A total of 275 "effect size rs" were then calculated. An effect size r is a number ranging from -1.0 to +1.0 that represents the strength of the relationship between two variables. An effect size of +1.0 would mean a perfect positive correlation between two variables - e.g., the more air you put in your tires, the greater the air pressure. An effect size r of .10 is considered small in size, comparable to the difference in height between 15 and 16 year old girls (.5 inches). An effect size r of .24 is considered medium in size, comparable to the average height difference in 14 to 18 year old girls of one inch. And an effect size r of .37 is considered large in size, equivalent to the difference in IQ between typical college freshmen and those who have attained doctorates.

A set of 10 meta-analyses were conducted. A meta-analysis combines and compares effect sizes across groups of studies that address similar research questions. Statistical analyses are then used to determine whether the effect size can be generalized to new studies on the same research question. If the effect size cannot be generalized to new studies, we must conclude that the finding is not reliable and is not likely to hold up.

The effect sizes found in the REAP analyses ranged from small to large. Although small or medium differences may seem trivial, they may in fact turn out to be of practical importance. For instance, the relationship between taking an aspirin a day and reduction of heart attack risk is only r = .03! Thus when judging the value of any intervention, we must attend not only to the size of the effect but also to the importance of the outcome compared to the cost of the intervention in effort and dollars. If a small effect size is due to scores on standardized tests of mathematics increasing an average of 3 points, that seems of little consequence. However, if a small effect size is due to even a few children staying in school as opposed to dropping out, that is of great consequence.

We found three areas in which clear causal links could be demonstrated between the arts and achievement in a non-arts, academic area.

In two areas, causal links were found but only on the basis of a very few studies. Because of the small number of studies in these two areas, we urge caution and more rigorous research before conclusions can be confidently drawn.

In five areas, no clear causal links were found. The lack of findings in these five areas is attributable to one or more of three factors: in some cases the failure to find a causal link probably reflects the fact that there is in fact no causal link; in some cases a causal link was found but it was not strong enough to be reliably generalized to other studies; and in other cases, the lack of findings may have been due to the small number of studies carried out on a given research question.

Listening to Music and Spatial-Temporal Reasoning
Based on 26 reports (36 effect sizes), a medium-sized causal relationship was found between listening to music and temporary improvement in spatial-temporal reasoning. However, there was wide variation in the studies, with some showing the effect clearly and many not showing the effect at all. Moreover, the existing research does not reveal conclusively why listening to music affects spatial-temporal thinking. For education, such a finding has little importance, since it is temporary and not consistently found. Scientifically, however, this finding is of interest because it suggests that music and spatial reasoning are related psychologically (i.e., they may rely on some of the same underlying skills) and perhaps neurologically as well (i.e. they may rely on some of the same, or proximal, brain areas). Further research is needed to understand the mechanism by which certain types of music influence spatial skills.

Learning to Play Music and Spatial Reasoning
Based on 19 reports (29 effect sizes), a large causal relationship was found between learning to make music and spatial-temporal reasoning. The effect was greater when standard music notation was learned as well, but even without notation the effect was large. The value for education is greater here, since the effect works equally for both general and at risk populations, costs little since it is based on standard music curricula, and influences many students (69 of every 100, 3-to-12 year old students). Of course we must still determine the value of improved spatial skills for success in school. Spatial skills might or might not be of benefit to students, depending on how subjects are taught. For example, mathematics or geography might be taught spatially, and if they are, then students with strong spatial abilities should have an advantage in these subjects. Sadly, many schools offer few chances to apply spatial abilities.

Classroom Drama and Verbal Skills
Based on 80 reports (107 effect sizes), a causal link was found between classroom drama (enacting texts) and a variety of verbal areas. Most were of medium size (oral understanding/recall of stories, reading readiness, reading achievement, oral language, writing), one was large (written understanding/recall of stories), and one was small and could not be generalized to new studies (vocabulary). In all cases, students who enacted texts were compared to students who read the same texts but did not enact them. Drama not only helped children's verbal skills with respect to the texts enacted; it also helped children's verbal skills when applied to new, non-enacted texts. Thus, drama helps to build verbal skills that transfer to new materials. Such an effect has great value for education: verbal skill is highly valued, adding such drama techniques costs little in terms of effort or expense, and a high proportion of students are influenced by such curricular changes.

