4
TEACHING OLD DOGS NEW TRICKS
Myths about Intelligence and Learning
Myth #15 Intelligence (IQ) Tests Are Biased against Certain Groups of People
Few icons of popular psychology are the subject of as many misconcep tions as are tests of the intelligence quotient (IQ; Gottfredson, 1997). So before addressing what’s perhaps the most widespread misconcep tion, a tad bit of history is in order.
More than a century ago, Charles Spearman showed that scores on measures of many diverse cognitive abilities tend to be positively corre lated. In a classic paper, he proposed a “general intelligence” factor to account for the commonality underlying these capacities (Spearman, 1904). Although Spearman recognized the existence of more specific abilities too, massive amounts of data show that mental abilities are underpinned by this factor (Carroll, 1993). Other terms for the general intelligence factor are general mental ability, IQ, and—in honor of its early pro ponent—Spearman’s g. Most IQ tests, like the widely used Wechsler Adult Intelligence Scale (Wechsler, 1997), now in its fourth version, contain multiple subtests, like vocabulary and arithmetic. The positive associations among these subtests on these tests are consistent with Spearman’s g, supporting the use of a single IQ score for many import ant purposes.
Far from being an arbitrary construct that depends entirely on how we choose to measure it, there’s consensus among most experts that intel ligence is:
a very general mental capability that, among other things, involves the ability to reason, plan, solve problems, think abstractly, comprehend com plex ideas, learn quickly and learn from experience. It is not merely book learning, a narrow academic skill, or test-taking smarts. Rather, it reflects a broader and deeper capability for comprehending our surroundings— “catching on,” “making sense” of things, or “figuring out” what to do. (Gottfredson, 1997, p. 13)
Some critics have charged that IQ tests predict performance only on other IQ tests. In a lively Internet discussion among faculty members regarding IQ tests, one participant commented that “IQ is a notoriously weak predictor of anything other than IQ” (http://chronicle.com/blogs/ election/2456/can-iq-predict-how-well-a-president-will-perform; September 19, 2008). Yet the data show otherwise. Although far from perfect measures, IQ tests yield scores that are among the most valid and cost-effective predictors of academic achievement and job performance across just about every major occupation studied—factory worker, waiter, secretary, police officer, electrician, and on and on (Neisser et al., 1996; Sackett, Schmitt, Ellingson, & Kabin, 2001; Schmidt & Hunter, 1998). Dean Keith Simonton (2006) even showed that U.S. presidents’ estimated IQs are good predictors of their success in office, as rated by historians. Because of their utility, decision-makers frequently use IQ tests in “high-stakes” (important in their real-world consequences) selection contexts, including admissions and hiring.
As the civil rights movement gathered steam in the 1960s, many researchers examined IQ score differences across racial and ethnic groups. It became popular to attribute differences among groups to test bias: Most researchers assumed that IQ tests favored white males (Anastasi & Urbina, 1997). The commonplace use of IQ tests and the weight assigned to applicants’ IQ scores mean that if these tests are biased against women or minority group members, widespread and unfair discrimina tion could result. Potential test bias is far more than a question of hair splitting or political correctness.
What’s test bias, and how would we know it if we saw it? One wide spread misunderstanding is that if any two groups score differently, the test is biased. We can find this misconception in a host of popular writ ings. It’s a particularly frequent refrain among critics of IQ testing and other standardized tests. In the early 1980s, consumer advocate (and later multiple-time presidential candidate) Ralph Nader and his colleagues argued that the SAT (then called the Scholastic Aptitude Test) should be banned because poorer students and many students from minority groups tend to do worse on it than other students (Kaplan, 1982). Writing in The Nation magazine, Jay Rosner (2003) contended that consistent differences in SAT item performance between majority and minority students demonstrate that standardized tests are biased.
Many judges have similarly ruled that differences in the test scores of two groups, such as a majority versus a minority group, imply test bias. In the influential ruling of Larry P. v. Riles (1980), the 9th District court of Appeals in California ruled that an unbiased test by definition yields “the same pattern of scores when administered to different groups of people” (p. 955) and placed strict limits on the use of intelligence tests for classifying children as mildly mentally retarded for educational purposes (Bersoff, 1981). In another early court case, the Golden Rule Insurance Company sued the state licensing board and test publisher because a smaller proportion of black than white examinees responded correctly to some items on the licensing tests (Golden Rule Insurance Company et al. v. Washburn et al., 1984). Many lawyers later filed court cases on the grounds that differences in test scores across groups prove that this test is biased.
