Reading and IQ: what the twin study, the meta-analyses and the long follow ups show

0 Quick Answer

Reading ability in childhood predicts later intelligence even between identical twins raised in the same home, print exposure explains a growing share of language ability from preschool to university, and the effects are real, modest and concentrated on the verbal side of what an IQ test measures. Ritchie, Bates and Plomin tested 1,890 twin pairs from the Twins Early Development Study at ages 7, 9, 10, 12 and 16 and reported in Child Development in 2015 that, within pairs of monozygotic twins, the twin with the better reading ability at an earlier age went on to score higher on intelligence measures at later ages. Because identical twins share their genes and their family environment, the design removes both as explanations, and the associations were not explained by differences in how much the twins read and were not restricted to verbal measures. It is the strongest evidence in the literature that something about reading, rather than something about the kind of child who reads, contributes to later ability.

The correlational literature points the same way and shows the size of the relationship. Mol and Bus's meta-analysis of print exposure across 99 studies and 7,669 participants found moderate to strong correlations between leisure reading and comprehension, technical reading and spelling at every age, with print exposure explaining 12 percent of the variance in oral language in preschool and kindergarten, 13 percent in primary school, 19 percent in middle school, 30 percent in high school and 34 percent at university. Cunningham and Stanovich's ten year follow up found that reading ability in first grade predicted how much a person read in eleventh grade, and that reading volume predicted vocabulary and general knowledge beyond what ability alone explained. The mechanism most of this literature proposes is the one Stanovich named the Matthew effect in 1986: children who read well read more, and reading more builds the vocabulary and knowledge that verbal tests reward.

Two limits belong in the answer. Reading difficulty is not low intelligence, and Tanaka and colleagues' imaging study of 131 children found the same brain signature of dyslexia in poor readers with high and with low IQ scores. And the most publicized claim in the field, that reading a passage of literary fiction improves theory of mind, did not survive a large replication attempt. What survives is narrower and better founded: reading builds crystallized ability, learning to read changes how the brain processes print and speech, and the effects on nonverbal reasoning are real in the twin data but small.

1 What Reading Could Change, and What a Test Measures

An IQ test measures several abilities, reading plausibly affects some of them directly and others hardly at all, and the question whether reading raises IQ has a different answer for each. A Full Scale score summarizes verbal knowledge, fluid reasoning, quantitative reasoning, visual spatial ability, working memory and processing speed, and the page on the six cognitive domains sets out what each index samples. The verbal domain is where reading operates most directly. In ACIS, the Vocabulary subtest asks for typed definitions and carries a reliability of .950 and a general factor loading of .732; Antonyms and Synonyms sample the precision of stored word meanings; Information asks general knowledge questions ordinarily acquired through schooling and everyday exposure, and Similarities asks for the relation between two concepts. The Verbal Comprehension Index built from them has a composite reliability of .9745 and a general factor loading of .830, and it correlates .801 with the Fluid Reasoning Index in the ACIS technical sample, which is why a person who reads widely tends to score well on the whole battery and not only on words.

The page on fluid versus crystallized intelligence explains the distinction the reading literature turns on. Crystallized ability is acquired knowledge and skill, and vocabulary is its purest indicator; fluid ability is reasoning with novel material, which a matrices test measures. Reading feeds the first directly, one word and one fact at a time, and the meta-analytic evidence below shows how strongly. Whether it feeds the second is the open question, and the twin study is important precisely because it found effects on nonverbal measures that a pure knowledge account would not predict.

The page on what IQ measures adds the caution that a test score is a position within a norm sample on a particular day. A person who reads a great deal will typically hold a larger vocabulary and more general knowledge than a person of the same reasoning ability who does not, and will therefore score higher on the verbal indices and, through them, on the Full Scale score. That is not an artefact; vocabulary is part of intelligence as every major battery defines it. But it means the sentence "reading raises IQ" is partly a statement about what IQ includes, and the honest version of it names the domain.

