Requests for copies should beaddressed to:TGroenewald,Technikon SA, Private Bag X6, Florida,1710 THE CONSTRUCTION AND EVALUATION OFA NORMATIVE LEARNING STYLE PREFERENCE QUESTIONNAIRE MJ VILJOEN JM SCHEPERS K VAN ZYL Department of Human Resource Management Rand Afrikaans University ABSTRACT Various authors have indicated the need for and value of identifying the learning style preferences of individual learners. Similar needs have been voiced in the South African context.The focal point of this study was the deve- lopment of a normative instrument for predicting the preferred learning styles of individuals. Secondary aims were to determine whether there are di¡erences between groups formed on the basis of gender, academic quali¢cations and functional disciplines as far as their learning style preferences are concerned. Based on a review of the literature and an existing questionnaire, namely the Learning Style Inventory (LSI 85), the Learning Style Preference Ques- tionnaire (LSPQ) consisting of136 items was developed and administered to respondents (N = 542) in a large orga- nisation.The LSPQ was subjected to a principal factor analysis and six factors were obtained.The six factors were rotated to simple structure by means of the Direct Oblimin procedure. The matrix of intercorrelations of the six factors was subjected to a second-order factor analysis and yielded a single factor. Six scales were constructed corres- ponding to the six factors.These scales were subjected to item analysis and yielded reliabilitycoe⁄cients that ranged from 0,809 to 0,939 according to Cronbach’s coe⁄cient alpha.The implications of the ¢ndings are discussed. OPSOMMING Verskeie outeurs het na die behoefte aan asook die waarde van identi¢kasie van leerstylvoorkeure van individuele leerders verwys. Soortgelyke behoeftes is ook in Suid-Afrikaanse verband geopper. Die fokus van hierdie studie was die ontwikkeling van’n normatiewe instrument om die leerstylvoorkeure van individue te meet. Sekonde“ re doel- witte was om te bepaal of daar verskille tussen groepe is wat saamgestel is op grond van geslag, akademiese kwali- ¢kasies en funksionele dissiplines wat hul leerstylvoorkeure betref. Gegrond op ’n oorsig van die literatuur en ’n bestaande vraelys, te wete die‘‘Learning Style Inventory’’ (LSI 85), is die‘‘Learning Style Preference Questionnaire‘‘ (LSPQ), bestaande uit136 items, gekonstrueer en op 542 respondente in’n groot organisasie toegepas. Die LSPQ is aan’n hoo¡aktorontleding onderwerp en ses faktore is verkry. Die ses faktore is deur middel van die Direct Obli- min-prosedure na eenvoudige struktuur geroteer. Die interkorrelasiematriks van die ses faktore is aan ’n tweede- ordefaktorontleding onderwerp en het ’n enkelfaktor opgelewer. Daarna is ses skale gekonstrueer wat met die ses faktore ooreenstem. Hierdie skale is aan itemontleding onderwerp, en het betroubaarheidskoe« ⁄siente wat wissel van 0,809 tot 0,939, volgens Cronbach se alfa-koe« ⁄sient, opgelewer. Die implikasies van die bevindinge is bespreek. There has been avery strong interest by business in the concept of learning organisations and the capabilities required to build learningorganisations. A learningorganisation is de¢ned as an organisation continuously transforming through the acquisiti on, processing and dissemination of knowledge about mar- kets, products, technology, and business processes (Ellinger, Watkins & Bostrom, 1999; Senge, Roberts, Ross, Smith, & Kleiner,1994). Learning is seen to be a key to survival in a ra- pidly changing world, requiring organisations to be faster and more e¡ective in the way they learn (Prokesch, 1997; Senge, Kleiner, Roberts, Ross, Roth, & Smith, 1999). A learning or- ganisation needs to re£ect a learning approach implicitly or explicitly in its vision, strategy and values to ensure that all employees have a shared focus (Abernathy,1999; Pedler & As- pinwall,1998; Senge et al.,1999) In practice this requires an un- derstanding of how the organisation, i.e. the individuals and teams in the organisation acquire, share and utilise knowledge (Dibella, Nevis & Gould,1996). A true learning organisation intentionally makes use of indi- vidual and team learning (Dubois, 1993; Pedler & Aspinwall, 1998; Senge et al., 1999). What really matters is an under- standing of how individuals learn, what their preferred learn- ing styles are and how their preferred style will contribute to their own and team learning (Pedler & Aspinwall,1998; Senge et al.,1994). Senge et al. (1994), supported by Stuart (1992) and Newstrom and Legnick-Hall (1991) suggested the use of a diagnostic instrument such as the Learning Style Inventory (LSI) of Kolb to help individuals gain an understanding of their preferred learning styles. Kolb (1976, 1984) originated and pioneered the experiential learning theory (ELT), resulting in his learning style model and the LSI. Many other researchers such as Honey and Mum- ford (1982) and McCarthy (1987) were inspired by the work of Kolb, which made substantial contributions to the ¢eld of learning styles. The present study relies heavily on the work of Kolb because of its distinct contribution and frequent cita- tion in the literature. Experiential learning and Kolb’s model of learning styles In the development of the ELT Kolb (1984) maintained that learning is a process involving the resolution of dialectical con£icts between opposing modes of dealing with the world, either through action and re£ection or concrete or abstract in- terventions. He accredited his conceptualisation of the learn- ing process to Jung’s concept of style and to Lewin’s theoretical model which describes individual di¡erences in learning behaviour (Kolb,1984; Loo,1999;Verses, Sims & Loc- klear, 1991).Kolb described learning as cyclical. He suggested two primary dimensions to the learning process, namely con- crete experience (CE) as the polar opposite of abstract concep- tualisation (AC), and re£ective observation (RO) as the polar opposite of active experimentation (AE).These polar extremes are integrated into a four-stage cycle of learning ranging from perceiving and experiencing events (CE), to re£ecting on ex- periences from di¡erent perspectives (RO), constructing theories which integrate observation (AC), actively using theories to make decisions, and solving problems (AE) (Kolb, 1984; Loo,1999;Willcoxson & Prosser,1996). Based on the ELTand the four-stage model of learning, the LSI was developed by Kolb to assess an individual’s learning style preferences (Kolb, 1995; Verses et al., 1991). The LSI, a self-administered questionnaire, consists of 12 statements followed by four word endings corresponding to the four learning orientations (Geiger, Boyle & Pinto, 1992; Kolb, 1995). Respondents are required to rank the four word en- dings for each set of 12 statements according to how well Journal of Industrial Psychology, 2001, 27(3), 51-60 Tydskrif vir Bedryfsielkunde, 2001, 27(3), 51-60 51 the word endings characterise their learning orientation (Allison & Hayes,1988; Kolb,1995).This methodology ren- ders the LSI (85) as an ipsative instrument. Kolb de¢ned the four learning styles according to the learner’s preference for a particular phase of the learning cycle, namely converger emphasises abstract conceptualisation and active experimentation as the dominant