But there’s a serious problem with this popular view: The groups may actually differ in the trait being assessed (Anastasi & Urbina, 1997). Almost surely, a physician’s records would show that the average weight of her adult male patients is greater than that of her adult female patients. This fact doesn’t suggest that the scale used to measure patients’ heights is biased, because men tend to be heavier than women. Differences between groups don’t necessarily demonstrate bias, although they might suggest it in some cases. At least some of the reason for this misunderstanding may stem from a misapplication of the representativeness heuristic (see Introduction, p. 15). For much of American history, many outcomes that showed large group differences, like differences in school achievement across races or differences in job status between men and women, were due largely to societal bias. So today, when people see that a test yields group differences, they may automatically equate these differences with bias.
How can we know whether group differences in test scores are due to bias? The trick is to focus on the validity of a test’s predictions. If we use an IQ test to predict performance in school or the workplace, we must collect data on the IQ scores of applicants and their performance. If group differences in IQ test scores are accompanied by roughly comparable differences in performance, the test is unbiased. An unbiased test neither underpredicts nor overpredicts performance for the members of any group. In contrast, if groups score differently on the IQ test but perform similarly, we can conclude that the test is biased. One consequence could be unfair discrimination in favor of the group whose performance is over-predicted and against the group whose performance is underpredicted.
Fortunately, many researchers have studied the possibility that IQ test scores are biased against women or minorities. Two panels assembled by the National Academy of Science (Hartigan & Wigdor, 1989; Wigdor & Garner, 1982) and a Task Force of the American Psychological Association (Neisser et al., 1996), each of which contained individuals representing a diverse range of expertise and opinions, reached the same conclusion: There’s no evidence that IQ tests or other standardized tests, like the SAT, underpredict the performance of women or minor ities. Today, most experts agree that the question of IQ test bias has been settled about as conclusively as any scientific controversy can be (Gottfredson, 1997, 2009; Jensen, 1980; Sackett et al., 2001; Sackett, Borneman, & Connelly, 2008).
It’s crucial to understand, though, that the absence of test bias doesn’t say anything about the causes of group differences in IQ; these differ ences could be due largely or entirely to environmental influences, like social disadvantages or prejudice. To the extent that we blame group differences in IQ on test bias, we may ignore the genuine causes of these differences, some of which we may be able to remedy with social and educational programs.
Despite the research evidence, some psychologists argue that the test bias claim contains a kernel of truth. Here’s why. Researchers can evaluate potential bias not only at the level of a whole test, but at the level of the items making up a test. Just as a biased test would under predict one group’s ability relative to that of another, a biased test item would do the same. Psychologists refer to this phenomenon as differ ential item functioning, or DIF (Hunter & Schmidt, 2000). For any pair of groups (such as women versus men, or blacks versus whites), we can examine each item on an IQ test for DIF. If members of two groups perform about the same on the rest of the test but score differently on a particular item, this finding provides evidence of item bias. Researchers commonly find that a number of IQ test items meet criteria for DIF. Roy Freedle and Irene Kostin (1997) found DIF for a number of verbal analogy items on the SAT and GRE tests, including those with easy stems like “canoe: rapids” and hard stems like “sycophant: flattery.” At first blush, finding DIF for many test items seems to call into question the verdict of no test bias. After all, how can the items themselves demon strate DIF without scores on the whole test being biased?
It turns out that many or most instances of DIF are trivial in size (Sackett et al., 2001). Even among items that exhibit DIF, the direction of bias is inconsistent. Some items favor one group and other items favor the other group, so the effects tend to cancel out when the items are com bined into the total score (Sackett et al., 2001). So DIF doesn’t neces sarily produce test bias (Freedle & Kostin, 1997).
As we’ve discovered throughout this book, the gulf between research and popular opinion is often wide, and this is especially the case in the domain of intelligence (Phelps, 2009). IQ tests validly predict perform ance in many important realms of everyday life, with no evidence of bias against women or minorities. The real bias occurs when we blame the “messengers”—that is, the IQ tests themselves—and neglect potential environmental explanations, such as cultural disadvantage, for differences in test scores across groups.