2 The Twin Design: Ritchie, Bates and Plomin, 2015

The identical twin differences design is the closest thing to an experiment that a study of reading and intelligence can have without assigning children to read, and it found that the twin who read better earlier was more intelligent later. Monozygotic twins share all of their genes and, when raised together, their family, school and neighbourhood. Any difference between them in reading ability at seven therefore cannot be caused by genes or by the shared environment; it must come from something specific to one twin, a teacher, a friend, an illness, a book. The cross-lagged design then asks whether that within pair difference in reading at one age predicts a within pair difference in intelligence at the next age, over and above the intelligence difference that already existed. Ritchie, Bates and Plomin ran that analysis on 1,890 pairs at 7, 9, 10, 12 and 16 with reading tests and both verbal and nonverbal intelligence measures at each wave.

The results showed several such associations, in which the earlier reading difference predicted the later intelligence difference. Two features of the result matter more than its existence. The associations were not explained by differences in reading exposure, the amount the twins reported reading, which argues against the simplest volume account. And they were not restricted to verbal cognitive domains: the twin who read better earlier also scored higher on nonverbal measures later, which is the finding a knowledge only account cannot easily accommodate. The authors describe the influence as driven by the nonshared environment and raise, without settling, the question of mechanism, whether reading trains abstract thinking, sustained attention, or the habit of working through structured material.

The limits are the usual ones and the authors state them. The effects are modest in size, the design shows that reading ability differences precede intelligence differences but not what caused the reading differences, and reading was measured by tests rather than by observed practice. The page on what heritability actually means explains why a twin design can isolate environmental influence without identifying it, and the page on IQ and academic achievement sets out the larger correlational picture within which this causal thread sits. What the study establishes is that at least part of the association between reading and intelligence runs from reading to intelligence, and that the part is not confined to words.

3 Print Exposure Across Development: Mol and Bus, 2011

The correlational literature is large, consistent and shaped like a spiral, with the association between reading and language growing at every stage from preschool to university. Mol and Bus synthesized 99 studies with 7,669 participants in Psychological Bulletin in 2011, grouped by preschoolers and kindergartners, pupils in grades 1 to 12, and university students, using measures of print exposure such as recognition of author and title names that are less vulnerable to social desirability than self-reported hours. For every outcome in the domains of comprehension, technical reading and spelling, the correlations with print exposure were moderate to strong. The variance in oral language explained by print exposure rose from 12 percent in preschool and kindergarten to 13 percent in primary school, 19 percent in middle school, 30 percent in high school and 34 percent at university, which is the pattern of an accumulating advantage rather than a fixed one.

The authors read the pattern as an upward spiral of causality: children who are more proficient readers read more, and because they read more their comprehension and technical skills improve more with each year of schooling. That is Stanovich's Matthew effect, named in his 1986 paper after the parable in which those who have are given more, and it is the model that most of the field works from. The meta-analysis also found moderate associations between print exposure and academic achievement, and, in a finding the authors flagged as interesting, that poor readers also appeared to benefit from independent leisure reading, which argues against the view that the spiral only lifts those already at the top.

Oral language is not IQ, and the outcomes in the meta-analysis are reading and language measures rather than intelligence tests. The link to IQ runs through vocabulary and verbal comprehension, which are the language measures that every battery includes and which load heavily on the general factor, as the ACIS figures above show. The page on the vocabulary subtest describes what a definitions task requires, and the page on intelligence versus knowledge draws the line that the Mol and Bus results sit on: print exposure builds knowledge and skill, and a test that counts knowledge and skill as intelligence, which every battery does, will register the difference.

4 Reading Volume: The Arithmetic Behind the Spiral

The reason reading differences compound is arithmetic, and the classic measurement of it shows a gap of orders of magnitude between children at the top and the bottom of the reading time distribution. Anderson, Wilson and Fielding asked fifth graders to keep diaries of their time outside school and reported in Reading Research Quarterly in 1988 that the amount of book reading varied enormously and predicted reading achievement growth. A child at the 90th percentile of reading time read books for about 40 minutes a day, which the authors estimated at more than two million words a year; a child at the 50th percentile read for about thirteen minutes a day, around 600,000 words; a child at the 10th percentile read for well under two minutes a day, on the order of 50,000 words a year, by the authors' estimates. Over a school career the difference in words encountered is measured in tens of millions, and a share of the unfamiliar words met in text are learned incidentally, which is how vocabularies of tens of thousands of words are built without anyone teaching them.