learning mode; diverger emphasises concrete experience and re£ective observation as the dominant learning mode; assimilator ^ emphasises abstract conceptualisation and re£ective observation as the dominant learning mode; and accommodator ^ emphasises concrete ex- perience and active experimentation as the dominant learning mode (Geiger et al.,1992; Kolb,1984). Application of the LSI (85) The LSI (85) has been used to determine learning style prefe- rences in cultural, social and cross-cultural studies (Chi-Ching & Moi, 1994; Hong & Suh, 1995), gender studies (Hickson, 1996; Wilson, 1996), academic quali¢cation studies (Kolb, 1984) and functional discipline studies (Kolb, 1984; Smedley, 1987;VanWyk,1992) to mention but a few. For the purpose of this study it was decided to focus on the learning style pre- ferences of three speci¢c subgroups of the sample, namely ge- nder groups, academic groups (academic quali¢cations of individuals) and functional disciplines (work areas of em- ployees in an organisation). Learning styles and gender Several studies have been conducted to determine whether the- re are learning style di¡erences between men and women. Se- veriens and Ten Dam (1997, p.80) found small but consistent gender di¡erences in respect of the LSI (85). Men showed a stronger preference for the abstract conceptualisation learning mode than women. Men also seemed to be more interested in academic quali¢cations and their value. Women, by contrast, were more interested in the content and the value of learning. In a study byWilson (1996) the variance of preferred learning styles owing to gender, race and study course was examined. From the results it was evident that African-American women showed a much lower preference for active learning than men from the same group. Hickson (1996, p.65), in another study of gender and learning style preferences, found that women were more likely to be visual learners than men and that women preferred quiet learning environments, were more teacher- motivated and more persistent. Learning styles and academic quali¢cations Early educational experiences shape individual learning styles (Kolb, 1984). Although the early years of education are for the most part generic, there is an increasing specialisation that deve- lops in earnest in high school. For those who continue on to ter- tiary institutions, this specialisation develops in greater depth in the undergraduate years (Kolb, 1984). Jonassen and Garabowski (1993) also found that students tend to enter a ¢eld of study that matches their respective learning styles from school. However, should there be no match, the tendency is to change their ¢eld of studyor choose a career outside their ¢eld of study.The reason for this is that the young employee’s job is usually a continuation of his/her quali¢cation and a re¢nement of his/her specialised skills and knowledge (Kolb,1984). Learning styles and functional disciplines According to Kolb (1984), employees practising di¡erent functio- naldisciplines are inclinedtopreferdi¡erentlearningstyles.This is evident from the variations among their primary tasks, technolo- gy, products, criteria for academic excellence, productivity, lear- ning methods, research methods and methods for recording and portraying knowledge. Over time, owing to exposure to a ¢rm way of doing things and socialisation pressures, a homogeneous disciplinaryculture develops,directing studentstowards a particu- lar learning style (Kolb,1984). Honey and Mumford (1995), Kolb (1984), Slaat, Lodewijks andVan der Saden (1999) andWillcoxson and Prosser (1996) support this notion and found in their respecti- ve studies that chemists, medical students, marketers, researchers, engineers, personnel and ¢nancial people are homogeneous in their learning style preferences. In South Africa, it was found that apprentices from the same trade (De Klerk,1993), black teaching students from the same ¢eld (VanWyk,1992) and managers from the same disciplines (Heymans,1988) tend to be more alike than di¡erent in their approach to learning. A critical look at the ELTand the LSI (85) From the literature it is clear that there are three schools of thought concerning the validity and reliability of the ELTand the LSI (85). At the one extreme researchers such as Freedman and Stumpf (1980) and Reynolds (1997) refer to the ELT and the LSI (85) as failures, displaying insu⁄cient evidence of relia- bility and validity. At the other extreme, Ferrell (1983) and Loo (1999), amongst others, refer to the ELTand LSI (85) as well-es- tablished, with considerable attraction and interest for applica- tion. Somewhere in the middle, researchers such as Geiger et al. (1992) and Merritt and Marshall (1984) support the ELTand LSI (85) with reservations, suggesting some changes to enhance the validity and reliability of the LSI (85). Of course, this is not for- getting Kolb’s own defence of the ELTand LSI (85). In his res- ponse to criticism by Freedman and Stumpf, Kolb (1981) stated that they inappropriately assessed the validityof the ELT by ba- sing their judgement primarily on an analysis of the internal characteristics of the LSI, with only the most super¢cial review of research on the theory. Furthermore, he also stated that their criticism of the reliability and structure of the LSI represented misapplications of statistical assumptions of stability and inde- pendence to a theory based on variability and interdependence. In spite of Kolb’s defence, researchers continued to level three speci¢c criticisms at the psychometric properties of the LSI. Firstly, factoring an ipsative correlation matrix produces an in- valid factor solution. Secondly, the two dimensions of the LSI account for only some 21% of the total variance.Thirdly, test- retest studies display a lack of stability, a ¢nding that con£icts with the ELT’s position that learning styles are a relatively sta- ble and enduring characteristic of the learner (Cornwell & Manfredo, 1994; Verses et al., 1991). Allison and Hayes (1988) also refer to the reliability of the two scales reported by Kolb. They found the reliability of the abstract-concrete dimension and the active-re£ective dimension to be only 0,40 and 0,52 (Cronbach alpha). Bycontrast, Kolb (1995) reports a Cronbach alpha of 0,88 for the abstract-concrete dimension and 0,81 for the active-re£ective dimension of his LSI (85). Merrit and Marshall (1984) also report on the psychometric qualities of the LSI and stress the ipsative nature of the instru- ment, warning of the dangers of using it as a normative instru- ment. In their quest to ¢nd a normative instrument, they redesigned the LSI, using the same word list of the inventory, asking respondents to rate the degree to which each word en- ding was characteristic of their preferred learning style. The four response choices for each word were ‘‘characteristic’’, ‘‘somewhat characteristic’’, ‘‘somewhat uncharacteristic’’ and ‘‘uncharacteristic’’. The results of the study suggested that the normative form of the LSI was consistent with the learning style model proposed by Kolb. The factor structure demon- strated construct validity and a moderate level of concurrent validity was found (Merrit & Marshall,1984). In a similar study Geiger et al. (1993) used a 7-point Likert scale ranging from‘‘very much like me’’to‘’not