Learning to Play Music and Mathematics
Based on 6 reports (6 effect sizes), a small causal relationship was found between music training and math. However, while 3 of these studies produced medium effects, 3 produced either very small effects or none at all. If the two studies measuring pre-school math rather than school math had not been included (because these were measures of spatial recognition and perception), a reliable effect would have resulted. However, more studies are needed before any firm conclusions can be drawn.

Dance and Nonverbal Reasoning
Based on 3 reports (4 effect sizes), a small to medium sized causal relationship was found between dance and improved visual-spatial skills. Again, however, more studies are needed before any firm conclusions can be drawn.

Arts-Rich Education and Verbal and Mathematics Scores/Grades
Based on 31 reports (66 effect sizes), a small to medium correlation was found between studying the arts and academic achievement as measured primarily by test scores. However, no evidence was found that studying the arts causes academic indicators to improve. The correlational findings can be explained by non-causal mechanisms. For example, high achieving students (no matter what their ethnic or racial group, no matter what their social class) may choose or be guided to study the arts. This would then result in the finding that students who take arts courses are also high-achieving, high test-scoring students.

Arts-Rich Education and Creative Thinking
Based on 4 reports (6 effect sizes), no relationship was found between studying arts and verbal creativity test measures. A small to medium sized relationship was found between studying arts and figural creativity tests (which themselves are visual tests) but this relationship could not be generalized to new studies.

Learning to Play Music and Reading
Based on 6 reports (6 effect sizes), a small relationship was found between music and reading but this relationship could not be generalized to new studies.

Visual Arts and Reading
Based on 5 reports in which visual arts was taught separately from reading (7 effect sizes), a very small relationship between visual arts and reading was found, but this relationship could not be generalized to new studies. This effect was entirely due to reading readiness outcomes (which are themselves visual), and did not hold up for reading achievement outcomes. Based on 4 reports in which visual arts were integrated with reading instruction (4 effect sizes), a medium sized relationship was found between integrated arts/reading instruction and reading outcomes. However, this result could not be generalized to new studies.

Dance and Reading
Based on 4 reports (4 effect sizes), a small relationship between dance and reading was found, but this relationship could not be generalized to new studies.

Let's stop justifying the arts instrumentally. This is a dangerous (and peculiarly American) practice. Anyone who looks closely, as we have done, will see that these claims do not hold up unequivocally. Those who live by instrumental claims risk dying by such claims.

The arts offer a way of thinking unavailable in other disciplines. The same might be said of athletics. Suppose coaches began to claim that playing baseball increased students' mathematical ability because of the complex score keeping involved. Then suppose researchers set out to test this and found that the claim did not hold up. Would school boards react by cutting the budget for baseball? Of course not. Because whatever positive academic side effects baseball might or might not have, schools believe sports are inherently good for kids. We should make the same argument for the arts: the arts are good for our children, irrespective of any non-arts benefits that the arts may in some cases have. Just as a well-rounded education requires education of the body through physical education, a balanced education requires study of the arts.

Let's bet on history. Of course, we do not know for sure what is the best education for children to ensure that they will grow up to lead productive and happy lives. But the arts have been around longer than the sciences; cultures are judged on the basis of their arts; and most cultures and most historical eras have not doubted the importance of studying the arts. Let's assume, then, that the arts should be a part of every child's education and treat the arts as seriously as we treat mathematics or reading or history or biology. Let's remember why societies have always included the arts in every child's education. The reason is simple. The arts are a fundamentally important part of culture, and an education without them is an impoverished education leading to an impoverished society. Studying the arts should not have to be justified in terms of anything else. The arts are as important as the sciences: they are time-honored ways of learning, knowing, and expressing.

We recommend two kinds of studies to advance our understanding of the relationship between arts and non-arts outcomes: theory-building studies and theory-driven experiments. Both types require rigorous methods. Here is an example of each type.

One possibility is that the same schools that treat the arts seriously institute other kinds of innovations that are favorable to academic learning. For instance, these schools may become more inquiry-oriented, more project-based, more demanding of high standards, and more focussed on processes that lead to excellence. Educators and policy makers need to understand what comes along with the arts.

To discover this, researchers need to carry out ethnographic studies of exemplary schools that grant the arts a serious role in the curriculum. What kinds of innovations have been made in these schools to foster excellence? If certain innovations are always found in schools that grant the arts a serious role, this finding could account for why schools with serious arts programs have high academic performance.