That arithmetic is what makes vocabulary a good indicator of intelligence and an imperfect one. It is good because the rate at which a person learns words from context depends on reasoning, memory and existing knowledge, so that a vocabulary score summarizes a lifetime of cognitive work. It is imperfect because the raw material, the number of words met, depends on opportunity, and two people of the same ability who have read very different amounts will hold different vocabularies. The page on average IQ by education shows the size of the gradient that schooling, which is largely a reading regime, produces in test scores, and the page on the Flynn effect discusses whether rising exposure to print and abstract material across the twentieth century is part of why raw scores rose.

For a reader who wants the practical implication, it is the plain one: the verbal indices of any battery are sensitive to reading history, and a person who has read little and scores modestly on them has a score that reflects the reading as well as the ability. A person who wants to raise a vocabulary score has, in the reading volume literature, the only method that reliably works, and it takes years.

5 Ten Years Later: First Grade to Eleventh

The longest follow up in the literature started with first graders and found that how well they read at six predicted how much they read, and what they knew, at sixteen. Cunningham and Stanovich reported in Developmental Psychology in 1997 on children first tested in first grade and followed to eleventh grade. First grade reading ability was a strong predictor of eleventh grade reading comprehension, vocabulary and general knowledge, which is expected. The less expected finding was that first grade reading ability predicted eleventh grade print exposure even after eleventh grade reading comprehension was controlled, which means that the early start on reading shaped the habit independently of the skill it eventually produced. And print exposure in eleventh grade predicted vocabulary and general knowledge beyond what reading comprehension and general ability explained, which is the volume effect operating in adolescence.

The study is small by the standards of the meta-analyses and it is correlational, so it cannot separate the child who reads from the reading. Its value is the length of the window and the separation of ability from exposure at both ends. The pattern it describes, early ability setting up a habit that then feeds knowledge for a decade, is exactly the spiral that Mol and Bus later quantified across 99 studies and that the twin study then tested against genetic confounding. Three designs with different weaknesses point the same way, which is the ordinary standard for believing a finding in developmental research.

The page on IQ and memory is relevant to the general knowledge result: a person who reads a great deal encounters facts repeatedly in varied contexts, which is the condition under which long term memory consolidates them, and the Information subtest in any battery rewards that consolidation. The page on whether IQ changes with age explains why crystallized measures keep rising into late middle age while fluid measures do not, and a decade of accumulated reading is the mechanism behind the crystallized curve.

6 What Literacy Does to the Brain and to Test Performance

Learning to read reorganizes the visual and language systems of the brain and creates abilities that do not arise without it, which is why illiterate adults perform differently on almost every neuropsychological test, including nonverbal ones. Dehaene and colleagues measured the brain responses of illiterate adults, adults who learned to read as children and adults who learned to read as adults in Brazil and Portugal, and reported in Science in 2010 that reading acquisition was associated with the appearance of a region specialized for written words in the left occipitotemporal cortex, with enhanced early visual processing of horizontally oriented stimuli, and with a reduction in the response to faces in the same territory, a gain of function with a measurable cost. The changes were present in the adults who learned late as well as in those who learned as children.

Morais and colleagues had shown three decades earlier, in Cognition, that adults who had never learned to read could not add or delete a single sound at the start of a spoken word, a task that literate adults find trivial, and that the awareness of speech as a sequence of sounds does not arise spontaneously but comes with learning an alphabet. Ardila and nine colleagues' review of the neuropsychology of illiteracy in Archives of Clinical Neuropsychology in 2010 concluded that literacy is significantly associated with virtually all neuropsychological measures, that learning to read reinforces and modifies verbal and visual memory, phonological awareness and visuospatial and visuomotor skills, and that the effect of literacy on test performance reflects both changed abilities and unfamiliarity with the formats and demands of testing itself.

That last point reaches beyond the individual. Any test score from a population with limited literacy measures test familiarity along with ability, and the page on average IQ in India shows what that means for national tables built from samples of schoolchildren scored against norms from fully literate countries. The page on culture fair tests explains why even a nonverbal matrices task is not immune, since the habits of scanning rows, working left to right and treating abstract patterns as problems to be solved are themselves products of schooling and print. Reading does not only add words; it installs the routines that testing assumes.