like me’’and rando- mised the order of the 48 independent learning style state- ments of the LSI in an attempt to create a normative instrument.Their factor analysis did not support the existence of any bipolar dimensions but did support the four separate learning styles. In conclusion, Geiger et al. (1993) were of the opinion that their modi¢ed instrument gave a better assess- ment of the four separate learning abilities than the LSI (85). Problem statement The LSI (85) is an ipsative instrument and should be valued as such and applied in proper context. As an ipsative instrument, it has value when the focus is on a single individual’s relative strengths in terms of his/her learning preferences. Its focus is VILJOEN, SCHEPERS,VAN ZYL52 TABLE 1 BIOGRAPICAL INFORMATION OF THE RESPONDENTS 1. GENDER FREQUENCY PERCENTAGE Male 351 64,8% Female 191 35,2% Total 542 100,0% 2. AGE 24 62 years 3. MARITAL STATUS FREQUENCY PERCENTAGE Single 101 18,6% Married 360 66,4% Divorced 53 9,8% Widowed 2 0,4% Living together 26 4,8% Total 542 100,0% 4. ACADEMIC QUALIFICATIONS FREQUENCY PERCENTAGE Less than GR 12 (std10) 15 2,8% Grade 12 (std 10) 143 26,4% Technikon diploma 145 26,8% Bachelor’s degree 121 22,3% Post-graduate diploma/degree 104 19,2% Other 14 2,6% Total 542 100,0% 5. CULTURAL GROUP FREQUENCY PERCENTAGE Asian/Indian 18 3,3% Black 74 13,7% Coloured 43 7,9% White 403 74,4% Unknown 4 0,7% Total 542 100,0% 6. CURRENT FUNCTIONAL DISCIPLINE IN WORK PLACE FREQUENCY PERCENTAGE Distribution 81 14,9% Finance 80 14,8% Human Resources 44 8,1% Information Technology 19 3,5% Marketing 47 8,7% Production 165 30,4% Risk Control 13 2,4% Sales 68 12,5% Other 20 3,7% Unknown 5 0,9% Total 542 100,0% intra-individual. Applied for this purpose, the LSI (85) is seen to be a useful instrument. However, the LSI (85) should not be used to compare individuals with one another. A normative instrument, i.e. an inter-individually focused instrument, is required for this purpose. On inspection of the original LSI (85) and adapted versions of it, several problems were iden- ti¢ed. It was found that some of the learning concepts in the12 state- ments and four word endings of the LSI (85) are in need of simpli¢cation in certain cases. Multiple learning concepts we- re found tobe grouped in the same word endings, causing pos- sible confusion for the respondent. At face value it seems that there is a broader area of learning to be described than is cover- ed by the LSI (85). The adapted questionnaires merely linked existing versions of the LSI (85) to a seven-point scale with little or no expansion of the existing questionnaire. A norma- tive instrument, determining learning style preferences under local conditions, does not exist for South Africans. This is problematic.There is a de¢nite need to objectively de- termine the preferred ways of learning of individuals and teams in an organisation. This would assist in establishing a learning approach for an organisation and help the organi- sation become a true learning organisation. It would create a better understanding of how individuals and teams prefer to learn, matching learning style preferences and the training needs of individuals and teams with e¡ective training metho- dology. The aims of the study are as follows: * To develop a new normative instrument for determining the learning style preferences of South Africans, in contrast to Kolb’s ipsative instrument.The new instrument should also de- scribe a broader ¢eld of learning than the four modes of Kolb. * To determine whether there are any di¡erences in respect of gender, academic quali¢cations and functional disciplines as far as learning style preferences are concerned. Hypotheses 1. The vectors of means of men and women di¡er statistically signi¢cantly from one another in respect of their learning style preferences. 2. The vectors of means of the various academic groups di¡er statistically signi¢cantly from one another in respect of their learning style preferences. 3. The vectors of means of the various functional disciplines di¡er statistically signi¢cantly from one another in respect of their learning style preferences. METHOD Sample The questionnaire was distributed to the entire population of 2099 members of a large organisation operating in the ¢eld of ‘‘fast-moving consumer goods’’. All the incumbents, from super- visory to executive management level, were included.Thisgroup was selected inviewof the literacy level required to complete the questionnaire. Employees from all over the country were inclu- ded in the target sample. Participants operating in diverse func- tional disciplines, such as production, marketing, sales, engineering and distribution, were included. Atotal of 542 com- pleted questionnaires were returned. No information is available as to why only 25,82% responded. No evidence of bias in the sample could be detected fromthe results.The biographical back- ground information of the participants is given inTable1. The ages of the participants range from 24 to 62 years.The sam- ple indicates a greater response by the male respondents (64,8%), with female respondents accounting for 35,2% of the sample. The majorityof the respondents areWhite (403), with Blacks se- cond (74), Coloureds next (43) and Asians/Indians last (18). The educational level of the sample varies from less than Grade12 to post-graduate quali¢cations.The majorityof the respondents ha- ve Grade12,Technicon diploma and Bachelor’s degree quali¢ca- tions. Participants with quali¢cations less than Grade 12 (N=15) were excluded from the analysis. 53LEARNING STYLE PREFERENCE QUESTIONNAIRE TABLE 2 EIGENVALUES OF UNREDUCED INTERCORRELATION MATRIX ROOT EIGENVALUE PERCENTAGE OF VARIANCE ROOT EIGENVALUE PERCENTAGE OF VARIANCE 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 10,348 2,342 1,852 1,679 1,357 1,187 1,083 1,022 0,979 0,946 0,871 0,803 0,741 0,734 0,707 0,693 0,645 0,618 28,995 6,507 5,144 4,664 3,769 3,297 3,009 2,838 2,719 2,628 2,420 2,230 2,058 2,039 1,963 1,925 1,791 1,716 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 Trace 0,601 0,589 0,538 0,518 0,501 0,481 0,468 0,453 0,428 0,408 0,392 0,366 0,345 0,318 0,308 0,262 0,170 0,158 36,000 1,669 1,637 1,494 1,439 1,391 1,335 1,301 1,258 1,190 1,133 1,088 1,016 0,957 0,882 0,855 0,728 0,473 0,440 TABLE 3 ROTATED FACTOR MATRIX OF THE LEARNING STYLES PREFERENCE QUESTIONNAIRE (DIRECT OBLIMIN ROTATION) VARIABLE FACTOR 1 FACTOR 2 FACTOR 3 FACTOR 4 FACTOR 5 FACTOR 6 FACTOR 7 FACTOR 8 Subtest 1 0,302 -0,040 0,288 -0,132 0,257 -0,587 0,049 -0,060 Subtest 2 0,223 -0,090 0,187 0,020 -0,070 -0,249 0,553 0,009 Subtest 3 0,455 -0,030 0,292 -0,133 0,079 -0,252 0,186 -0,040 Subtest 4 -0,040 -0,128 0,421 0,317 0,095 0,088 0,251 0,115 Subtest 5 0,452 -0,163 0,328 0,005 -0,142 0,037 0,211 0,039 Subtest 6 0,013 -0,010 -0,009 0,305 -0,109 -0,529 0,087 0,019 Subtest 7 0,257 -0,264 0,142 0,152 0,238 -0,080 0,321 0,089 Subtest 8 0,008 -0,122 0,611 0,125 -0,157 -0,030 0,122 0,026 Subtest 9 0,037 0,007 0,411 -0,090 0,083 -0,380 0,049 0,114 Subtest 10 0,201 0,416 -0,113 0,187 -0,132 -0,162 0,024 0,057 Subtest 11 -0,030 0,186 0,041 0,103 -0,010 -0,007 0,531 -0,080 Subtest 12 0,108 -0,010 0,332 0,092 -0,040 -0,177 0,359 -0,060 Subtest 13 0,298 -0,070 0,070 -0,106 0,055 -0,326 -0,080 0,368 Subtest 14 0,393 -0,040 -0,162 -0,050 