight (≥ 2500 g), every 500 g increase in birth weight was associated with an approximate increase of 0.7 points in IQ scores measured in the early grades.

“Also, maternal smoking during pregnancy was accompanied by a 3.3-point deficit in children’s intellectual abilities. We also found that the later criminal behaviour of boys was associated with lower cognitive ability, grade point average, the lack of agreeableness and conscientiousness, as well as higher neuroticism.”

The study, “How accurately does self-reported intelligence reflect psychometrically measured IQ?“, was published June 21, 2025.