Can we generalize about the learning style characteristics of high academic achievers?

In 1980 Dunn and Price used their Learning Style Inventory (Dunn, Dunn & Price, 1975) to investigate differences between the learning style preferences of high academic achieving students and the preferences expressed by same-age students with average or below average academic achievement. The purpose of the study described in this article was to determine if and how the learning style preferences of a different group of high academic achieving students, inventoried at a later date, but with the same instrument, differed from those identified in the original study. A discriminant function analysis analyzed the learning styles data obtained from 500 students in grades 4 - 8. While significant differences (p [is less than] .001) in the preferences distinguished between the average and above average achieving students groups, there was minimal overlap with the preferences identified in original investigation. The authors conclude that the differences within an achievement group may be as great as between groups, and that it is improper to prescribe instructional methods or categorize groups of learners by presuming that they have similar style preferences.

During the last 25 years, educators have collected and analyzed a great deal of information about students learning styles. Researchers such as Barbe and Swassing (1979), Dunn, Dunn and Price (1975), Gregorc and Ward (1977), Hill (1971), Hunt (1981), Kolb (1978), McCarthy (1980), Myers and Myers (1980), Renzulli and Smith (1978), and Schmeck (1977), developed instructional and theoretical models to explain differences in how students acquire and process information. Although the constructs that underlie these models vary (Ferrell, 1983), each researcher attempted to develop a system that allows teachers to identify, formally or informally, the special learning characteristics of these students and to modify instructional practices accordingly to improve the effectiveness of instruction and to increase academic achievement.

Although a number of studies suggest that there may be merit in addressing students' learning styles as a technique for improving achievement and attitude toward school (e. g., Cafferty, 1980; Carbo & Hodges 1988; Domino, 1979; Doyle & Rutherford, 1984; Lynch, 1981; Shands & Brunner, 1989; Shea, 1983), some educators remain skeptical about the value of such information (Kavale & Forness, 1987; Pask, 1988). Weak research designs, the lack of randomly selected samples, and a "premature rush into print and marketing with very early and preliminary indications of factor loadings based on one data set" (Curry, 1990, p. 51) also influence the degree to which findings can be generalized to broad student populations. In addition, the questionable reliability and validity of some learning style instruments (Stahl, 1988), and unanswered questions about the malleability or durability of students' learning style preferences (Davidman, 1981) still plague the field.

Despite these problems and a host of competing models and nomenclature within the field of learning style research, educators agree that attention to the individual learning style of a given student holds promise as a technique for improving school performance. This technique, however, may be an inappropriate and ineffective teaching tool if we assume that instruments designed to diagnose students' individual learning style preferences predict the learning styles of specific student groups such as Native American, reading-disabled, middle school, or high achieving students.

Learning Styles Research in Gifted Education 

In the past, investigators conducted a number of studies to address possible learning style differences between groups of students. Researchers in the field of gifted education investigated the degree to which a consistent pattern of learning style preferences distinguish high achieving students from the general population. Most of these studies used one of three diagnostic instruments (Dunn, Dunn & Price, 1975; Barbe & Swassing, 1979; Renzulli & Smith, 1978) appropriate for students in grades K-12.

Stewart (1981) administered Renzulli and Smith's Learning Style Inventory to a sample of general education and high achieving students in grades 4, 5, and 6, and found academically able students exhibited preferences for independent study, discussion, and lecture; factors significantly different from the preferences of the general education students. Wasson (1980) used the same instrument and found that the grade 4, 5, and 6 gifted achieving and underachieving students differed in their preferences for teaching games, independent study, peer teaching and programmed instruction. Although both of these studies used the same grade level subjects and instruments, the learning styles identified as significant in the studies differed in three out of four factors.