7 Dyslexia: Reading Difficulty Is Not Low Intelligence

The relationship between reading and intelligence runs in both directions in the population and in neither direction for a person with dyslexia, whose reading difficulty has the same basis whatever their IQ. For decades dyslexia was defined by a discrepancy between reading and IQ, on the assumption that a poor reader with a low IQ was simply a slow learner while a poor reader with a high IQ had a specific disorder. Tanaka and colleagues tested that assumption with functional imaging in two independent samples totalling 131 children and reported in Psychological Science in 2011 that discrepant and nondiscrepant poor readers showed the same pattern of reduced activation in left parietotemporal and occipitotemporal regions during phonological processing. The brain basis of the phonological deficit was independent of IQ, converging with behavioural evidence that poor readers of any IQ have the same kind of difficulty.

The page on dyslexia and IQ sets out the consequences for assessment, which are that reading disability is diagnosed on reading measures rather than on a gap, and that a person with dyslexia can hold any position in the intelligence distribution. It also complicates the reading and intelligence story in a specific way. The twin study found that better early reading predicted later intelligence; a child whose reading is held back by a phonological deficit is therefore denied part of the print exposure through which vocabulary and knowledge accumulate, and the Matthew effect works against them regardless of their reasoning ability. Effective reading instruction for such a child is, on the evidence of this page, also an intervention on later crystallized ability.

For testing, the practical consequence is that a verbal index is not a reading test and should not be allowed to become one. Timed verbal subtests that require reading the items can penalize a slow decoder for reasons unrelated to vocabulary, and the page on twice exceptional profiles describes the pattern of high reasoning with a specific weakness that such a person shows. In ACIS the verbal subtests present short stimuli and the reading demand is modest, but a person with a diagnosed reading disability should read their verbal indices with that history in mind and weight the nonverbal indices accordingly.

8 Reading to Children: The Experiments

The one part of the reading literature with randomized experiments concerns reading to young children, and the effects on vocabulary are moderate, reliable and largest in the youngest children. Whitehurst and colleagues reported in Developmental Psychology in 1988 on an intervention they called dialogic reading, in which parents of children aged about two to three were trained to ask open questions, expand the child's answers and let the child become the teller of the story rather than the listener, for a month. Children in the trained group showed gains in expressive language relative to controls at the end of the month, and the advantage was still present at follow up. The design was small and the method was new; the result has since been replicated many times.

Mol, Bus, de Jong and Smeets pooled sixteen studies in Early Education and Development in 2008 and found an effect of dialogic reading on expressive vocabulary of d = 0.59 across nine studies with 322 children, a moderate effect by the usual conventions, with smaller effects on receptive vocabulary and substantially smaller effects for children aged four to five and for children at risk of language and literacy difficulties. The pattern says that the intervention works best when the child is youngest and the vocabulary to be built is expressive, and that it is not a remedy for established difficulty.

Vocabulary at three is not IQ at thirty, and no study has followed a dialogic reading experiment far enough to measure adult intelligence. What the experiments establish is the causal link that the correlational spiral assumes at its start: an adult reading interactively to a child produces a measurable gain in the child's language, and language at school entry is the strongest predictor of the reading ability that the twin study found predicting later intelligence. The page on breastfeeding and IQ reviews another early life influence with a similar evidential structure, and the page on birth order and IQ discusses why the firstborn, who receives more of this kind of attention, tends to score a point or two higher.

9 Fiction and the Mind: A Finding That Did Not Replicate

The most widely reported claim about reading in the last fifteen years, that a few pages of literary fiction improve the ability to read other minds, did not survive a large replication attempt, and the episode is a useful guide to how reading claims should be weighed. Kidd and Castano reported in Science in 2013 that participants assigned to read short passages of literary fiction scored higher on tests of theory of mind, including the Reading the Mind in the Eyes task, than participants who read popular fiction, nonfiction or nothing. The finding was small in absolute terms, immediately famous and consistent with a hopeful theory about what literature does. Panero and six colleagues at several laboratories then attempted a direct replication with a much larger sample and reported in the Journal of Personality and Social Psychology in 2016 that reading a single passage of literary fiction did not improve theory of mind scores; what did predict those scores was lifetime exposure to fiction, a correlational relationship that runs in both directions.

