The Reliability and Use of Kirton Adaption-Innovation Inventory with Teenagers
The Reliability and Use of Kirton Adaption-Innovation Inventory with Teenagers
M.J. Kirton, Occupational Research Centre
M.E. Tefft, ESDS Division, University of Hertfordshire
Cognitive style is a concept that has been much explored in the field of education but results of studies using measures of it are not always clear. One reason may be that the measures used depended on definitions that did not clarify some of the key elements they contained. Kirton Adaption-Innovation Theory, initially developed in 1976 outside the education field, posits a number of precise elements (that are listed) and the validity and reliability of its measure, the Kirton Adaption-Innovation Inventory (KAI), is well documented among adult populations in several countries; some of its published findings are applicable to education. So, there is a growing interest among educational researchers to apply the Adaption-Innovation Theory to young people, giving rise to a need to establish securely the validity and reliability of the KAI among younger populations. Reliability is the first priority and this paper examines the reliability results of five studies that used the KAI in secondary schools involving nearly 2,500 pupils in all. The results suggest the reliability of KAI to be so robust and the first findings relating to its validity so encouraging that the way is open for major studies of the Adaption-Innovation concept in younger age groups. To this end research design guidelines on key variables are provided to assist the researcher in obtaining dependable results with secondary school pupils. Indications of possible future research that are both challenging and potentially useful are also given.
The use of the Kirton Adaption-Innovation Inventory with Teenagers
Cognitive styles are generally thought to be potentially important factors in education since these influence one’s approach to problem solving and learning (Guilford, 1980; Messick, 1976). Unfortunately earlier studies attempting to link thinking styles with education were, in general, less conclusive than initial promise seemed to indicate (e.g., child, 1986; Head 1995). One reason might well be the use of imprecise and overlapping terms leading to weak measures. Adaption-Innovation (A-I) Theory (Kirton, 1976; 1994) defines some key elements with precision and it may now offer a better chance of conclusive results. Kirton’s A-I Theory posits, firstly, a sharp distinction between (preferred) cognitive style and (potential) cognitive level and (achieved) manifest capacity, i.e., there is no “better” style than another. This assertion has been the subject of voluminous study, the latest review of which is found in Mudd (1996). Secondly, A-I Theory also makes a clear distinction between behaviour, which is assumed as highly flexible and preferred style, for which there is now ample evidence that it is not at all flexible, an assertion recently reviewed by Kirton (1994).
A-I Theory posits that everyone can be located on a cognitive style continuum ranging from a preference for doing things better (adaptive style) to a preference for doing things differently (innovative style) (Kirton, 1976; 1994). The more adaptive individual prefers, in the resolution of problems, more structure than one who is more innovative and also prefers more of that structure to be consensually agreed. This stable preference for varying degrees of structure is distinct from any variety of cognitive ability, be it intelligence, cognitive complexity or management competency. The immediate implication of this assertion is that people prefer to solve problems with and within different amounts of supporting cognitive structure. Applied to education this assertion predicts that people also prefer to learn and teach within varying amounts of structure and, therefore, find themselves more naturally drawn to a particular subject (Kirton et al., 1991), type of teaching method, learning material and learning environment (Head, 1996). For instance, if a teacher is inclined towards adaption and so is the curriculum but the student’s style is markedly more innovative, learning is expected to be more difficult for the student. Difficult but not impossible, for, at a cost, the student can engage in “coping behaviour”, which A-I theory defines as the behaviour needed to close the gap between stable preferred style and needed behaviour; “coping” costs more energy than behaviour in a preferred mode and is fuelled by motive (Kirton 1994; Hayes & Allinson, 1994). Adaption-Innovation Theory, therefore, suggests that there is not an ideal curriculum, teaching style or learning style; all problem solving variables are contingent on each other and the specific nature of the problem to be solved, e.g., a particular teacher needing to teach effectively a particular student a particular curriculum in a particular environment. One role of the teacher, it can now be seen, is to mitigate the “cognitive style gap” between any student and the nature of the material by teaching skill and by raising motivation. This is hardly a new set of assumptions, but now clearly supported by a theory, one that has shown positive results in adults (e.g., Hammerschmidt 1995), and eminently, therefore, open to test on younger populations, as long as the measure of the theory can be relied upon for the sample being used.