Kreitner (1981) and Kirchoff (1980) administered the Swassing-Barbe Modality Index to assess modality preference (e.g. visual, auditory, tactile kinesthetic) with two different groups of high achieving subjects. Kirchoff concluded that modality strength is not a fixed characteristic. Instead, it usually changes over time, with high academic achievers demonstrating an integration of modalities at an earlier age. Surprisingly enough, Kreitner also found that musically talented adolescents did not demonstrate a preference to learn auditorally, though they did indicate strong perceptual preferences. In addition, his subjects preferred not to learn through lecture. Again, neither researcher found similar learning styles when comparing two different groups of high achieving students.

Researchers also used the Dunn, Dunn and Price Learning Style Inventory (LSI) (1975, 1979, 1985, 1987, 1989) to examine possible differences between the learning style preferences of academically able students and the general student population. Dunn and Price (1980) conducted a study that involved gifted education students. They compared the learning style preferences of high achieving grade 4 - 8 students with counterparts in the general population. They concluded that high achieving students perceived themselves to be more persistent and had stronger preferences for tactile and kinesthetic learning than students in the general population. In addition, they found that high achieving students preferred a formal classroom design and less structure. These students also saw themselves as less responsible than their peers.

Griggs and Price also administered the 1975 version of the Dunn, Dunn and Price LSI Inventory to students who were identified for participation in a gifted education program and to students who were not identified. The LSI scores for the grades 7 - 9 academically able students indicated that these students perceived themselves to be less teacher-motivated. They preferred quiet and a less formal design, perceived themselves to be more persistent, preferred learning through visual, tactile or kinesthetic means rather than through auditory means, and preferred to learn alone rather than with adult or peers.

Ewing and Yong's (1992) study used a 1987 version of the LSI with 125 gifted African-American, Mexican-American and Chinese-American students in grades 6 - 8. Preferences for noise, light, visual modality, studying in the afternoon and perceptions of their own persistence significantly differed for students in the three groups.

The Need for Additional Studies

Although some of the variables associated with the learning style characteristics of high achieving students seem consistent from study to study, no clearly defined pattern has emerged. To some degree, the differences in these studies' findings are related to the differences in the constructs measured by the three instruments; for example, the Barbe and Swassing Inventory analyzes modality preferences, while the Renzulli inventory assesses learner preferences for a variety of instructional strategies. Yet, even within the groups of studies that used the same instrument, conflicting sets of significant variables emerge from the findings.

Some of these differences are explained by the fact that the grade levels of the students tested were not identical in all studies, nor were the social, cultural, experiential or economic backgrounds of the students. The results of other learning styles studies, outside the field of gifted education, suggest that preference differences between students may be related to gender, learning disabilities, reading achievement, culture, age, and a host of other factors. Viewed together, the studies in general education and the studies in gifted education bring to question whether differences in learning style preferences are just as great within the gifted education population as they are between gifted education students and students in the general population.

Despite the issues raised when generalizing from these studies, some educators suggest that high achieving students do share similar learning style preferences (Barbe & Milone, 1982; Dunn, 1993; Griggs & Price, 1980; Ricca, 1984; Ross & Wright, 1987), and that these preferences may account for differences in their academic achievement. However, before educators can generalize from the work of these researchers to the population of academically able students at large, one must first verify that the variables identified in the above studies apply to additional, but similar groups of high achieving students.

Subsequent studies are warranted in light of the many recommendations, presented in print and in oral presentations, that have been made over the years regarding the learning style preferences of all gifted education students. However, in reviewing more than a dozen studies on the learning style preferences of gifted education students, no such replication or extension studies. For these reasons, we attempted to duplicate, as closely as possible, the sample, the instrument, and the data analysis techniques used in one of these studies; the 1980 research study by Dunn and Price.

A Description of the Original Study

The Dunn and Price study involved a total of 109 high achieving students and 160 randomly selected students from the general population of grades 4 - 8, in eight different schools in the Eastern portion of the United States. Researchers in the original study classified students as gifted if they possessed an academic aptitude score of 130 or above, or an academic aptitude score between 120 and 129 with scores in the 95th percentile or above on the math and/or reading subtests of standardized norm-referenced achievement tests. The researchers used the Otis-Lennon Academic Aptitude Test to assess academic aptitude, and the 1975 version of the Dunn, Dunn and Price Learning Style Inventory to assess learning style preferences.