0,115 0,061 0,385 0,246 Subtest 15 0,468 -0,124 -0,060 -0,060 -0,050 -0,060 0,201 0,208 Subtest 16 0,101 0,836 0,076 0,643 -0,040 0,030 -0,080 -0,040 Subtest 17 0,070 0,256 0,046 0,100 0,112 -0,008 0,164 0,164 Subtest 18 0,611 0,085 0,028 0,055 0,016 -0,050 -0,125 0,030 Subtest 19 0,202 0,045 0,031 0,216 0,289 -0,139 0,070 0,056 Subtest 20 -0,114 0,020 -0,148 0,299 0,147 -0,150 0,318 0,186 Subtest 21 0,134 -0,010 -0,080 -0,040 0,050 -0,050 0,560 0,048 Subtest 22 0,191 -0,122 0,053 -0,006 0,390 -0,163 0,224 0,013 Subtest 23 0,044 0,022 0,063 0,049 -0,480 -0,208 0,287 0,127 An inspection of Table 2 shows that there are eight eigenva- lues greater than unity. Accordingly, eight factors were extra- cted and rotated to simple structure by means of the Direct Oblimin procedure.The rotated factor matrix is given in Ta- ble 3. From an inspection of this table it is clear that six of the factors obtained are well determined, with three or more signi¢cantloa- dings. Factors 4 and 5 are poorly determined and were excluded fromthe study. Factor 4 loads ononly three items and factor 5 on six items. The intercorrelations between the factors are given at the end of the table. Next, the intercorrelation matrix of the six factors retained was subjected to a second-order factor analysis. This analysis yielded one factor only, as can be seen in Table 4. All calculations were done by means of the SPSS-Windows pro- gramme of SPSS ^ International. Measuring instruments For the purpose of this study, permission was granted by Kolb to use the LSI (85) as a basis for the research of learning style preferences under South African conditions. The rationale of Kolb’s four learning modes, of namelyconcrete experience, re- £ective observation, abstract conceptualisation and active ex- perimentation, was used as a guideline for the design of a new questionnaire. In the design, the ¢elds of learning style descriptions were extended by adding new descriptions. Exis- ting learning style descriptions that were found to be too com- plex were simpli¢ed and care was taken to ensure that each item consisted of only one learning style description to pre- vent any chance of confusion. The newly constructed Lear- ning Style Preference Questionnaire (LSPQ) consists of 136 items. The items are formulated as questions linked to a 7- point response scale ranging from 1 (not at all) to 7 (to a great extent). Biographical questions relating to the participant’s personal and employment history were also included in the questionnaire. The initial form of the questionnaire was ap- plied to a small group of individuals, representative of the tar- get population. Comments on the degree of complexity, ambiguity and overlap of items were invited and considered in the construction of the ¢nal version of the questionnaire. Procedure The questionnaires were distributed to the participants via the company’s intranet. Inviewof the numberof questionnaires to be mailed, and the e¡ect this could have on the company’s net- work, special permission had to be obtained. Clear instruc- tions were given as to the electronic completion and return of the questionnaires. All completed questionnaires received we- re stored electronically for future reference. RESULTS To determine the factor structure of the LSPQ a procedure de- veloped by Schepers (1992) was followed.The 136 items of the LSPQ were intercorrelated and the eigenvalues of the unredu- ced intercorrelation matrix were calculated. Owing to limited space, the intercorrelation matrix (136 x136) is not reproduced here. Thirty-six factors were postulated according to Kaiser’s (eigenvaluesgreaterthanone) criterion (1961), and extracted by means of a principal factor analysis.The factor matrix obtained was rotated to simple structure by means of avarimax rotation. Next, subscores were formed for each of the factors by adding together the scores of items with substantial loadings (> 0,30) on a factor. The subscores were then intercorrelated and sub- jected to a principal factor analysis. Owing to limited space, the matrix of intercorrelations of the subscores (36 x 36) is not reproduced here * The eigenvalues of the unreduced intercorrelation matrix are given inTable 2. VILJOEN, SCHEPERS,VAN ZYL54 TABLE 4 SECOND-ORDER FACTOR MATRIX Factor 1 Factor 2 Factor 3 Factor 4 Factor 5 Factor 6 FACTOR I 0,860 0,309 0,644 -0,879 0,808 0,717 h2j 0,739 0,095 0,414 0,774 0,653 0,514 TABLE 5 ITEM STATISTICS IN RESPECT OF SCALE 1OF THE LEARNING STYLES PREFERENCE QUESTIONAIRE ITEM DESCRIPTION OF ITEM MEAN OF STANDARD INDICES OF ITEM-TOTAL ITEM DEVIATION RELIABILITY CORRELATION OF ITEM Xg sg rgx sg rgx 1a. How often do you think about ideas while learning? 5,288 1,273 0,722 0,567 1b. How e¡ectively do you learn when you think about ideas? 5,183 1,292 0,708 0,548 2a. To what extent do you learn while doing things? 5,828 1,140 0,517 0453 2b. How e¡ectively do you learn while doing things? 5,812 1,165 0,575 0,494 3a. To what extent do you learn when you are emotionally involved in the situation? 5,290 1,358 0,569 0,419 3b. How e¡ectively do you learn when you are emotionally involved in the situation? 5,185 1,351 0,550 0,407 5a. To what extent do you learn by ¢rst analysing and then generalising? 5,212 1.090 0,569 0,522 5b. How e¡ectively do you learn by ¢rst analysing and then generalising? 5,266 1,161 0,768 0,584 7a. To what extent do you rely on logical thinking while learning? 5,906 1,096 0,590 0,539 7b. How e¡ectively do you learn when you rely on logical thinking? 5,941 0,963 0,573 0,594 11a. To what extent do you learn by following a practical approach to matters? 5,851 0,929 0,484 0,521 11b. How e¡ectively do you learn by following a practical approach to matters? 5,924 0,948 0,478 0,505 13a. To what extent do you tend to reason things out while learning? 5,919 1,009 0,660 0,654 14a. How often do you try out things while learning? 5,472 1,276 0,688 0,538 19a. To what extent do you try out things while learning? 5,768 0,979 0,614 0,627 19b. How e¡ectively do you learn by thinking? 5,755 1,074 0,650 0,605 22b. How e¡ectively do you learn by considering all aspects of an issue? 5,755 1,074 0,650 0,605 23a. How important is it to understand the basic principles of an issue while learning? 6,498 0,813 0,385 0,473 23b. How e¡ectively do you learn when you understand the basic principles of an issue? 6,489 0,771 0,379 0,492 25a. To what extent do you like to analyse things while learning? 5,851 0,991 0,681 0,687 25b. How e¡ectively do you learn when you analyse things? 5,956 0,958 0,664 0,693 30a. To what extent do you like breaking things down into parts while learning? 5,581 1,211 0,656 0,541 30b. How e¡ectively do you learn when you break things down into parts? 5,747 1,156 0,614 0,531 31a. To what extent do you take responsibility for your actions while learning? 5,998 0,991 0,569 0,574 31b. How e¡ectively do you learn when taking responsibility for your actions? 6,070 0,976 0,619 0,634 39. How readily would you change your style from learning through feelings to learning through thinking if required? 5,190 1,354 0,670 0,495 40a. To what extent do you like dealing with ideas while learning? 5,627 1,075 0,804 0,748 40b. How e¡ectively do you learn by dealing with ideas? 