My th #16 If You’re Unsure of Your Answer When Taking a Test, It’s Best to Stick with Your Initial Hunch
Few phrases instill more fear into the hearts and minds of college students than those three dreaded words: “multiple choice test.” Probably because many undergraduates would prefer sitting on a bed of nails to taking a multiple choice test, they’re always on the lookout for tips to boost their performance on most professors’ favorite weapon of intellectual torture. Fortunately, a handful of these test-taking pointers actually boast some scientific support. For example, on multiple-choice tests, longer answers are slightly more likely than other answers to be correct, as are more precise answers (for example, in response to the stem “The U.S. Constitution was adopted in ____”, “1787” is more precise than “between 1770 and 1780”) and “all of the above” answers (Geiger, 1997; Gibb, 1964).
Yet perhaps the most widely accepted piece of test-taking folklore is to stick with your original answer, especially if you’re unsure whether it’s right or wrong. Across various surveys, large proportions—between 68% and 100%—of college students say that changing their initial answers on a test won’t improve their score. About three fourths say that chang ing their answers will actually lower their score (Ballance, 1977; Benjamin, Cavell, & Shallenberger, 1984). This myth—sometimes called the “first instinct fallacy”—isn’t limited to undergraduates. In one study, among professors who gave their college students advice about changing answers on tests, 63% told them not to do so because it would tend to lower their scores. Among science and liberal arts professors, only 5–6% said that changing answers would tend to increase students’ scores; the per centage among education professors was 30% (Benjamin et al., 1984).
What’s more, scores of websites, including those designed to provide students with test-taking advice, inform readers that changing their initial answers is a bad strategy and encourage them to trust their first hunches. One website tells students, “Don’t keep on changing your answer—usually your first choice is the right one, unless you misread the question” (TestTakingTips.com) and another advises them to “Trust your first hunch. When you answer a question, go with your first hunch —don’t change your answer unless you’re absolutely sure you’re right” (Tomahawk Elementary School). Another goes further, even citing research support for this belief: “Be wary of changing your mind: There is evidence to suggest that students more frequently change right answers to wrong ones than wrong answers to right ones” (Fetzner Student-Athlete Academic Center).
What do the scientific findings actually say? With over 3 million high school students taking the SAT and ACT (interestingly, in the case of both tests the letters don’t stand for anything) each year, this question is hardly trivial. In fact, the research evidence is surprisingly consistent, and it points to the opposite conclusion presented on these websites (Benjamin et al., 1984; Geiger, 1996; Skinner, 1983; Waddell & Blankenship, 1994). More than 60 studies lead to essentially the same verdict: When students change answers on multiple-choice tests (typically as judged by their erasures or cross-outs of earlier answers), they’re more likely to change from a wrong to a right answer than from a right to a wrong answer. For each point that students lose when changing from a right to a wrong answer, they gain between two and three points on average in changing from a wrong to a right answer (Benjamin et al., 1984; Foote & Belinky, 1972; Geiger, 1996). In addi tion, students who change more answers tend to receive higher test scores than other students, although this finding is only correlational (see Intro duction, p. 13) and may reflect the fact that frequent answer-changers are higher test performers to begin with (Geiger, 1997; Friedman & Cook, 1995). All of these conclusions hold not merely for multiple choice tests given in classes, but for standardized tests like the SAT and Graduate Record Exam (GRE).
Admittedly, there are two qualifications to the “when in doubt, change your answer” strategy. First, research suggests that students shouldn’t change their answer if they’re merely guessing this answer might be wrong; changing one’s answer is beneficial only when students have a good reason to suspect their answer is wrong (Shatz & Best, 1987; Skinner, 1983). Second, there’s some evidence that students who do poorly on multiple choice tests may benefit less from changing their answers than other students (Best, 1979). So these students may want to change their answers only when they’re fairly certain these answers are wrong.
