Research Article: Instability in longitudinal childhood IQ scores of Guatemalan high SES individuals born between 1941-1953

Date Published: April 25, 2019

Publisher: Public Library of Science

Author(s): Liina Mansukoski, Eef Hogervorst, Luis Fúrlan, J. Andres Galvez-Sobral, Katherine Brooke-Wavell, Barry Bogin, Agustin Martínez Molina.


Childhood IQ has been used to predict later life outcomes across disciplines in epidemiology, education, and psychology. Most often only a single childhood IQ test is available or is used for these purposes in the belief that IQ is stable across the life course. The primary aim of this study was to examine the longitudinal stability of individuals’ IQ test scores derived from school-age tests. The secondary aim was to investigate the association of the pre-adult scores with later life intelligence scores. The longitudinal pre-adult IQ scores of 42 high socioeconomic status Guatemalans born 1941–1953 were analysed and showed low stability of longitudinal test scores. Fluctuations of >1SD were found for 59.5% of the sample. The same participants, aged 64–76 years, were re-assessed and average pre-adult IQ explained 12% of variance in the older age intelligence score. The reasons behind the longitudinal instability in test scores reported in this study remains unknown but the results suggest single point measurements of intelligence before adulthood should be regarded with some caution.

Partial Text

Previous studies have associated childhood general mental ability, often measured as Intelligence Quotient (IQ) scores, with various later life health outcomes, including risk of dementia and mortality[1–7]. Studies connecting childhood or adolescent mental ability/IQ to later life outcomes assume that there is relative stability in cognitive performance test scores over the developmental period, as often only a single, un-replicated IQ measurement is available or selected to represent early life general mental ability [3,6,8,9]. A systematic review identified a total of nine studies where IQ scores from childhood or early adulthood were investigated longitudinally in relation to later life mortality[1]. None of the studies had repeated childhood IQ measurements; instead, tests were administered only at a single time point, commonly around 11 or 12 years of age [1]. Much literature shows evidence for longitudinal stability of IQ across the life course [10,11], and individuals appear to follow their own ‘tracks’ in terms of broad cognitive ability and are thought to maintain their position relative to others in IQ during the age-related decline–individuals who scored highly compared to peers early in life seem to do so later on as well [12]. Other work has investigated stability of ‘absolute’ IQ scores–that is, repeated tests taken by an individual and reported high levels of fluctuation in scores, particularly during pre-adult years [13–16]. Possible reasons for instability are adverse life events, larger than expected deviations of individual developmental level at the time of the testing and differences between the testing instruments [13–16]. Due to these fluctuations it is important to investigate whether a single timepoint measure of IQ in pre-adult years can be considered representative of childhood mental ability, and how this relates to old age scores.

Descriptive data on each childhood IQ test, mean scores and ages of participants are presented in Table 1. Neither age at testing nor year of measurement were significantly related to pre-adult IQ test scores (age: rho = 0.12, p = 0.15; for year of measurement: rho = 0.1, p = 0.22). There were some differences in score distributions between the tests used (Fig 2), and most between-test correlations were low (Table 3). The range in mean outcome scores between tests was up to 8.4 points. The Pintner Durost A test significantly differed in mean from the Otis Intermedio (mean difference = -8.45, p = 0.025). All other between test comparisons were non-significant.

This study examined whether 1) IQ, as derived from age-appropriate standardized tests, remains relatively stable across childhood and adolescence–i.e., within subject variance and range are smaller than between subject variance and range; and 2) older age cognitive function or IQ may be predicted from pre-adult IQ. We found large intraindividual variability in IQ test scores in this sample of Guatemalan high SES individuals tested over their pre-adult years, and that the within individual distribution of IQ test scores varied greatly between participants. With regard to the second hypothesis, we found that the pre-adult IQ test scores were only modestly related to older age fluid intelligence test scores but not to WAT.