5,641 1,096 0,782 0,713 44. How readily would you change your style from learning through watching to learning through thinking if required 5,100 1,335 0,615 0,461 54. How readily would you change your style from learning through doing things to learning through thinking if required? 4,862 1,313 0,550 0,419 TABLE 3 (continued) ROTATED FACTOR MATRIX OF THE LEARNING STYLES PREFERENCE QUESTIONNAIRE (DIRECT OBLIMIN ROTATION) VARIABLE FACTOR 1 FACTOR 2 FACTOR 3 FACTOR 4 FACTOR 5 FACTOR 6 FACTOR 7 FACTOR 8 Subtest 24 0,151 0,144 0,017 -0,168 0,201 -0,020 0,094 0,413 Subtest 25 0,105 0,163 -0,213 0,063 0,008 -0,324 0,298 0,070 Subtest 26 0,134 0,014 -0,070 0,064 -0,009 -0,276 -0,090 0,276 Subtest 27 0,098 0,144 0,023 0,018 -0,002 0,023 0,192 0,394 Subtest 28 0,053 0,137 0,567 0,006 0,048 -0,050 -0,129 0,040 Subtest 29 0,439 0,053 0,051 0,203 0,113 0,023 0,038 -0,080 Subtest 30 0,030 0,060 0,053 0,002 -0,134 -0,060 -0,159 0,577 Subtest 31 -0,040 0,780 0,068 -0,020 -0,007 0,022 0,046 0,066 Subtest 32 -0,020 0,098 0,107 -0,050 -0,040 -0,485 -0,030 0,109 Subtest 33 0,017 0,041 0,074 0,043 0,101 -0,309 0,135 0,310 Subtest 34 0,069 -0,103 -0,132 0,067 0,145 -0,285 0,402 0,023 Subtest 35 -0,040 -0,116 0,099 0,252 0,173 -0,080 -0,020 0,351 Subtest 36 0,095 -0,070 0,126 -0,080 -0,060 -0,050 0,534 0,142 MATRIX OF INTERCORRELATIONS OF FACTORS 1 2 3 4 5 6 1 1,000 0,018 0,302 -0,473 0,392 0,409 2 0,018 1,000 -0,123 -0,050 -0,070 0,189 3 0,302 -0,123 1,000 -0,254 0,152 0,193 4 -0,473 -0,050 -0,254 1,000- 0,387 -0,482 5 0,392 -0,070 0,152 -0,387 1,000 0,297 6 0,409 0,189 0,193 -0,482 0,297 1,000 Next, six scales were formed by allocating the items with sub- stantial loadings (> 0,30) on a particular factor, to a scale. Each of the six scales was subjected to item analysis using the NP 50 programme of the National Institute for Personnel Research. The item statistics of Scale1are given inTable 5. Owing to limited space, only a summary of the item statistics of the other ¢ve scales will be given here. From an inspection of the items in Scale1it can be seen that respondents with high scores on Scale 1 learn through practical involvement in the * The relevant matrices are available from the author on request learning situation, analysing detail. This conclusion was rea- ched by inspecting the subtests of Scale 1 (Table 5).The subtests forming Scale 1 are Subtest 3 with 12 items (Q7B, Q13A, Q19A, Q19B, Q22B, Q25A, Q25B, Q39, Q40A Q40B, Q44, Q54), Subtest 5 with 5 items (Q1A, Q1B, Q2A, Q2B, Q14A), Subtest14 with 4 items (Q11A, Q11B, Q23A, Q23B), Sub- test15 with 4 items (Q30A, Q30B, Q31A, Q31B), Subtest18 with 2 items (Q5A, Q5B) and Subtest 29 with 3 items (Q3A, Q3B, Q7A). Items from this scale link learning to notions such as practical, basic, breakingthingsdown intoparts,analyticalandthinking.Thedomi- nant subtest, Subtest 3, links learning to concepts such as reasoning thingsout,consideringanddealingwith ideas.Insummary,thissca- le portrays learning through reasoning. The summary statistics in respect of Scales 1to 6 appear in Ta- ble 6 and a summary of the items included in the subtests of the scales appear in Table 7. From Table 6 it is clear that the mean of Scale 1 is 169, 956 and the standard deviation is 18,305. All 30 items were retained and the reliability according to Cronbach’s coe⁄cient alpha is 0,919. 55LEARNING STYLE PREFERENCE QUESTIONNAIRE TABLE 6 MEANS AND STANDARD DEVIATIONS OF THE ITEM STATISTICS OF THE SIX SCALES OF THE LSPQ SCALE SCALE 1 SCALE 2 SCALE 3 SCALE 4 SCALE 5 SCALE 6 Mean Standard Deviation Mean Standard Deviation Mean Standard Deviation Mean Standard Deviation Mean Standard Deviation Mean Standard Deviation MEAN OF ITEMS 5,665 0,394 4,299 0,626 4,913 0,343 5,532 0,218 5,863 0,280 5,367 0,351 STANDARD DEVIATION OF ITEMS 1,107 0,162 1,498 0,177 1,305 0,149 1,097 0,101 1,003 0,135 1,148 0,107 INDICES OF RELIABILITY rgx sg 0,610 0,099 0,948 0,217 0,785 0,130 0,659 0,110 0,590 0,073 0,677 0,065 ITEM-TOTAL CORRELATION rgx 0,556 0,090 0,625 0,084 0,601 0,058 0,600 0,081 0,594 0,077 0,592 0,058 DESCRIPTIVE STATISTICS OF THE SIX SCALES OF THE LSPQ Mean of Scale Standard Deviation Cronbach Alpha Number of Items SCALE 1 169,956 18,305 0,919 30 SCALE 2 38,692 8,534 0,809 9 SCALE 3 78,613 12,568 0,881 16 SCALE 4 138,290 16,467 0,925 25 TABLE 7 ITEMS INCLUDED IN THE SUBTESTS OF THESIX SCALES OF THE LSPQ Items included in subtests of Scale 1 Items included in subtests of Scale 2 Items included in subtests of Scale 3 Subtest 3: 7b, 13a, 19a, 19b, 22b, 25a, 25b, 39, 40a, 40b, 44, 54 Subtest 5: 1a, 1b, 2a, 2b, 14a Subtest 14: 11a, 11b, 23a, 23b Subtest 15: 30a, 30b, 31a, 31b Subtest 18: 5a, 5b Subtest 29: 3a, 3b, 7a Subtest 10: 4a, 4b, 18, 29, 34 Subtest 17: 63a, 63b, Subtest 31: 16a, 16b Subtest 4: 15a, 15b, 21a, 21b, 45a, 45b Subtest 8: 46a, 46b, 74a, 74b Subtest 9: 48a, 48b, 69a, 69b Subtest 28: 65a, 65b Items included in subtests of Scale 4 Items included in subtests of Scale 5 Items included in subtests of Scale 6 Subtest 1: 22a, 35a, 35b, 50a, 50b, 53a, 53b, 55a, 55b, 60a, 60b, 67a, 67b, 70a, 70b Subtest 6: 72a, 72b, 73a, 73b Subtest 25: 52a, 52b Subtest 26: 10a, 10b Subtest 32: 58a, 58b Subtest 2: 9a, 9b, 13b, 14b, 42a, 42b, 51b, 56a, 56b, 62a, 62b, 68a, 68b, 71a, 71b Subtest 7: 36a, 36b, 37a, 37b Subtest 11: 49, 59, 64 Subtest 12: 51a, 61a, 61b Subtest 20: 41a, 41b Subtest 21: 26a, 26b Subtest 34: 57a, 57b Subtest 36: 20a, 20b Subtest 13: 33a, 33b, 38a, 38b Subtest 24: 17a, 17b Subtest 27: 8a, 8b An inspection of Table 6 shows that the mean of Scale 2 is 38,692 and the standard deviation is 8,534. All 9 items were re- tained and the reliability according to Cronbach’s coe⁄cient alpha is 0,809. It is evident from an inspection of the items in Scale 2 that respondents with high scores on Scale 2 learn throughwatching and observing, internalising the learning si- tuation and content. Items from Scale 2 typically refer to no- tions such as watching and observing and reviewing processes. Subtests 10,17 and 31 form the basis of this scale, with no sub- test dominating. In summary, this scale measures learning through observation. FromTable 6 it is clear that the mean of Scale 3 is 78,613 and the standard deviation is12,568. All16 items were retained and the reliability according to Cronbach’s coe⁄cient alpha is 0,881. An inspection of the items in Scale 3 shows that people with high scores on Scale 3 learn through exploring new and untes- ted terrain. Items from Scale 3 typically refer to notions such as risk-taking, breaking rules, trying the new, relying on gut fee- lings, exploring and testing alternatives. Subtests 4, 8,9 and 28 form the basis of this scale, with no subtest dominating. In summary, this scale measures learning through exploring al- ternatives. The mean of Scale 4 is 138,290 and the standard deviation is 16,467, as shown inTable 6. All 25 items were retained and the reliability according to Cronbach’s coe⁄cient alpha is 0,925. An inspection of the items contained in Scale 4 indicates that respondents with high scores on Scale 4 learn by establishing a broad base, including all aspects of the learning situation. Items from Scale 4 typically refer to notions such as under- lying theory, basic assumptions, gaining an overall view and considering all aspects of the learning situation. Subtests 1, 6, 25, 26 and 32 form the basis of this scale, with Subtest 1 being the dominant subtest. In summary, this scale