The contrast with the findings above is instructive. The twin study, the print exposure meta-analysis and the ten year follow up all concern years of reading measured against broad abilities, and their effects are modest, cumulative and consistent across designs. The fiction and theory of mind claim concerned minutes of reading measured against a single task, and its effect vanished when the sample grew. Reading appears to change what a person knows and can do slowly, through volume, rather than quickly, through content, and a claim that a particular kind of book does something specific to the mind in a single sitting should be held to the standard the replication set.

The page on emotional intelligence versus IQ explains why theory of mind and social cognition are measured separately from the abilities an IQ test samples, and why a gain on one would not have been a gain in IQ in any case. The page on IQ and creativity takes up a related set of claims about what reading and the arts do for thinking, with the same conclusion: the durable effects are the slow ones.

10 Reading in Adulthood and Later Life

In adults the evidence shifts from ability to decline, and it shows that a lifetime of reading is associated with slower cognitive ageing and, in one national cohort, with longer survival, through a route that runs through cognition. Wilson and colleagues followed older participants in a clinical pathological cohort with annual cognitive testing and, for 294 who died and underwent autopsy, related their rated cognitive activity in childhood, young adulthood, middle age and late life to the rate of decline, adjusting for infarcts, Lewy bodies, amyloid burden and tangle density. They reported in Neurology in 2013 that more frequent late life activity and more frequent early life activity were each associated with slower decline, that the two together accounted for 14 percent of the variability in decline not explained by neuropathology, and that the early life association was attributable to activity in childhood and middle age. Reading, writing and library use are the core of the activity measures.

Bavishi, Slade and Levy analysed 3,635 participants aged 50 and over in the Health and Retirement Study and reported in Social Science and Medicine in 2016 that book readers had a survival advantage over non-readers, a hazard ratio of 0.80 over the follow up, that the advantage was larger for books than for newspapers and magazines, that it held across sex, wealth, education and health, and that it was mediated by a cognitive measure, which is to say that reading was associated with better cognition and better cognition with survival. The page on IQ and longevity reviews the broader relationship between measured ability and mortality, of which this is one strand.

Both studies are observational, and reverse causation is the first alternative: people whose cognition is declining read less, and people who are ill read less. Wilson's design addresses the first by measuring pathology directly and finding the association in the residual, and Bavishi's addresses the second with baseline health adjustment; neither can close the question. What they support is the cognitive reserve hypothesis, that a lifetime of cognitive activity builds capacity that delays the expression of pathology, and the page on whether IQ changes with age explains why the crystallized abilities that reading builds are the ones that hold up longest.

11 Schooling as the Largest Reading Programme Ever Run

The strongest causal evidence that anything raises intelligence concerns schooling, and the first thing school does is teach and require reading, so the education effect is in part a reading effect measured at scale. Ritchie and Tucker-Drob's meta-analysis of 142 effect sizes from 42 datasets and more than 600,000 participants, using designs that control for earlier intelligence, exploit compulsory schooling reforms or use school entry cut offs, found that each additional year of education raises intelligence by about one to five points, with effects persisting across the lifespan and present on all broad categories of ability studied. The authors describe education as the most consistent, robust and durable method yet identified for raising intelligence, and the page on whether you can improve your IQ places it beside the methods that do not work.

What a year of school contains is, for most of the years that the reforms extended, reading: textbooks, set texts, written examinations, and the accumulation of vocabulary and general knowledge through print. It also contains arithmetic, instruction in abstraction, and practice at the timed, seated, silent performance that a test requires, which is why the education effect appears on nonverbal as well as verbal measures, just as the reading effect did in the twin study. The two literatures are describing overlapping mechanisms at different grains, and the page on average IQ in the UK shows what the raising of the British school leaving age looks like in the population data.

The reverse case makes the same point. Where schooling delivers little reading, test scores are low regardless of the children, and the page on average IQ in India reports survey data in which fewer than a quarter of third year pupils in government schools could read a second year text, alongside national IQ estimates built from those children. A reading effect on intelligence that is real at the level of twins is also real at the level of nations, and in both cases the honest statement is the same: the effect is on what a test measures, it is modest per year, and it accumulates.