There’s surprisingly little research addressing the question of why students believe that changing their initial answers is usually a bad idea. But three likely explanations come to mind. First, as we’ve seen, most professors who give their students advice about changing their answers advise them not to do so (Benjamin et al., 1984). So this mistaken belief is probably spread partly by word-of-mouth (Higham & Gerrard, 2005). Second, research suggests that students are more likely to remember items whose answers they changed from right to wrong than those they changed from wrong to right (Bath, 1967; Ferguson, Kreiter, Peterson, Rowat, & Elliott, 2002). Because the bitter taste of incorrect decisions lingers longer than the memory of correct decisions (“Why on earth did I change that answer? I had it right the first time”), our test-taking mistakes typically stick in our minds. As a consequence, a phenomenon called the availability heuristic may lead students to overestimate the risk of committing errors when changing answers. As we learned earlier (see Introduction, p. 15), a heuristic is a mental shortcut or rule of thumb. When we use the availability heuristic, we’re estimating the likelihood of an event by how easily it comes to our minds. Indeed, research shows that students who change right answers to wrong answers recall these decisions much better than do students who change wrong answers to right answers, largely because the former changes create a more lasting emotional impact (Kruger, Wirtz, & Miller, 2005). Third, research indicates that most students overestimate how many answers they get right on multiple choice tests (Pressley & Ghatala, 1988), so they may assume that changing answers is likely to lower their score.
So to cut to the bottom line: When in doubt, we’re usually best not trusting our instincts. After all, our first hunches are just that—hunches. If we have a good reason to believe we’re wrong, we should go with our head, not our gut, and turn that pencil upside-down.
Myth #17 The Defining Feature of Dyslexia Is Reversing Letters
Humor often reveals our conceptions—and misconceptions—of the world. For example, few psychological conditions are the butt of as many jokes as dyslexia: “I’m an agnostic dyslexic with insomnia. I lay awake all night trying to work out if there really is a Dog.” Or, “Dyslexics of the world, untie!”
Yet to people with dyslexia, these jokes aren’t especially funny. Not only do they poke fun at people with a disability, but they reinforce inaccurate stereotypes of people with a genuine psychological condition. They also underscore just how distant the public’s conception of dyslexia is from reality. Most people believe that the defining feature of dyslexia is “mirror writing” or “mirror reading” (Fiorello, 2001; Gorman, 2003). Indeed, many laypersons believe that dyslexics literally see letters backward. Two types of reversals are commonly associated in the public mind with dyslexia: (1) reversing letters themselves, like writing or seeing “b” instead of “d,” and (2) reversing the order of letters within words, like writing “tar” instead of “rat.” Even among educators, including university faculty, special education teachers, and speech therapists, 70% believe that the second problem is a defining fea ture of dyslexia (Wadlington & Wadlington, 2005). In another survey, about 75% of basic education teachers identified odd spellings, especially reversals of the order of letters within words, as a key sign of dyslexia (Kerr, 2001).
The belief that dyslexia is underpinned by letter reversals has early roots (Richardson, 1992). In the 1920s, American neurologist Samuel Orton (1925) coined the term strephosymbolia (meaning “twisted symbol”) to refer to the tendency to reverse letters, and hypothesized that it was the underlying cause of dyslexia. He also claimed that some children with this condition could read more easily if they held writing up to a mirror. Orton’s views helped to perpetuate the longstanding belief that letter reversals are central to dyslexia (Guardiola, 2001).
This view, or variants of it, is bolstered by media portrayals of—and jokes about—dyslexia. A 1984 ABC movie, Backwards: The Riddle of Dyslexia, stars a 13-year-old child, Brian Ellsworth (portrayed by the late River Phoenix), who reverses letters in words. The 1994 comedy film, Naked Gun 33 1/3, shows lead character Frank Drebin (portrayed by Leslie Nielsen) reading a newspaper featuring the headline “Dyslexia for Cure Found.” In the 2001 film, Pearl Harbor, Captain Rafe McCauley (portrayed by Ben Affleck) informs the nurse administering an eye exam that he can’t read letters because “I just get ‘em backward sometimes.” And on a National Public Radio show on dyslexia in 2007, the host stated that the “simplest explanation, I suppose, is that you see things backwards” (National Public Radio, 2007).
But what is dyslexia, anyway? Dyslexia (meaning “difficulty with words”) is a learning disability marked by difficulties in processing written language (Shaywitz, 1996). Most often, dyslexics experience problems with reading and spelling despite adequate classroom instruction. Often, they find it challenging to “sound out” and identify printed words. About 5% of American children suffer from dyslexia. Despite what many people believe, dyslexia isn’t an indicator of low mental ability, because dyslexia occurs in many highly intelligent people (Wadlington & Wadlington, 2005). Indeed, the formal psychiatric diagnosis of dyslexia (or more technically, “reading disorder”) requires that children’s over all intellectual ability be markedly superior to their reading ability (American Psychiatric Association, 2000).