Step-wise discriminant analysis measured differences in learning style preferences for the gifted and non-gifted education students in their study, with the Learning Style Inventory variables serving as predictor variables. The statistical analysis determined which of the Learning Style Inventory variables discriminated significantly between the two groups of students, minimizing Wilks' lambda coefficient. The F for inclusion and deletion was set at 1.0, with a tolerance level of p [is less than] .001. Bartlett's chi square (52.97, 9 df, p [is less than] .001) was significant when Wilks' lambda of .817 was tested for one discriminant function.

The discriminant analysis identified six variables that significantly differed between the two groups. Gifted students in the Dunn and Price study preferred to learn through their tactile and kinesthetic senses, and indicated less of a preference than the non-gifted students for using their auditory sense of learning. The non-gifted students preferred an informal design, required structure, saw themselves as more responsible, but not as persistent, and preferred to use their tactile and kinesthetic senses less and their auditory senses to a greater extent than the gifted students.

On the basis of these six variables it was possible to predict membership in the gifted education group with 53 percent accuracy and predict with 81 percent accuracy membership in the non-gifted group. The authors concluded that gifted education students do possess identifiable learning style preferences that are different from the preferences of students in the general population. The significant variables that differed for the two groups included style preferences related to independence, persistence, perception and motivation. The authors also suggested that the LSI might be used as an alternative identification instrument.

Extending the Original Study

In reviewing the Dunn and Price findings, the question arose about whether these results would be similar with another sample of gifted education students. The inferred hypothesis was that if these findings were consistent from sample to sample, then recommendations for modifying identification or instructional strategies within gifted education programs might be justified. For this reason, an extension study was conducted in order to investigate the following questions:

* Do high academic achieving students differ from average and low achieving students with respect to their learning style preferences?

* If statistically significant discriminating variables do exist, are they the same as the variables identified in Dunn and Price's study?

Limitations The sample for this extension study involved 500 students, 54 percent more students than the sample in Dunn and Price's original study. The ratio of high achieving students to students in the general population was 19.8 to 100, as compared to Dunn and Price's 40 to 100 ratio. The procedures used to identify students for the gifted and the non-gifted groups in both studies involved similar, but not identical, criteria. Although the students in both samples were enrolled in grades 4 - 8 in the public schools, they live in different regions of the country. Due to the constraints common to most research studies conducted in the public schools, a random selection of gifted students was not possible in either study.

Subjects The sample involved in the extension study included 99 high achieving students and 401 students in the general population of grades 4 - 8. These students were enrolled in the elementary, junior high, and middle schools of nine public school districts in one Midwestern county of the United States. Communities within the county included metropolitan, agricultural, industrial, and suburban settings. Students categorized as gifted for the purposes of this study had scores at or above the 95th percentile on the Comprehensive Test of Basic Skills or the Iowa Test of Basic Skills or academic aptitude quotients of 125 or above on the California Test of Academic Aptitude. All parents and students received invitations to participate in the study and did so in order to acquire more information about the students' learning style preferences. Three local classroom teachers, two administrators and a gifted education coordinator received training in the administration of the 1975 version of Dunn, Dunn and Price's Learning Styles Inventory prior to using the instrument with students.