12 The Evidence by Design

Read by design rather than by headline, the reading literature sorts into one genetically controlled longitudinal study, a large correlational base, a long follow up, brain and neuropsychological evidence, a small set of experiments in early childhood, one failed replication and two late life cohorts, and the column that matters is what each measured as the outcome.

The column to read is the outcome. The twin study measured intelligence and found reading ability predicting it. The meta-analyses measured language and reading and found volume predicting them strongly. The experiments measured vocabulary in toddlers and found a moderate effect. The brain studies measured cortical organization and found literacy changing it. The failed replication measured a single task after a single passage. The cohorts measured decline and survival. Each answers a different question, and only the first answers the one in the title of this page.

13 What the Evidence Supports, Stated Narrowly

Stated as narrowly as the designs allow, the evidence supports six sentences about reading and IQ, and each is more useful than the slogan. First, in identical twins, the twin with the better reading ability at an earlier age scores higher on intelligence measures at later ages, on nonverbal as well as verbal tests, and the effect is not explained by genes, family or the amount read; it is modest in size. Second, the amount a person reads correlates moderately to strongly with their language, comprehension and knowledge at every age, and the association grows from preschool to university, consistent with a spiral in which skill drives volume and volume drives skill. Third, an early start on reading predicts a lifelong habit and, through the habit, vocabulary and general knowledge beyond what ability explains, over a decade of follow up. Fourth, learning to read reorganizes the brain's visual and language systems and creates abilities such as phoneme awareness that do not arise without it, and literacy is associated with performance on virtually every neuropsychological test, so that test scores from populations with limited literacy measure familiarity with testing along with ability. Fifth, reading interactively to children aged two and three produces a moderate gain in expressive vocabulary, and reading difficulty in a child has the same basis whatever the child's IQ. Sixth, a lifetime of reading and related cognitive activity is associated with slower cognitive decline in late life after accounting for brain pathology, and with longer survival in a national cohort, through a route that runs through cognition.

What the evidence does not support is a figure for how many points reading adds to an IQ, because no study has measured a dose of reading against a normed battery in adults; the twin effect is expressed as a standardized association, and the schooling literature, which is the closest thing to a dose response, attributes one to five points to a whole year of education. It does not support the claim that a particular kind of reading, literary fiction in particular, does anything specific to the mind in a single sitting, because the one such claim that was tested at scale failed. It does not support the inference that a person who reads little has a low IQ or that a person who reads a great deal has a high one, because the correlations are moderate and the causation runs both ways. And it does not support the reading of any verbal index as a pure measure of ability independent of reading history, because the vocabulary and knowledge that such an index counts are built by reading.

The narrow statement leaves a person with a clear picture. Reading builds the crystallized part of intelligence directly and, on the twin evidence, contributes something to the rest; it does so slowly and cumulatively; and it is the only everyday activity for which a genetically controlled design has found such an effect.

14 What This Does Not Say About You

No study on this page can tell a reader how many points their own reading has added or would add, and the reasons are the same ones that limit every population finding. A within pair twin association says that reading ability differences precede intelligence differences on average across 1,890 pairs; it does not say that a given person who reads an hour more a day will move by a stated amount. A meta-analytic correlation says that print exposure and language ability travel together; a person who reads little and has a large vocabulary, or the reverse, is not contradicted by it. The schooling literature's one to five points per year is an average across reforms, cohorts and countries, and the page on the standard deviation of 15 explains why a few points is within the error of a single measurement of one person.

What a person can know is where their verbal knowledge stands now, on a stated scale with a stated error. An ACIS session scores five verbal subtests, including open ended Vocabulary and Information, alongside fluid, quantitative, spatial, memory and speed indices, against an adult reference frame of 3,243 English speaking records aged 16 to 90 described in the technical manual, and prints the standard error of each index. A reader whose verbal indices sit well above their fluid indices is seeing, in one plausible reading, the deposit of years of print; a reader with the reverse pattern is seeing reasoning that reading has not yet been asked to furnish. The page on Full Scale IQ explains how the indices combine, and the page on score versus percentile explains how to read the result.