The causes of dyslexia are controversial, although most researchers believe that dyslexics experience difficulty with processing phonemes, the smallest units of language that contain meaning (Stanovich, 1998; Vellutino, 1979). The English language, for example, contains 44 phonemes, such as the “c” in “cat” and the “o” in “four.” Because dyslexics find it difficult to parse words into their constituent phonemes, they often make mistakes when identifying words (Shaywitz, 1996). Some researchers believe that a subset of dyslexics is marked by visual deficits in addition to deficits in phoneme processing (Badian, 2005; Everatt, Bradshaw, & Hibbard, 1999), but this view is not universally accepted (Wolff & Melngailis, 1996). In any case, there’s no evidence that dyslexics literally “see” letters backward or in reverse order within words. Research on twins strongly suggests that dyslexia is partly influenced by genetic factors (Pennington, 1999).
More important, research conducted over the past few decades demonstrates that letter reversals are hardly distinctive to dyslexia. Both backward writing and letter reversals are commonplace in the early phases of spelling and writing of all children age 6 and younger (Liberman et al., 1971; Shaywitz, 1996), not merely dyslexic children. These errors decrease over time in both groups of children, although less so among dyslexic children. In addition, most research suggests that letter reversals are only slightly more frequent, and in some studies no more frequent, among dyslexic than non-dyslexic children (Cassar, Treiman, Moats, Pollo, & Kessler, 2005; Lachman & Geyer, 2003; Moats, 1983; Terepocki, Kruk, & Willows, 2002). Letter reversals also account for only a small minority of the errors that dyslexic children make, so they’re hardly a defining feature of the condition (Guardiola, 2001; Terepocki et al., 2002). Finally, although dyslexic children are worse spellers than other children of their age, teachers who’ve worked extensively with dyslexic children can’t distinguish their spellings from those of non-dyslexic, but younger, writers (Cassar et al., 2005). This finding supports the view that normal children make similar spelling errors to those of dyslexic children, but typically “outgrow” them.
So the next time someone asks you if you’ve heard the joke about the person with dyslexia who answers the phone by saying “O hell,” you can politely reply that this view of dyslexia is now a few decades out of date.
Myth #18 Students Learn Best When Teaching Styles Are Matched to Their Learning Styles
In the headline story “Parents of Nasal Learners Demand Odor-based Curriculum,” writers at the satirical newspaper, The Onion (2000), poked good-natured fun at the idea that there is a teaching style to unlock every underperforming student’s hidden potential (http://www.runet.edu/~thompson/obias.xhtml). We’ve all observed students in the same classes learning in different ways. Many people believe that all students could achieve at the same level if only teachers would tailor their teaching styles to each student’s learning style. As one parent in The Onion story put it, “My child is not stupid. There simply was no way for him to thrive in a school that only caters to traditional students who absorb educational concepts by hearing, reading, seeing, discussing, drawing, building, or acting out.” An educational researcher noted that “Nasal learners often have difficulty concentrating and dislike doing homework … If your child fits this description, I would strongly urge you to get him or her tested for a possible nasal orientation.” According to the story, we don’t need to consider ability or motivation, because all students are equally capable. Any failure to learn means only that teachers haven’t accommodated adequately to a student’s learning style.
Of course, the nasal story was fiction, but it’s not all that far from reality. Plug the words “learning styles” into an Internet search engine, and you’ll find any number of websites claiming to diagnose your preferred learning style in a matter of minutes. One informs visitors that “Learning styles are a way to help improve your quality of learning. By understanding your own personal styles, you can adapt the learning process and techniques you use.” It also directs them to a free “Learning Styles Inventory” that over 400,000 people have taken (http://www.learning-styles-online.com). There, you can find out whether you’re primarily a visual learner, a social learner, an auditory (sound) learner, a physical learner, and so on. These sites are premised on a straightforward and widely accepted claim: Students learn best when teaching styles are matched to their learning styles.