The Kirton Adaption-Innovation Inventory (KAI) is a self-report instrument designed to measure a person’s degree of adaptive-innovative preference thereby locating one on the A-I continuum. The KAI was developed for use with literate and work experienced adults and has withstood rigorous study into its validity and reliability using numerous samples and languages (Kirton, 1976, 1994; Prato Previde & Massimini, 1984; Mudd, 1987 and 1996; Kirton & Kubes, 1992; Clapp, 1993; Murdock et al, 1993). One intriguing result that is emerging from Adaption-Innovation research, using both general population and comparative samples, is that this inventory seems to be unaffected by even such powerful environmental influence as culture and may, therefore, be tapping cognitive functional aspects that are deep seated in individuals (Prato Previde, 1991; Goldsmith, 1994; Tullett & Kirton, 1995). Just how early these preferences emerge has yet to be determined, but Adaption-Innovation Theory proposes that cognitive style expresses itself in early childhood and may even be an innate component of personality (Kirton, 1994; van der Molen, 1994); hence its relative inflexibility compared to behaviour.
The precision in key elements of definition used in the A-I theory, the encouraging report of its validity and the well established reliability and stability, over a wide range of situations, of the KAI measure, reopens the possibility of investigating the effects of cognitive style within the educational context (Kirton et al, 1991). There is also now a growing interest (Hammond,1994, Martin 1987, Selby,1992 Taylor, 1994a and 1994b) in applying A-I theory and the KAI to younger populations, particularly in the classroom environment and, therefore, the need to establish the general reliability and validity of the KAI with younger populations has arisen. This paper’s primary consideration is to examine the reliability of the KAI with secondary school pupils. The authors consider the question of the KAI’s reliability with younger populations to be the necessary prelude to its further validation with young people particularly in light of the already well documented validity of the instrument with adults. Without first establishing the KAI’s reliability among secondary pupils, any attempts to apply the Adaption-Innovation concept to younger non-adults could lead to the kind of confusing results that have so far dogged cognitive style studies in education. However, while the emphasis of this paper is to report on the reliability of the KAI with younger populations, two studies examining the measure’s validity are also presented. It is hoped that the strong reliability of the KAI found with adults will hold up satisfactorily when used with the younger samples. The concept itself is, however, a sophisticated one and it is still not yet certain that younger students and pupils (whatever may be the cut-off point) can understand the kind of items of measurement to which they need to be asked to respond. The main purpose of this paper, then, is to explore to what extent the theory and measure can be effectively used with teenagers, and if so where are the limits and the limitations and what cautions can be offered in the use of KAI with this age range. This has been done by analysing the results from five studies involving (a) comparisons of different age groups of secondary school pupils, ranging between 13-19 years and (b) comparisons of test administrators with differing classroom experience in the use of psychological measures. In addition the validity of KAI with younger groups is reviewed, followed by suggestions for future work, including repeating useful studies, on issues relevant to teaching in general.
Five researchers from four English speaking countries conducted studies using the KAI with secondary school pupils and kindly made either their complete KAI data or their reliability results available to the authors to allow a thorough examination of the reliability of the measure with younger populations. A brief description of the sample and the conditions of data collection for each contributing study are given in Table 1.
Table 1. Description of the five studies reported
ˡThird form=grade 8; fourth form=grade 9; fifth form=grade 10; sixth form=grades 11 and 12
²Experienced in the classroom environment and the administration of psychometric measures or
inexperienced by virtue of being either a researcher lacking experience in the classroom environment or a teacher lacking experience in the administration of psychometric measures
It should be noted that in England, Eire and New Zealand, the sixth form contains a higher proportion of pupils intent on university than do the eleventh and twelfth grades in the United States. Pupils in the first three countries can leave school after the fifth form with basic qualifications either to enter the job market or to obtain further training in technical colleges, while those planning on attending university remain in school to complete two additional academic years. In the United States, however, the basic qualification is not obtained until after completing grade 12 and, so, most pupils remain in secondary school until then, whether or not they intend to pursue a tertiary education. Therefore, the sixth forms of England, Eire and New Zealand are a more homogeneous group of pupils than the eleventh and twelfth grades of the United States with respect to academic interest and ability.
The full KAI data from Taylor (1994a), Selby (1992) as well as Tefft (reported here) were made available to the authors who then conducted a detailed item analysis on each sample and calculated the internal reliability as measured by Cronbach alpha. The full KAI data from the studies by Martin (1987) and Hammond (1987, 1994) were not available to the authors, but the original researchers provided Cronbach alpha data.