Instrumentation Although a 1989 edition of the inventory is now available, the goal was to eliminate any problems that might arise from analyzing data from two different versions of the instrument. For this reason, researchers searched for and used student data from 1981; data that used the same 1975 version of the LSI as Dunn and Price's 1980 study. This version of the LSI assessed grades 3-12 students' beliefs about 34 variables within 13 categories related to learning conditions:

* noise level - quiet or sound;

* light - low or bright;

* temperature - cool or warm;

* design - informal or formal;

* motivation - unmotivated or motivated;

* persistence - impersistent or persistent;

* responsibility - irresponsible or responsible;

* structure - more structure or less structure;

* sociological - learning with peers, alone, in pairs, in a team, with an authority figure, or varied;

* perceptual - visual, auditory, tactile, kinesthetic or combined;

* intake - requires food or does not require food;

* time - morning, late morning, afternoon, or evening; and

* mobility - prefers mobility or no mobility.

Students completed the inventory in a 30 - 40 minute session and responded to each of the 100 items in the inventory with a dichotomous, true or false, rating. Students rarely exhibit a strong preference for all variables. Instead, the inventory analysis usually identifies the 6 - 14 variables of greatest influence to the learning process. The inventory analysis identifies students' extreme preferences. Scores range from 20 - 80 with a mean of 50 and a standard deviation of 10. Scores of 60 or higher indicate a strong preference for a given factor, while scores of 40 or below indicate a negative preference.

Principal components analysis with the 1975 instrument identified 32 factors, not 34, each with an eigenvalue greater than 1.0, which collectively explained 62 percent of the variance. The authors report reliability coefficients on the internal consistency for the various scales that range from .40 to .84. Subsequent editions of the instrument used a Likert scale and fewer variables to increase reliability coefficients.

Data Analysis

Researchers used SPSS step-wise discriminant analysis to find the linear combination of variables that best discriminated between the high achieving and the general sample of students using the 34 variables on the 1975 Learning Style Inventory. The analysis identified which factors on the Learning Style Inventory significantly discriminated between the two groups of students minimizing Wilks' lambda coefficient. The F for inclusion and deletion was 1.0, and the tolerance level was p [is less than] .001. Prior probability for each group was set at .50. Bartlett's chi square (55.031, 14 df, p [is less than] .0001) was significant when Wilks' lambda of .894 was tested for one discriminant function for high achieving students versus students in the general population.

Results

In addressing the first research question, "Do high academic achieving students differ from average and low achieving students with respect to their learning style preferences?", it was found that of the 34 predictor variables assessed in this study, 14 variables discriminated significantly between the subjects in the high achieving group and students in the general population. Table 1 summarizes the order in which these variables entered the discriminant equation, the F to enter, and Wilks' lambda. The standardized discriminant function coefficients are indicated in Table 2.

Summary of the Learning Style Variables
That Entered the Discriminant Equation

        Step           Variable               F to      Wilks
       number          entered                enter     lambda

         1             Little structure       11.48     .966
         2             Not responsible         4.87     .951
         3             Late morning            4.51     .941
         4             Needs quiet             8.31     .934
         5             Sound acceptable        4.28     .928
         6             Adult motivated         2.57     .923
         7             Persistent              2.66     .918
         8             Bright light            4.50     .914
         9             Informal design         2.27     .909
         10            Auditory preference     2.59     .905
         11            Needs little mobility   2.28     .902
         12            Low light               1.79     .899
         13            Learn alone             1.55     .896
         14            Afternoon               1.12     .894

       Step               Significance
       number               level

         1                .0001
         2                .0001
         3                .0001
         4                .0001
         5                .0001
         6                .0001
         7                .0001
         8                .0001
         9                .0001
         10               .0001
         11               .0001
         12               .0001
         13               .0001
         14               .0001

Table 1

Standardized Discriminant Function
Coefficients for the Learning Style Inventory

Learning Style Inventory Factor Name       Coefficient

       Quiet                                 -.595
  Little Structure                            .477
  Sound Acceptable                           -.437
    Bright Light                              .430
    Late Morning                             -.336
Not Very Responsible                          .324
     Low Light                                .272
     Persistent                               .233
      Auditory                                .231
  Adult Motivated                             .228
    No Mobility                               .219
  Informal Design                             .213
       Alone                                 -.182
     Afternoon                               -.149