Two cautions apply to the verbal indices specifically. They depend on exposure to the English register the items sample, so a person whose reading has been mostly in another language should read them with that in mind; the page on average IQ by language sets out the size of that effect. And they are not reading tests, so a person with a reading disability should note that the modest reading demand of short verbal items is still a demand, and weight the nonverbal indices accordingly. Beyond that, the evidence on this page offers one piece of practical advice, and it is the slow kind: the only documented way to raise the verbal side of a profile is to read, for years.

15 Sources Behind This Page

Every figure above is traceable to one of the following, and each is linked at the point where it is used. Effect sizes, sample sizes and percentages are quoted from the abstracts and results of the papers as published. ACIS reliability and loading figures are from the technical manual, version 1.4.

• Ritchie S J, Bates T C and Plomin R. Does learning to read improve intelligence? A longitudinal multivariate analysis in identical twins from age 7 to 16. Child Development, 2015, volume 86, issue 1, pages 23 to 36.
• Mol S E and Bus A G. To read or not to read: A meta-analysis of print exposure from infancy to early adulthood. Psychological Bulletin, 2011, volume 137, issue 2, pages 267 to 296.
• Stanovich K E. Matthew effects in reading: Some consequences of individual differences in the acquisition of literacy. Reading Research Quarterly, 1986, volume 21, issue 4, pages 360 to 407.
• Anderson R C, Wilson P T and Fielding L G. Growth in reading and how children spend their time outside of school. Reading Research Quarterly, 1988, volume 23, issue 3, pages 285 to 303.
• Cunningham A E and Stanovich K E. Early reading acquisition and its relation to reading experience and ability 10 years later. Developmental Psychology, 1997, volume 33, issue 6, pages 934 to 945.
• Dehaene S, Pegado F, Braga L W, Ventura P, Nunes Filho G, Jobert A, Dehaene-Lambertz G, Kolinsky R, Morais J and Cohen L. How learning to read changes the cortical networks for vision and language. Science, 2010, volume 330, issue 6009, pages 1359 to 1364.
• Morais J, Cary L, Alegria J and Bertelson P. Does awareness of speech as a sequence of phones arise spontaneously? Cognition, 1979, volume 7, issue 4, pages 323 to 331.
• Ardila A, Bertolucci P H, Braga L W, Castro-Caldas A, Judd T, Kosmidis M H, Matute E, Nitrini R, Ostrosky-Solis F and Rosselli M. Illiteracy: The neuropsychology of cognition without reading. Archives of Clinical Neuropsychology, 2010, volume 25, issue 8, pages 689 to 712.
• Tanaka H, Black J M, Hulme C, Stanley L M, Kesler S R, Whitfield-Gabrieli S, Reiss A L, Gabrieli J D E and Hoeft F. The brain basis of the phonological deficit in dyslexia is independent of IQ. Psychological Science, 2011, volume 22, issue 11, pages 1442 to 1451.
• Whitehurst G J, Falco F L, Lonigan C J, Fischel J E, DeBaryshe B D, Valdez-Menchaca M C and Caulfield M. Accelerating language development through picture book reading. Developmental Psychology, 1988, volume 24, issue 4, pages 552 to 559.
• Mol S E, Bus A G, de Jong M T and Smeets D J H. Added value of dialogic parent-child book readings: A meta-analysis. Early Education and Development, 2008, volume 19, issue 1, pages 7 to 26.
• Kidd D C and Castano E. Reading literary fiction improves theory of mind. Science, 2013, volume 342, issue 6156, pages 377 to 380.
• Panero M E, Weisberg D S, Black J, Goldstein T R, Barnes J L, Brownell H and Winner E. Does reading a single passage of literary fiction really improve theory of mind? An attempt at replication. Journal of Personality and Social Psychology, 2016, volume 111, issue 5, pages e46 to e54.
• Wilson R S, Boyle P A, Yu L, Barnes L L, Schneider J A and Bennett D A. Life-span cognitive activity, neuropathologic burden, and cognitive aging. Neurology, 2013, volume 81, issue 4, pages 314 to 321.
• Bavishi A, Slade M D and Levy B R. A chapter a day: Association of book reading with longevity. Social Science and Medicine, 2016, volume 164, pages 44 to 48.
• Ritchie S J and Tucker-Drob E M. How much does education improve intelligence? A meta-analysis. Psychological Science, 2018, volume 29, issue 8, pages 1358 to 1369.

16 Frequently Asked Questions