It’s understandable why this view is so popular: Rather than imply ing that some students are “better” or “worse” learners overall than others, it implies that all students can learn well, perhaps equally well, given just the right teaching style (Willingham, 2004). In addition, this view dovetails with the representative heuristic: like goes with like (see Introduction, p. 15). Advocates of this hypothesis claim that verbally oriented students learn best from teachers who emphasize words, visually oriented students learn best from teachers who emphasize images, and so on.
Ronald Hyman and Barbara Rosoff (1984) described the four steps of the learning styles (LS) approach: (1) Examine students’ individual learning styles, (2) classify each style into one of a few categories, (3) match it to the teaching style (TS) of a teacher or request that teachers adjust their TS to match the student’s LS, and (4) teach teachers to perform steps 1-3 in their training programs. These authors noted that each step imposes a requirement for the approach to work. These requirements include (a) a clear concept of LS, (b) a reliable and valid way to assess and classify students’ LS, (c) knowledge of how LS and TS interact to influence learning, and (d) the ability to train teachers to adjust their TS to match students’ LS. Writing in 1984, Hyman and Rosoff didn’t believe that any of these requirements had been met. We’ll soon see if their negative verdict has stood the test of time.
The notion that assessing students’ LS is effective has become a virtual truism in educational theory and practice. It’s been extolled in many popular books, such as Teaching Students to Read through Their Individual Learning Styles (Carbo, Dunn, & Dunn, 1986), and Discover Your Child’s Learning Style: Children Learn in Unique Ways (Willis & Hodson, 1999). In an article entitled “Dispelling outmoded beliefs about student learning” in a popular educational journal, the authors debunked 15 myths about student learning, but began by proclaiming that the belief that “Students learn best when instruction and learning context match their learning style” was well supported (Dunn & Dunn, 1987, p. 55). In many school districts, questions about matching TS to LS are routine in interviews for aspiring teachers (Alferink, 2007). Many teachers share the field’s enthusiasm: The results of one survey of 109 science teachers revealed that most displayed positive attitudes toward the idea of matching their TS to students’ LS (Ballone & Czerniak, 2001). Not surprisingly, workshops on educating instructors about matching their styles to students’ learning styles are popular, often attracting hundreds of teachers and principals (Stahl, 1999). In some schools, teachers have even asked children to wear shirts emblazoned with the letters V, A, K, which, as we’ll soon learn, stand for three widely discussed learning styles—visual, auditory, and kinesthetic (Geake, 2008).
The prevalence of these beliefs is underscored by the sheer volume of articles published in the educational literature on LS, the vast number of LS models proposed, and the enormous commercial success of LS measures. An August, 2008 search of the ERIC database, which catalogues educational scholarship, revealed a whopping 1,984 journal articles, 919 conference presentations, and 701 books or book chapters on LS. In the most comprehensive review of the LS literature, Frank Coffield and his colleagues (Coffield, Moseley, Hall, & Ecclestone, 2004) counted no fewer than 71 LS models. For example, the “VAK” model targets visual, auditory, and kinesthetic learners, who allegedly learn best by seeing and reading, listening and speaking, or touching and doing, respectively. Peter Honey and Alan Mumford’s (2000) model classifies students into four categories: “activists,” who immerse themselves in new experiences, “reflectors,” who sit back and observe, “theorists,” who think through problems logically, and “pragmatists,” who apply their ideas to the real world.
The LS movement has even embraced models and measures developed for very different purposes. Howard Gardner’s (1983) influential theory of multiple intelligences is often considered an LS classification, and some teachers use the Myers-Briggs Type Indicator (Briggs & Myers, 1998), which was developed as a psychoanalytically oriented personality inven tory (Hunsley, Lee, & Wood, 2003), to classify students’ LS. Honey and Mumford’s (2000) Learning Styles Questionnaire is popular, as are two different measures both called the Learning Styles Inventory (Dunn, Dunn, & Price, 1999; Kolb, 1999).
Among the 3,604 ERIC entries related to LS, less than one quarter are peer-reviewed articles. Likewise, Coffield et al. (2004) compiled a database of thousands of books, journal articles, theses, magazine articles, websites, conference papers, and unpublished literature. Few were published in peer-reviewed journals and fewer still were well-controlled studies. In other words, much of LS literature is flying “under the radar,” bypassing anonymous critical feedback by expert scholars.