Table 2

The classification procedure correctly classified 68 percent of the students. Based on the 14 variables that significantly discriminated between the two groups, 62 percent of the high ability students and 70 percent of the students in the general population were correctly classified. This yields a discriminant power above chance of 12 and 20 percent, respectively. Table 3 summarizes the classification procedures for these students. It must be noted however, that, as in the Dunn and Price study, the jackknife procedure for classification (Tabachnick and Fidell, 1983) was not used and that bias did enter this classification procedure. The square of the canonical correlation of .326 for the function equation demonstrated that 10.60 percent of the variance between the two groups can be accounted for by the significant variables in the discriminant equation.

Percent of Students Properly Classified
by Group Using the Function Equation

                                     Percent of   Percent of
Group           n    Hits   Misses   hits using   hits beyond
                                       DFA          chance
High
achieving       99     61       38    61.60%      11.60%
students

General
population     401    279      122    69.60%      19.60%
students

Table 3

An analysis of the significant variables and the group means showed that the high achieving group preferred little structure and an informal design, accepted sound, low mobility and bright light in the learning environment, and perceived themselves to be more persistent than their classmates in the general population. The students in the general population preferred a quiet learning environment with low light. They perceived themselves as less responsible, more adult motivated, and preferred to learn alone through auditory means in the late morning or early afternoon. Structure and sound preferences accounted for the greatest differences between the two groups.

Discussion

Although both studies produced a statistically significant function equation that successfully discriminated between the learning style preferences of high achieving students and those in the general student sample, neither study identified the same set of variables. The only variables consistent for the high achieving students in both studies were the personality trait of persistence and a preference for little structure. Students in the general population of both groups showed preferences for learning through auditory means.

More disturbing, however, is the fact that the two studies produced contradictory results in two crucial categories. Dunn and Price's high achieving students preferred a formal classroom design, while the students in this study preferred an informal design. Dunn and Price's high achieving students saw themselves as less responsible than their peers in the general population, while this study suggested that the students in the general population, not the high achieving students, saw themselves as less responsible. Seven variables; light, mobility, auditory and tactile/kinesthetic learning, time preferences, adult motivation and the desire to learn alone, were identified in one study, but not the other.

The discriminant function equation correctly classified the high achieving students in Dunn and Price s study at a level only 3 percent above chance. The high achieving students in this study were correctly classified, above chance, in only 12 percent of the cases. Even though students in the general population constituted the majority of the research population, and chance classification alone should have worked in their favor, only 21 percent in Dunn and Price's study and 20 percent in this study were correctly classified. In all likelihood this occurred because the magnitude of the difference between the two groups was insufficient to facilitate predictions much greater than chance.

On the basis of these findings, the low percentage of shared variance, and the fact that the discriminant equation in this study yielded classifications only 18 percent above chance prediction, it is difficult to accept the idea that the population of academically able students share common learning style preferences. Certainly the current learning environment of the students tested, the quality of their gifted education program, their specific educational experiences, attitudes toward school, and the demographic makeup of their community accounts for some of the discrepancies noted above. These facts weaken the argument that most high achieving students share common learning style preferences.

The discriminant function analysis yielded interesting information, but it should not be taken as clear and irrefutable evidence that consistent differences in learning style preferences do or do not exist between achievement groups. Plainly, we are on shaky ground if we continue to assume that certain learning style preferences are associated with achievement test score levels. In the wrong hands, this conclusion might be construed as evidence in favor of a new identification technique, reminiscent of the characteristics checklists that were popular in gifted education just a few years ago. The conflicting data between subgroups suggests at least two possibilities. Either the original instrument was flawed, or, individual differences between students accounts for more variance in style preferences than group differences.

Although the 1975 LSI was subsequently revised to improve the factor structure and the reliability of the instrument, no replication or extension studies have been found for either the older or newer versions. At the very least, the findings from this extension study strongly suggest the need to conduct additional studies with the newer instruments. Until such time, researchers should not attempt to synthesize the results of learning styles studies that used more than one version of the LSI instrument.