Experimental studies thus far have not tested this hypothesis. What is needed are comparisons of academically strong vs. academically at-risk students taught the same subject matter with and without the arts as entry points. Can we identify students who first experience success in the art form and subsequently go on to show heightened interest and effort in the academic subject matter? And do levels of interest and/or motivation predict later achievement in that subject matter?

It is also possible that all students would benefit from an arts-integrated approach, even those who are high achievers to begin with, simply because an arts-integrated approach makes any subject more interesting. This hypothesis also deserves a rigorous test.

Research in the two directions suggested here can help us to understand the puzzling finding that when the arts are granted a serious role in our schools, academic achievement often rises. It is time to look seriously at the possibility that the arts are associated with academic achievement because of other academic innovations that are made in schools that bring in the arts, and/or because the arts provide engaging and motivational entry points into academic study for the many students who do not thrive in the structures and cultures of our schools today.

Ellen Winner, Principal Investigator of REAP, Reviewing Education and the Arts Project, is Professor of Psychology at Boston College and Senior Research Associate at Project Zero, Harvard Graduate School of Education. She is the author of numerous scholarly articles as well as three books, Invented Worlds: The Psychology of the Arts, published in 1982, The Point of Words: Children's Understanding of Metaphor and Irony, published in 1988, and Gifted Children: Myths and Realities, published in 1996. Her work focuses on the developmental psychology of the arts.

Lois Hetland, Project Manager for REAP, Reviewing Education and the Arts Project, was a teacher at the elementary and middle school levels before enrolling in the doctoral program at the Harvard Graduate School of Education. She is a researcher at Harvard Project Zero, directs Project Zero's Summer Institutes for Educators, and consults on professional and curriculum development.

Kristin Burger recently received her masters degree concentrating in Arts in Education from the Harvard Graduate School of Education. She has been a Research Assistant at Harvard Project Zero and is entering a teaching apprenticeship program in September 2000.

Ron Butzlaff is Assistant Professor of Psychology at University of Oakland, in Michigan. He received his doctorate from Harvard University where he studied with Robert Rosenthal and learned the art of meta-analysis. In addition to authoring one of the REAP articles, he served as the statistical consultant to the research team whose work is represented here.

Monica Cooper holds a doctoral degree in neurobiology and was a member of the REAP research team. She is continuing her studies in psychology at the Harvard Graduate School of Education.

Mia Keinanen, has worked professionally as a dancer/choreographer in the United States and Europe and is currently undertaking doctoral studies in the area of art and cognition at the Harvard Graduate School of Education.

Erik Moga is a graduate of Interlochen Arts Academy has served as an intern at Harvard Project Zero, Harvard Graduate School of Education. He is currently studying music performance and creativity at the University of Michigan.

Ann Podlozny received her masters from the Harvard Graduate School of Education and served as a research assistant at Project Zero, where she worked with arts organizations in evaluation their programs. She has also worked as an actor and arts educator with Shakespeare and Company and currently works in the movie industry.

Kathryn Vaughn received a doctorate in Musicology from University of California at Los Angeles and was Associate Professor of Music at the Berklee College of Music and Principal Research Scientist at the Media Lab at Massachusetts Institute of Technology before undertaking a second doctorate in educational research measurement and evaluation at Boston College. Her writings have appeared in Music Perception and Ethnomusicology.

Research Assistants
Lisa French, Kimberlee Garris, Nandita Ghosh, Jessica Gordon, Joanna Holtzman, Jenny Martin, Elisabeth Moriarty-Ambrozaitis, Brian Moss, Melissa Mueller, Leah Okimoto, Nina Salzman, and Daniel Schneider served as research assistants during various periods of this project.

Howard Gardner, John H. and Elisabeth A. Hobbs Professor of Cognition and Education, Harvard Graduate School of Education

Richard Light, Professor, Harvard Graduate School of Education and John F. Kennedy School of Government

David N. Perkins, Professor, Harvard Graduate School of Education

Robert Rosenthal, Edgar Pierce Professor of Psychology, Emeritus, Harvard University, Distinguished Professor, University of California, Riverside

Judith D. Singer, Academic Dean and Professor of Education, Harvard Graduate School of Education

The Bauman Foundation supported the research reported here. We thank John Landrum Bryant, Co-director of the Bauman Foundation, for his vision and generous support.

Copyright 2000 by the President and Fellows of Harvard College.

Back to The Arts in the NEWS!

Top of Page