Fortunately, theory and research are available to address each of the four requirements spelled out by Hyman and Rosoff (1984). First, is there a clear concept of LS? The answer appears to be no. Among the most popular of the LS models Coffield et al. (2004) reviewed, the differences are much more striking than the similarities. For example, the VAK model is based on learners’ preferred sensory modalities (visual, auditory, or kinesthetic), whereas the Honey–Mumford model, which divides students into activists, reflectors, theorists, and pragmatists, doesn’t even address the issue of sensory modalities. There’s no agreement on what LS is, despite decades of study.
Second, is there a reliable and valid way to assess students’ LS? Again, the answer seems to be no (Snider, 1992; Stahl, 1999). Gregory Kratzig and Katherine Arbuthnott (2006) found no relationship between LS classifications and memory performance on visual, auditory, and kinesthetic versions of a task. Supposedly visual learners did no better at the visual version of the task than the auditory or kinesthetic versions, and the same was true for each preferred sensory modality. Perhaps one reason for the unsatisfactory reliability and validity of LS inventories is that these measures usually assess learning preferences devoid of context (Coffield et al., 2004; Hyman & Rosoff, 1984). In other words, models and measures of LS don’t come to grips with the possibility that the best approaches to teaching and learning may depend on what students are trying to learn. Consider the first question on the Paragon Learning Style Inventory (http://www.oswego.edu/plsi/plsi48a.htm): “When you come to a new situation you usually (a) try it right away and learn from doing, or (b) like to watch first and try it later?” It’s difficult to answer this question without knowing the type of new situation. Would you learn to read a new language, solve mathematical equations, and perform gymnastics routines using the same methods? If so, we’d certainly be concerned. Most LS models don’t place learning into a meaningful con text, so it’s not surprising that measures based on these models aren’t especially reliable or valid.
Third, is there evidence to support the effectiveness of matching instructors’ TS to students’ LS? From the 1970s onward, at least as many studies have failed to support this approach as have supported it (Kavale & Forness, 1987; Kratzig & Arbuthnott, 2006; Stahl, 1999; Zhang, 2006). That’s mostly because certain TSs often yield better results than all others regardless of students’ LS (Geake, 2008; Zhang, 2006). The 2007 film Freedom Writers, starring Hilary Swank as real-life teacher Erin Gruwell, illustrates this point. After a shaky beginning as a teacher with students torn by boundaries of race, Gruwell became engrossed in her students’ lives and immersed them in the study of the Holocaust. By adopting a teaching style that went beyond ordinary classroom methods, she helped all of her students to appreciate and avoid the pitfalls of prejudice. Yet Gruwell didn’t match her TS to students’ LS. Instead, like many great teachers, she achieved outstanding results by developing an innovative TS to which the entire class responded enthusiastically.
Fourth, can educators train teachers to adapt their TS to match stud ents’ LS? Again, the commercial claims outstrip the scientific evidence. Coffield et al. (2004) noted minimal research support for this possibility, and positive results for using LS inventories to guide teaching training are at best weak. There are no clear implications for teaching practices because few well-conducted studies provide evidence, and those that do offer inconsistent advice.
So the popular belief that encouraging teachers to match their TS to students’ LS enhances their learning turns out to be an urban legend of educational psychology. To the extent that this approach encourages teachers to teach to students’ intellectual strengths rather than their weak nesses, it could actually backfire. Students need to correct and compensate for their shortcomings, not avoid them. Otherwise, their areas of intellectual weakness may grow still weaker. Because life outside the classroom doesn’t always conform to our preferred styles of learning, good teaching must prepare us to confront real-world challenges. We agree with Frank Coffield, who said that “We do students a serious disservice by implying they have only one learning style, rather than a flexible repertoire from which to choose, depending on the context” (Henry, 2007).