The evidence suggests that the learning style preference differences within an academic achievement group may be as great as the differences between the groups (Barbe & Milone, 1982). All style preferences may be equally appropriate (Fischer & Fischer, 1979), and care must be taken to refrain from placing value judgments on one preference over another. Educators must recognize the emerging nature of learning style preferences (Hunt, 1981) and come to grips with the seemingly topical and temporal nature of such preferences. Students change. They grow and adapt, and hopefully, become increasingly adept at functioning with a variety of styles.

The real issue involves educators' ability to modify the learning environment to deal appropriately with individual preferences. Some researchers believe teachers should consistently teach to a student's preferred learning style. Others believe that such modifications should occur primarily during initial instruction or times of learning difficulty (Barbe & Swassing, 1979). A third point of view advocates teaching all students in all style variables in an attempt to foster independence (Hunt, 1981; McCarthy, 1980). Perhaps the truth lies somewhere between the extremes of this continuum.

Learning style preferences may or may not account for part of what identifies a student as academically superior. However, the interaction of style preferences and the learning environment (Ricca, 1984) precludes a unilateral approach to instructional modification (Stewart, 1982). We recognize the fact that learners are different. We also believe that in general, it is helpful to recognize and accommodate these differences. Based on our findings however, we have concluded that it would not be prudent to prescribe instructional methods or categorize groups of learners by presuming that they have similar style preferences on the basis of singular research studies.

After additional replication or extension studies, we may find that learning style inventories should be used as they were originally intended; as informative diagnostic instruments to measure the learning style preferences of an individual student. In other words, the instrument should be used to take a "snapshot" of an individual in a particular situation, at a specific point in time. It should not be used to take a group portrait. Within gifted education, this information can be used during curriculum compacting, content acceleration, or during self-directed student investigations or research; not as an identification device, nor as a blanket recommendation to view or to teach all students with similar achievement levels in the same manner.

REFERENCES

Barbe, W. B., & Swassing, R. H. (1979). The Swassing-Barbe Modality Index. Columbus, OH: Zaner-Bloser.

Barbe, W. B., & Milone, M. N. (1982, January - February). Modality characteristics of gifted children. G/C/T, 5, 2-5.

Cafferty, E. (1980). An analysis of student performance based upon the degree of match between the educational cognitive style of the teachers and the educational cognitive style of the students. An unpublished doctoral dissertation, University of Nebraska.

Carbo, M. and Hodges, H. (1988). Learning styles strategies can help students at risk. Teaching Exceptional Children, 20, 55-58.

Curry, L. (1990). A critique of the research on learning styles. Educational Leadership, 48, 50-55.

Davidman, L. (1981). Learning style: The myth, the panacea, the wisdom. Phi Delta Kappan, 62, 641-645.

Domino, G. (1970). Interactive effects of achievement orientation and teaching style on academic achievement. ACT Research Report, 39, 1-9.

Doyle, W. & Rutherford, B. (1984). Classroom research on matching learning and teaching styles. Theory into Practice, 2, 20-25.

Dunn, R. (1993). Learning styles of the multiculturally diverse. Emergency Librarian, 20 (4), 24-32.

Dunn, R., Dunn, K., & Price, G. E. (1975). Learning Style Inventory. Lawrence, KS: Price Systems.

Dunn, R., & Price, G. E. (1980). The learning style characteristics of gifted students. Gifted Child Quarterly, 24, 33-36.

Ewing. N.J., & Yong, F.L. (1992). A comparative study of the learning style preferences among gifted African-American, Mexican-American and American Born Chinese middle grade students. Roeper Review, 14, 120-23.

Ferrell, B.G. (1983). A factor analytic comparison of four learning style instruments. Journal of Educational Psychology, 75, 33-39.

Fischer, B. B., & Fischer, L. (1979). Styles in teaching and learning. Educational Leadership, 36, 245-254.