portrays learning through relying on a sound basis of fact and theory. An inspection of Table 6 shows that the mean of Scale 5 is 193,487 and the standard deviation is 19,450. All 33 items were retained and the reliability according to Cronbach’s coe⁄cient alpha is 0,939. It is clear from an inspection of the items in Scale 5 that respondents with high scores on Scale 5 learn through active involvement in the learning situation, responding spon- taneously. Items from Scale 5 typically refer to notions such as spontaneous learning, practising new skills, trying things out, seeing results and being active in learning, suggesting a conti- nuous involvement in learning. Subtests 2, 7, 11, 12, 20, 21, 34 and 36 form the basis of this scale, with Subtest 2 being the dominant subtest. In summary, this scale portrays learning through practising new skills. FromTable 6 it is clear that the mean of Scale 6 is 75,138 and the standard deviation is 9,471. All 14 items were retained and the reliability according to Cronbach’s coe⁄cient alpha is 0,854. From an inspection of the items in Scale 6 it is evident that people with high scores on Scale 6 learn through considering current and future situations and carefully exercising options. Items from Scale 6 typically refer to notions such as consi- dering all facts, being open to conviction, relying on facts and taking time before acting, suggesting that learning takes place by focusing on current events projected onto future events. Subtests 13, 24, 27, 30, 33 and 35 form the basis of this scale, with no subtest dominating. In summary, this scale por- trays learning through considering all facts at hand. From the six scales it appears that people do learn in more VILJOEN, SCHEPERS,VAN ZYL56 TABLE 8 HOTELLING T2 : COMPARISON OF THE MEANS OF MEN AND WOMEN IN RESPECT OF THE SIX SCALESOF THE LSPQ MEN WOMEN VARIABLE X1 S1 N1 X2 S2 N2 Levene F df p(F) t-value df p(t) SCALE 1 169,90 17,4577 350 170,12 19,9197 189 0,262 1& 537 0,609 -0,129 537 0,898 SCALE 2 38,1857 8,3705 350 39,5759 8,7776 191 1,033 1 & 539 0,310 -1,815 539 0,070 SCALE 3 78,0029 12,0874 350 79,6737 13,4093 190 3,348 1 & 538 0,068 -1,475 538 0,141 SCALE 4 138,70 15,3128 349 137,47 18,5432 189 4,804 1 & 536 0,029 0,859 536 0,391 SCALE 5 191,44 19,6071 345 196,75 18,7752 187 1,426 1 & 530 0,233 -3,029 530 0,003* SCALE 6 74,9029 9,2245 350 75,5904 9,9285 188 0,476 1 & 536 0,490 -0,802 536 0,423 * Statistically signi¢cant Hotelling T2 = 0,067 df = 6 and 514 F = 5,707 p = 0,001 (Signi¢cant) TABLE 9 MANOVA AND ASSOCIATEDANOVAS: COMPARISON OF THE MEANS OF THE VARIOUS ‘‘ACADEMIC GROUPS’’ IN RESPECT OF THE SIX SCALES OF THE LSPQ WILKS’ COEFFICIENT LAMBDA F df p 0,834 SCALE 1 SCALE 2 SCALE 3 SCALE 4 SCALE 5 SCALE 6 5,190 4,175 12,055 4,127 2,514 0,043 2,543 18, 1411,9 3, 521 3, 523 3, 522 3, 520 3, 515 3, 520 0,001 0,006* 0,001* 0,007* 0,058 0,988 0,056 TABLE 10 DUNNETT’S POST-HOC MULTIPLE COMPARISONS: COMPARISON OF THE MEANS OF THE VARIOUS ‘‘ACADEMIC GROUPS’’IN RESPECT OF SCALE 1 OF THE LSPQ VARIABLE MEANS GROUPS SCALE 1 *p = 0,05 1 GRADE 12 OR EQUIVALENT N = 156 165,63 2 TECHNICON DIPLOMA N = 144 171,47 3 BACHELOR’S DEGREE N = 121 170,45 4 POST GRA- DUATE DEGREE/ DIPLOMA N = 104 172,93 1/2 * 1/3 1/4 * 2/3 2/4 diverse ways than is re£ected by the original instrument of Kolb.The LSPQ not only included and broadened the existing descriptions of learning by Kolb, but also added deeper levels of learning outside the existing descriptions. An ancillary aim of this study was to determine whether there are statistically signi¢cant di¡erences in the means of the sub- groups of the sample in respect of gender, academic quali- ¢cations and functional disciplines as far as learning style preferences are concerned. FromTable 8 it is apparent that the- re are statistically signi¢cantly di¡erences between the vectors of means of the men and women (Hotelling’sT2 = 0,067; F (6, 514) = 1,426; p < 0,001). However, only the di¡erence in respect of Scale 5 is statistically signi¢cant, the mean score of the wo- men being higher than that of the men. Thus there is partial support for hypothesis 1. The results of the multiple analysis of variance (MANOVA) for the four academic groups in respect of the six scales are shown inTable 9. From an inspection of the table it appears thatWilks’coe⁄cient lambda is equal to 0,834 with an associated F (18 and 1411,9) = 5,190; p < 0,001.The overall null hypothesis is there- fore rejected. From the one-way analyses of variance (ANO- VA), it appears that there are statistically signi¢cant di¡erences in the means of Scale1, F (3 and 521) = 4,175; p = 0,006; Scale 2, F (3 and 523) = 12,055; p < 0,001 and Scale 3, F (3 and 522) = 4,127; p = 0,007. Next, multiple comparisons were do- ne using Sche¡e¤ ’s technique if the variances did not di¡er and Dunnett’s technique if the variances di¡ered. Dunnett’s post- hoc multiple comparisons technique was used to determine which academic groups di¡ered from one another in respect of Scale 1. The results show that the group with Grade 12 or equivalent quali¢cations di¡er statistically signi¢cantly from the group with technicon diplomas and the group with post- graduate degrees/diplomas.The Grade12 mean scores were lo- wer, as shown inTable 10. Sche¡e¤ ’s post-hoc multiple comparisons technique was used to determine which academic groups di¡ered from one another in respect of Scales 2 and 3.The results for Scale 2 show that the group with Grade 12 or equivalent quali¢cations di¡ers statis- tically signi¢cantly from the groups with bachelor’s degrees and post-graduate degrees/diplomas, their mean scores being higher. Similarly, the group with technicon diplomas di¡ers statistically signi¢cantly from the group with bachelor’s de- grees and post-graduate degrees/diplomas, their mean scores being higher, as shown inTable11.The results for Scale 3 show that the group with Grade 12 or equivalent quali¢cations dif- fers statistically signi¢cantly from the group with post-gra- duate degrees/diplomas, their mean scores being lower, as shown in Table 11. There are no di¡erences in respect of the other scales. Hypothesis 2 is therefore only partially supported. The results of the MANOVA for the eight groupings of func- tional disciplines in respect of the six scales of the LSPQ are shown inTable 12. An inspection of Table 12 shows thatWilks’coe⁄cient lambda is equal to 0,831 (F (48, 2497) = 1,978; p < 0,001). The overall null hypothesis is therefore rejected. From the ANOVAs, it is apparent that there are statistically signi¢cantly di¡erences in the means of Scale 2 (F (8, 578) = 3,178; p = 0,002), Scale 3 (F (8, 526) = 2,013; p = 0,043) and Scale 6 (F (8, 524) = 2,882; 57LEARNING STYLE PREFERENCE QUESTIONNAIRE TABLE 11 SCHEFFE¤ ’S POST-HOC MULTIPLE COMPARISONS: COMPARISON OF THE MEANS OF THE VARIOUS ‘‘ACADEMIC GROUPS’’ IN RESPECT OF SCALES 2,3,4,5 AND 6 OF THE LAPQ VARIABLE MEANS GROUPS SCALE 2 SCALE 3 SCALE 4 SCALE 5 SCALE 6 *P = 0,05 1 GRADE 12 OR EQUIVALENT N = 157 41,07 76,50 135,48 193,10 75,38 2 TECHNICON DIPLOMA N = 145 39,83 77,51 139,62 193,35 76,41 3 BACHELOR’S DEGREE N = 121 35,98 80,30 137,86 193,28 73,22 4 POST GRA- DUATE DEGREE/ DIPLOMA N = 104 36,40 81,30 140,84 192,84 75,10 1/2 1/3 * 1/4 * * 2/3 * 2/4 * 3/4 TABLE 12 MANOVA AND ASSOCIATEDANOVAS: COMPARISON OF THE MEANS OF THE VARIOUS ‘‘FUNCTIONAL DISCIPLINES’’ IN RESPECT OF THE SIX SCALES OF THE LSPQ WILKS’ COEFFICIENT LAMBDA F df p 0,831 SCALE 1 SCALE 2 SCALE 3 SCALE 4 SCALE 5 SCALE 6 1,978 1,660 3,178 2,013 1,763 1,066 2,882 48, 2479 8, 525 8, 527 8, 526 8, 524 8, 519 8, 524 0,001* 0,106 0,002* 0,043* 0,082 0,386 0,004* TABLE 13 SCHEFFE’S POST-HOC MULTIPLE COMPARISONS: COMPARISON OF THE MEANS OF VARIOUS ‘‘FUNCTIONAL DISCIPLINES’’ IN RESPECT OF SCALES 1,2,4 AND 5 OF THE LSPQ VARIABLE SCALE 1 SCALE 2 SCALE 4 SCALE 5 1- H R 170,63 35,09 -91,93 190,81 2 DISTRIBUTION 172,73 39,28 -93,94 190,01 3- FINANCE 169,19 38,04 -91,44 187,18 4- INF SERVICES 174,53 38,26 -93,58 186,47 5- MARKETING 163,36 38,60 -87,34 182,53 6- PRODUCTION 170,37 38,12 -92,73 186,60 7-RISK CONTROL 160,77 39,92 -91,46 182,38 8- SALES 170,75 42,63 -97,84 190,04 1/2 1/3 1/4 1/5 1/6 1/7 1/8 2/3 2/4 2/5 2/6 2/7 2/8 3/4 3/5 3/6 3/7 3/8 4/5 4/6 4/7 4/8 5/6 5/7 5/8 6/7 6/8 7/8 *p = 0,05 G R O U P S M E A N S TABLE 14 DUNNETT’S POST-HOC MULTIPLE COMPARISONS: COMPARISON OF THE MEANS OF THE VARIOUS’’FUNCTIONAL DISCIPLINES’’ IN RESPECT OF SCALES 3 AND 6 OF THE LSPQ VARIABLE SCALE 3 SCALE 6 1- H R 82,75 74,19 2 DISTRIBUTION 79,27 76,86 3- FINANCE 76,43 73,36 4- INF SERVICES 77,95 78,61 5- MARKETING 78,30 73,28 6- PRODUCTION 76,80 74,29 7-RISK CONTROL 75,30 72,77 8- SALES 81,84 78,78 1/2 1/3 1/4 1/5 1/6 1/7 1/8 2/3 2/4 2/5 2/6 2/7 2/8 3/4 3/5 3/6 3/7 3/8 4/5 4/6 4/7 4/8 5/6 5/7 5/8 6/7 6/8 7/8 *p = 0,05 G R O U P S M E A N S p = 0,004). Next, Sche¡e¤ ’s post-hoc multiple comparisons technique was used to determine which functional disciplines di¡ered from one another in respect of Scale 2.The results are given inTable 13. It is clear that Human Resources di¡ers sta- tistically signi¢cantly from Sales, the mean score of Sales for Scale 2 being higher than that of Human Resources. Next,Dunnett’s post-hoc multiple comparisons technique was used to determine which functional disciplines di¡ered from one another in respect of Scales 3 and 6. In respect of Scale 3 no statistically signi¢cant di¡erences between the groups were found (seeTable14). Finally, in respect of Scale 6 it appears that Finance di¡ers statistically signi¢cantly from Sales, the mean score of Sales being higher than that of Finance (seeTable 14). This is shown in Table 14. Hypothesis 3 is therefore only par- tially supported. VILJOEN, SCHEPERS,VAN ZYL58 DISCUSSION At face value, learning styles and learning style instruments, and in particular the LSI (85), are appealing to people from business. Many researchers, however, have not endorsed the LSI (85) owing to its poor psychometric properties resulting mainly from its ipsative nature. The 136 items written for the LSPQ were subjected to a factor analysis, resulting in six clearly de¢ned factors. Next, six scales were constructed corresponding to the six factors that had been identi¢ed.These scales yielded reliabilities ranging from 0,809 to 0,939. The six scales that emerged were identi¢ed as Scale 1: learning through reasoning, Scale 2: learning through observing, Scale 3: learning through exploring alternatives, Scale 4: learning through relying on a sound basis of fact and theory, Scale 5: learning through practising new skills and Sca- le 6: learning through considering all facts at hand. From the second-order factor analysis that was done on the six scales, it is clear that there is a single second-order factor under- lying the six scales, with ¢ve of the scales having substantial loa- dings on the second-order factor, but with Scale 2 having avery lowcommonality. Furthermore, it has transpired that Scale 2 has a high speci¢city (0,714), i.e. it relates to a di¡erent kind of lear- ning than the other ¢ve scales. From the content, it is apparent that Scale 2 relates to relatively simple forms of learning, whereas the other ¢ve scales relate to more complex forms of learning, requiring higherlevels of insight. It is also interesting to notethat Scale 4, learning through relying on a sound basis of fact and theory, has a negative loading on this factor, indicating that peo- ple with high scores onthis scale tend to have lower scores onthe rest. On the surface it appears that there is a connection between the two bipolar dimensions of Kolb and the new scales of the LSPQ. It appears that there is a connection between Scale 4 (learning through relying on a sound basis of fact and theory) and Scale 6 (learning through considering all facts at hand) and the dimensions of abstract conceptualisation and concrete expe- rience. There also appears to be a connection between Scale 2 (learning through observing) and Scale 5 (learning through practising new skills) and the dimensions of re£ective observa- tion and active experimentation. Scale 1 (learning through rea- soning) and Scale 3 (learning through exploring alternatives) seem to be new and unique. It appears that Scales 1 and 3 are related to higher-order learning involving thinking processes instead of practical processes. The present study partially con- ¢rms the ¢ndings of Merrit and Marshall (1984) and Geiger et al. (1993). Merrit and Marshall’s ¢ndings suggest that the norma- tive form of the LSI is consistent with the learning style model proposed by Kolb. Geiger et al. (1993) found that their factor analysis did not support the existence of any bipolar dimen- sions, but did support the four separate learning styles.The pres- ent study also found support for four separate learning styles. However, it also identi¢ed two additional scales relating to hig- her-orderlearning.This was established by including more lear- ning modes than the LSI (85). As far as the ancillary aims are concerned, it was found that the mean score of the women in respect of Scale 5 (relating to lear- ning through practising new skills) is higher than that of men. The present study thus partially con¢rms the ¢ndings of Se- veriens andTen Dam (1997), who found that men and women prefer di¡erent learning styles, and that men show a greater preference for the abstract conceptualisation learning mode. The results are contrary to the ¢ndings of Wilson (1996), na- mely that women prefer not to be actively involved in the lear- ning situation. As far as the learning style preferences of the various academic groups are concerned, it was found that there are statistically signi¢cant di¡erences in respect of Scales 1, 2 and 3. As far as Scale 1 is concerned, it appears that people with Grade 12 or equivalent quali¢cations have lower mean scores than people with technicon diplomas and post-graduate degrees/diplomas, indicating that the groups with technicon diplomas and post- graduate degrees/diplomas prefer learning of a higher-order. In respect of Scale 2, it appears that people with Grade 12 or equivalent quali¢cations have higher mean scores than people with bachelor’s degrees and post-graduate degrees/diplomas, indicating that the group with Grade 12 or equivalent qua- li¢cations prefers learning of a lower complexity. Also the group with technicon diplomas has a higher mean score on Scale 2 than the group with bachelor’s degrees and post-gra- duate degrees/diplomas, indicating that the group with tech- nicon diplomas prefers learning of a lower complexity. In respect of Scale 3, it appears that people with Grade12 or equi- valent quali¢cations have a lower mean score