Chapter 4: Other Myths to Explore
| Fiction | Fact |
| Extremely intelligent people are more physically frail than other people. | With raro exceptions, extremely intelligent people tend to be in better physical health than other individuals. |
| IQ scores almost never change over time. | Although IQ scores tend to be quite stable in adulthood, they are unstable in childhood; moreover, even in adults, shifts of 5–10 points over a few months can occur. |
| IQ scores are unrelated to school performance. | IQ scores are moderately to highly predictive of grades in school, including high school and college. |
| The SAT and other standardized tests are highly coachable. | Most studies show that total SAT scores increase an average of only about 20 points as a consequence of coaching. |
| There’s a close link between genius and insanity. | There’s no evidence that high IQ predisposes to psychotic disorders; to the contrary, the IQ scores of people with schizophrenia tend to be slightly lower than those of people in the general population. |
| Mental retardation is one condition. | There are over 500 genetic causes of mental retardation in addition to environmental causes, such as accidents during birth. |
| Most mentally retarded individuals are severely retarded. | About 85% of mentally retarded individuals are classified as mildly retarded. |
| There is no association between brain size and IQ. | Brain size and IQ are moderately correlated in humans. |
| Women are worse drivers than men. | Even after controlling for the fact that men drive more than women, men get into 70% more car accidents than women, perhaps because men take more risks as drivers. |
| Creative breakthroughs occur in sudden bursts of insight. | Brain imaging studies reveal that well before people suddenly report a creative answer to a problem, brain areas involved in problem-solving, such as the frontal lobes, have already been active. |
| Very high levels of motivation usually help when solving difficult problems. | Very high levels of motivation typically impair performance on difficult problems. |
| Negative reinforcement is a type of punishment. | Negative reinforcement and punishment are opposite in their effects; negative reinforcement increases the frequency of a behavior by withdrawing an aversive stimulus, whereas punishment decreases the frequency of a behavior. |
| Punishment is a highly effective means of changing long-term behavior. | Although punishment inhibits behavior in the short term, it tends to be less effective than reinforcement for shaping behavior in the long term. |
| The best means of maintaining a behavior is to reward every response. | The best means of maintaining a behavior is to reward desired responses only intermittently. |
| B. F. Skinner raised his daughter in a “Skinner box,” contributing to her psychosis later in life. | Skinner raised his daughter in a specially designed crib, not a Skinner box; moreover, she never developed a psychosis. |
| Small class sizes consistently promote better student achievement. | The association between class size and achievement is mixed and inconsistent, although small class size may exert small positive effects among poorly performing children. |
| Grouping students in classes by their ability levels promotes learning. | Most studies show that “ability grouping” produces few or no effects on student learning. |
| Holding immature or underperforming students back a grade can be helpful. | Most research suggests that grade retention is largely ineffective in enhancing achievement, and may result in poorer emotional adjustment. |
| Standardized test scores don’t predict later grades. | Scores on the SAT and GRE are moderate to high predictors of later grades in samples with a broad range of SAT and GRE scores. |
| Direct and immediate feedback is the best means of ensuring long-term learning. | Irregularly provided feedback best promotes long-term learning. |
| “Discovery learning” (in which students must discover scientific principles on their own) is superior to direct instruction. | For tasks involving scientific reasoning, direct instruction is often superior to discovery learning. |
| The standardized test scores of U.S. students have been declining in recent decades. | Declines on the SAT and other standardized tests appear due largely or entirely to students with a broader range of abilities taking these tests in recent decades. |
| Students typically recall only 10% of what they read. | This is an urban legend with no scientific support. |
| Speed reading courses are effective. | Virtually all speed reading courses are ineffective, because they diminish comprehension. |
| Subvocalizing increases reading ability. | Subvocalizing slows down our reading speed, because we can read much more quickly than we can speak. |
| Deaf people can understand most of what other people say by reading lips. | Even the best lip-readers can understand only about 30–35% of what speakers are saying. |
| Some people “speak in tongues.” | There’s no scientific evidence for genuine “glossolalia,” that is, speaking in tongues. |
| Many identical twins have their own private language. | There’s no evidence that twins have “cryptophasia” (secret language); reports to the contrary appear due to the fact that twins often share similar language impairments, which they accommodate in each other. |
| Albert Einstein had dyslexia. | There’s no good evidence that Einstein was dyslexic. |
Sources and Suggested Readings
To explore these and other myths about intelligence and learning, see Alferink (2007); DeBell and Harless (1992); Della Sala (2007); Druckman and Bjork (1991); Druckman and Swets (1988); Ehrenberg, Brewer, Gamoran, and Willms (2001); Furnham (1996); Greene (2005); Jimerson, Carlson, Rotert, Egeland, and Sroufe (1997); Lubinski, Benbow, Webb, and Bleske-Rechek (2006); Phelps (2009); Sternberg (1996); Willerman (1979).