Let’s get this post started with a cognitive science themed trivia question. What’s the name of the intelligence component comprising declarative and procedural knowledge learnt by an individual throughout their life span?

The answer is: crystallized intelligence.

Information such as the number of different words a person knows (vocabulary size) and the ability to retrieve culturally relevant facts and data from memory (general knowledge) are parts of crystallized intelligence.

When one thinks of intelligence, the crystallised type is not necessarily what comes to mind. Instead, the other component, fluid intelligence probably does. Fluid intelligence refers to cognitive abilities that don’t depend on previously acquired knowledge such as pattern recognition, reasoning, and abstraction.

The distinction between these two components has a long story, and, while not fully independent from each other, they tend to relate to other variables in dissimilar ways. For example, fluid intelligence is a better predictor of academic success early in life, while crystallized intelligence predicts academic and professional success later.

Unlike tests of fluid intelligence, which, by definition, attempt to use questions that rely as little as possible on acquired knowledge (and hence often comprise a relatively small number of abstract items), reliably testing crystallized intelligence requires using a huge number of questions. After all, how many facts and information can a person know about domains as diverse as history, physics, economy, art, and countless other domains that one could find in trivia contests?

Francisco Buades-Sitjar, Roger Boada, Marc Guasch, Pilar Ferré,  José Antonio Hinojosa and Jon Andoni Duñabeitia, present results of their megastudy (~50,000 participants who responded to 1,000+ general knowledge questions) in their paper, “The predictors of general knowledge: Data from a Spanish megastudy,” published in the Psychonomic Society journal Behavior & Research Methods.

Data were collected online between July 13, 2020, and October 2, 2020. Initially, recruitment occurred via social media outlets, followed by announcements in traditional media (press, radio programs, etc.). The promotions corresponded to the peaks in the number of people who played the game, shown in the graph below. More than 85,000 response sets were collected.

The sample was comprised of 59% men, 40% women, and the rest did not identify as either. The average age was 42 years. The distributions of these variables are summarized in the figure below.

The study required only basic technological skills to participate, such as using a computer or smartphone (which 98% of the Spanish population claims to have). In line with this assumption, the range of years of education of the sample was wide, from 0 to 30, with a high number of responses indicating 19 years of education (corresponding to the years required for a university degree in Spain). Socioeconomic status was measured using the MacArthur scale of subjective social status. The scale is a single-item instrument to measure a person’s perceived rank relative to others in their group, with higher numbers indicating a higher perceived socioeconomic status. The following figure summarizes the years of education and perceived socioeconomic status of the sample.

Each participant responded to 60 multiple-choice questions from a pool of 1270 questions from more than 30 fields of knowledge, including

• To which art movement did Frida Kahlo belong?
• Which singer is known for the single “My Heart Will Go On”?
• What does the acronym “RAM” mean in computing?
• How many chambers comprise the stomach of a cow?

After removing participants who played more than once and people with nationalities other than Spanish, the authors calculated the correct response rates (i.e., percentage of correct) from about 50,000 participants. The average correct response rate was 60%, and the distribution is shown in the plot below.

Buades-Sitjar and colleagues analyzed the effects of the sociodemographic variables they collected (years of education, socioeconomic status, age, and gender) on the correct response rates. Senior author, Duñabeitia succinctly summarised the results,

After a series of statistical analyzes, it was found that the socioeconomic level, the educational level, age and the gender of the people were determining factors of their cultural level.

Indeed, all the socioeconomic variables analyzed significantly predicted correct response rates, with years of study and socioeconomic status showing tenuous linear relationships with rates, age acting as a positive predictor before the age of 50 years, but decreasing after 50 years, and people who identified as men or neither men nor women showing higher correct response rates than those who identified as women. These relationships are summarized in the following figure.

The inflection point at around 50 years of age (in the top right plot above) contrasts with results of other components of crystallized intelligence, such as vocabulary size, for which it has previously shown a positive relationship throughout the lifespan. Years of education were a reliable predictor of general knowledge, much in line with previous studies, and in line with intuitions.

Socioeconomic status was a significant, but weak predictor of correct response rates, which contrasts with results from previous studies that have found that it reliably predicts IQ (at least in younger populations). In fact, the authors suggest that its effect was so weak that it can be safely dismissed as a predictor altogether.

Maybe the most surprising results are those showing men and people who did not identify as men or women had higher correct response rates than women. While the results aligned with previous studies on crystallized intelligence, the authors caution against interpreting the exact nature of these differences because of the exploratory nature of the study, and because previous studies suggest that gender differences greatly vary with age and across countries and cultures, which calls for more detailed analyses in upcoming studies if one is to properly understand these relationships.

The data are available and if you feel comfortable enough in understanding Spanish, the test is available online for you to try.

Featured Psychonomic Society article:

Buades-Sitjar, F., Boada, R., Guasch, M. Ferré, P., Hinojosa, J.A., & Duñabeitia, J.A. (2021). The predictors of general knowledge: Data from a Spanish megastudy. Behavior Research Methods https://doi.org/10.3758/s13428-021-01669-4