than people with post-graduate degrees/diplomas, indicating that the group with post-graduate degrees/diplomas prefers learning of a hig- her-order. The present study thus partially con¢rms the ¢n- dings of Kolb (1984), namely that people with di¡erent academic quali¢cations prefer di¡erent learning styles. In this study it appears that the group with Grade 12 or equivalent quali¢cations di¡ers from the other academic groups. It ap- pears that they prefer learning of a lower complexity. Di¡erences in respect of the learning style preferences of the various functional disciplines are re£ected in statistically signi- ¢cant di¡erences on Scales 2 and 6. In respect of Scale 2, it ap- pears that people from Human Resources have a lower mean score than people from Sales, indicating that people from Sales prefer learning of a lower complexity. In respect of Scale 6, it appears that people from Finance have a lower mean score than people from Sales, indicating that people from Sales prefer learning of a more concrete nature. In conclusion, the present study opens up new research possi- bilities, for instance understanding and ¢nding the correct ba- lance in responding to the generic learning style preferences of teams and individuals when conducting training. Although the focus of this study was on the cognitive aspects of learning, the value of the a¡ective-social aspects of learning should not be overlooked and should be explored in future studies.The promi- sing psychometric properties of the LSPQ should encourage ot- her researchers to actively explore the validity of the various scales that were obtained. REFERENCES Abernathy, D. (1999). Leading-edge learning:Two views.Trai- ning & Development, 53 (3), 40-42. Allison, C.W. & Hayes, J. (1988).The Learning Style Question- naire: An alternative to Kolb’s inventory. Journal of Manage- ment Studies, 25, 269-281. Chi-Ching,Y. & Moi, L. (1994). Learning styles and their im- plications for cross-cultural management in Singapore. Journal of Social Psychology,134,596-600. Cornwell, J.M. & Manfredo, P.A. (1994). Kolb’s learning style theory revisited. Educational and Psychological Measurement, 54, 317-327. De Klerk, J. (1993).The connection of apprentices’ learning styles, lear- ning approaches, self-concept and academic achievement. Unpu- blished master’s dissertation, Universityof Pretoria, Pretoria. Dibella, A.J., Nevis, E.C. & Gould, J.M. (1996). Understanding organisational learningcapability. JournalofManagement,33, 361-379. Dubois, D.D. (1993). Competency-based performance improvement. Amherst, MA: HRD Press. Ellinger, A.D.,Watkins, K.E. & Bostrom, R.P. (1999). Mana- gers as facilitators of learning in learning organisations. Human Resource Development Quarterly,10,105-125. Ferrell, B.G. (1983). A factoranalysis comparison of fourlearning styles instruments. Journal of Educational Psychology,75,33-39. Freedman, R. & Stumpf, S. (1980). Learning style theory: Less than meets the eye. Academy of Management Review, 5, 445- 447. Geiger, M.A., Boyle, E.J. & Pinto, J.K. (1992). A factor analysis of Kolb’s revised learning style inventory. Educational and Psychological Measurement, 52, 753-759. Geiger, M.A., Boyle, E.J. & Pinto, J.K. (1993). An examination of 59LEARNING STYLE PREFERENCE QUESTIONNAIRE ipsative and normative versions of Kolb’s revised Learning Styles Inventory. Educational and Psychological Measurement, 53, 717-726. Heymans, B.S. (1988). The formation of an individualised mana- gement development model based on learning styles in a SouthAfri- can company. Unpublished master’s dissertation, University of theWitwatersrand, Johannesburg. Hickson, J. (1996). Gender related learning style patterns of middle school pupils. School Psychology International,17,59-70. Honey, P. & Mumford, A. (1982).The manual of learning styles. Maidenhead: Honey. Honey, P. & Mumford, A. (1995). Capitalizing on your learning style (3rd ed.). Carmarthen, Dyfed: Management Learning Resources, Ltd. Hong, E. & Suh, B.K. (1995). An analysis of change in Korean- American and Korean students’ learning styles. Psychologi- cal Reports,76,691-699. Jonassen, D.A. & Garabowski, B.L. (1993). Handbook of indivi- dual di¡erences, learning, and instruction. Hillsdale, NJ: La- wrence Erlbaum. Kaiser, H.F. (1961). A note on Guttman’s lower bound for the number of common factors. BritishJournal of Statistical Psy- chology,14,1. Kolb, D.A. (1976). Learning style inventory. Technical Manual. Boston, MA: McBer. Kolb, D.A. (1981). Experiential learning theory and the learn- ing style inventory: A reply to Freedman and Stumpf. Aca- demy of Management Review,6, 289-296. Kolb, D.A. (1984). Experiential learning. Englewood Cli¡s, NJ: Prentice-Hall. Kolb, D.A. (1985). Learning Style Inventory. Self-scoring Inventory and Interpretation Booklet. Boston, MA: McBer. Kolb, D.A. (1995). Learning Style Inventory. Technical speci¢- cations. Boston, MA: McBer. Loo, R. (1999). Con¢rming factor analyses of Kolb’s learning style inventory. British Journal of Educational Psychology, 69, 213-219. McCarthy, B. (1987).The 4 mat system. Barrington, IL: Excel. Merrit, S.L. & Marshall, J.C. (1984). Reliability and construct validity of ipsative and normative forms of the learning style inventory. Educational and Psychological Measurement, 44,463-472. Newstrom, J.W. & Legnick-Hall, M.L. (1991). One size does not ¢t all.Training & Development,45,43- 48. Pedler, M. & Aspinwall, K. (1998). A concise guide to the learning organisation. London: Lemos & Crane. Prokesch, S.E. (1997). Unleashing the power of learning. Har- vard Business Review,75,146-168. Reynolds, M. (1997). Learning styles: A critique. Management Learning, 28,115-133. Schepers, J.M. (1992). Toetskonstruksie:Teorie en praktyk. Johan- nesburg: RAU Press. Senge, P., Roberts, C., Ross, R.B., Smith, B.J. & Kleiner, A. (1994).The ¢fth discipline ¢eldbook. NewYork: Doubleday. Senge, P., Kleiner, A., Roberts, C., Ross, R.B., Roth, G. & Smith, B.J (1999).The dance of change. NewYork: Double- day. Senge, P. (1999).Why organisations still aren’t learning.Training & Development, 36,40-49. Severiens, S. & Ten Dam, G. (1997). Gender and gender iden- tity di¡erences in learning styles. Educational Psychology,17, 79-93. Slaat, A., Lodewijks, H.G.L.C. & Van der Saden, J.M.M. (1999). Learning styles in secondary vocational institu- tions. Learning and Instruction, 9,475-492. Smedley, L.C. (1987). Chemists as learners. Journal of Chemical Education,64, 321-323. Stuart, P. (1992). New direction in training individuals. Per- sonnelJournal,71, 86-94. Van Wyk, B.J. (1992). Cognitive learning styles of black teacher students. Unpublished master’s dissertation, University of Port Elizabeth, Port Elizabeth. Verses, J.G., Sims, R.R. & Locklear,T.S. (1991). Improving the reliability of Kolb’s revised Learning Style Inventory. Edu- cational and Psychological Measurement, 51,144-150. Willcoxson, L. & Prosser, M. (1996). Kolb’s Learning Style In- ventory (1985): Review and further study of validity and reliability. BritishJournal of Psychology,66, 247-257. Wilson,V. (1996). Scholars, active learners, and social butter£ies: pre- ferred learning styles of high school biology I students. Paper pres- ented at the annual meetingof the Mid-South Educational Research Association.Tuscaloosa, AL. VILJOEN, SCHEPERS,VAN ZYL60