Gregorc, A. F., & Ward, H. B. (1977, February). A new definition for individual. NASSP Bulletin, 2.

Griggs, S. A., & Price, G. E. (1980). A comparison between the learning styles of gifted versus average suburban junior high school students. Roeper Review, 3, 7-9.

Hill, J. (1971). Personalized education programs utilizing cognitive style mapping. Bloomfield Hills, MI: Oakland Community College.

Hunt, D. E. (1981). Learning style and the interdependence of practice and theory. Phi Delta Kappan, 62,647.

Kavale, K.A., & Forness, S.R. (1987). Substance over style: Assessing the efficacy of modality testing and teaching. Exceptional Children, 54, 228-239.

Kirchoff, S. (1980). Modality strengths of gifted students. An unpublished doctoral dissertation, Washington State University.

Kolb, D. A. (1978). Learning style inventory technical manual. Boston: McBer & Co.

Kreitner, K. R. (1981). Modality strengths and learning styles of musically talented high school students. An unpublished master's thesis, The Ohio State University.

Lynch, P. K. (1981). An analysis of the relationships among academic achievement, attendance and the individual learning style time preferences of eleventh and twelfth grade students identified as initial or chronic truants in a suburban New York school district. An unpublished doctoral dissertation, St. John's University.

McCarthy, B. (1980). The 4MAT system: Teaching to learning styles with right/left mode techniques. Barrington, IL: Excel, Inc.

Myers, I., & Myers, P. Gifts differing. Palo Alto, CA: Consulting Psychologists Press.

Pask, G. (1988). Learning strategies, teaching strategies and conceptual or learning style. In R. R. Schmeck, (Ed.), Learning strategies and learning styles, (pp. 83-100). New York: Plenum Press.

Renzulli, J. S., & Smith, L. H. (1978). The learning styles inventory: A measure of student preference for instructional techniques. Mansfield Center, CT: Creative Learning Press.

Ricca, J. (1984). Learning styles and preferred instructional strategies of gilled students. Gifted Child Quarterly, 28, 121-126.

Ross, E.P., & Wright, J. (1987). Matching teaching strategies to the learning styles of gifted readers. Reading Horizons, 28, 49-56.

Shands, R., & Brunner, C. (1989). Providing success through a powerful combination: Mastery learning and learning styles. Perceptions, 25, 6-10.

Schmeck, R. R. (1977). Development of a self-report inventory for assessing individual differences in learning process. Applied Psychological Measurement, 1, 413-431.

Shea, T. C. (1983). An investigation of the relationship among preferences for the learning style element of design, selected instructional environments and reading test achievement of ninth grade students to improve administrative determinations concerning effective educational facilities. An unpublished doctoral dissertation, St. John's University.

Stahl, S.A. (1988). Evidence to support matching reading styles and initial reading methods? A reply to Carbo. Phi Delta Kappan, 4, 317-22.

Stewart, E. D. (1981). Learning styles among gifted/talented students: Instructional technique preferences. Exceptional Children, 48, 134-138.

Stewart, E. D. (1982). Myth: One program, indivisible for all. Gifted Child Quarterly, 26, 27-29.

Tabachnick, B. G., & Fidell, L. S. (1983). Using multivariate statistics. New York: Harper and Row.

Wasson, F. R. (1980). A comparative analysis of learning styles and personality characteristics of achieving and underachieving gifted elementary students. An unpublished doctoral dissertation, Florida State University.

Deborah E. Burns is an associate professor of Educational Psychology and director of the Three Summer Graduate Program in Gifted Education at the University of Connecticut. Scott E. Johnson is a principal of a science, technology and global studies elementary magnet school that serves East Harford and Glastonbury, Connecticut. Robert K. Gable is a professor of educational psychology and associate director in the Bureau of Educational Research at the University of Connecticut.

Manuscript submitted January, 1996. Revision accepted October, 1997.