In the original validation of the KAI (Kirton, 1976, 1987, 1994), each item was shown (a) to be significantly related to the whole measure between the limits of correlation: +o.2 to +0.6; (b) to have no significantly negative correlation of equal to or greater magnitude than -0.2 with any other item. Helped by these restraints, the overall KAI was found to have an impressive internal reliability (alpha: 0.88, N: 808) with its target population of literate, work experienced adults. Similar results were obtained from the translation of the KAI into other languages (Prato Previde & Massimini 1984, Kirton & Kubes 1992, Tullett & Kirton 1995).
A number of items failed to relate to the whole measure to the standard set by the original validation study in each sample reported in this paper for which all KAI data were available. However, there was never secure agreement on which items failed, different items failing with different groups. Furthermore, when any particular failed item was removed from the scale, no significant improvements were made in the overall internal reliability of any group. The item integrity of the inventory gradually decreased year by year until by the youngest levels (third form/grade 8), in particular instances, it was just above the level set by the original KAI study of being adequate (i.e. a Cronbach alpha of 0.7). The type of item that failed in any sample rarely included those that seemed “adult work oriented”, such as those referring to “bosses” or “work colleagues”. The problems were more likely to be on personal evaluation and insight, such as whether respondents viewed themselves as “methodical” or “stimulating”.
As a consequence of the successive gradual weakening of item integrity by decreasing age, the three factor traits, into which the 32 KAI items can be grouped, operated progressively less well for these samples than for adults. By the fifth form/grade l0 it became doubtful whether these factor traits could be used with confidence; the smallest factor (7 items), relating to preference for method and system (see above), began to break up. Below the fifth form/grade 10, the factors could not be safely used, both because they became less consistent in content and because their internal reliabilities dropped below 0,7. However, for the KAI to have worked this well is impressive.
One other factor on performance needs to be mentioned here. In adult general population samples, there are only a very few respondents whose inventories need to be rejected for any reason (e.g. not responding to all the items), unless special factors pertain. Among the school populations this number rose with increasingly younger age groups. Under the most favourable conditions of older pupils (fifth and sixth forms/grades 10, 11 and 12) as well as experienced data collectors, wastage was around 10%. With the youngest pupils (below fifth form/grade 10), even under ideal conditions, (brighter pupils and experienced administrators) the wastage rate increased to around 2O%. With anything less than the most favourable circumstances (test administrators not personally briefed and know to be motivated) the wastage rate ranges from11% in the older samples to as much as 40% in the youngest. Fortunately, with care and these results in mind, most of this wastage can be avoided In discussion later, we shall argue that these are the limits of measures of this kind of concept.
Table 2 displays Cronbach’s alpha results of these studies broken down by school years and whether or not the administrator was experienced or not in use of psychometric instruments in schools. Both variables show significant effects.
Table 2. KAI internal reliabilities for secondary pupils
ˡMartin (1987) ²Taylor (1994a) ³Selby (1992) ⁴Hammond (1987) ⁵Tefft (reported here)
⁶pupils were 2 months away from entering fifth form
⁷pupils estimated to be brighter than average
⁸pupils estimated to be less bright than average
As expected the KAI did not perform as well with younger populations as with adults in terms of internal reliability. However, in most cases the internal reliability was higher than the generally accepted minimum standard of 0.7. In some cases the cost in wastage was high. On this problem, this study suggests that the bulk of this wastage can be reduced by insisting that the researcher retains control of the administration of the measure. Our results show that using teachers, whose motivation in the completion of the exercise is unknown, who are inexperienced in this sort of measure and who have not been personally briefed on the whole exercise, are prime causes for unacceptable high wastage rates. Conversely, when the administrator was the researcher who had teaching experience, that wastage was at easily bearable level. In fact, if a modest stage allowance is allowed for and care taken on the use of administrators, the KAI can yield useful, meaningful and, above all, reliable data when used with secondary school pupils as young as 14 with the appropriate administrators and with the brighter 13 year olds again using skilled, motivated administrators. Nevertheless, a further word of caution is that these results show that 14 year old or the brighter 13 year old are the youngest group we think can manage this material in paper and pencil form. The results yielded three very interesting trends with respect to variations in age and intelligence of the pupils and the experience of the measure’s administrators, which can now be discussed.
Table 2 clearly shows that with as the sample’s age was younger the reliability of KAI decreased. Nevertheless, if care is taken in administration and at the youngest age (13 years old) if the brighter subjects only are used, the decrease is certainly not fatal. The evidence so far suggests that this kind of measure is reaching its limit as far as age is concerned. At first it was thought that the “work related items” would cause problems and that finding replacements might be needed. The evidence is that (a) age by decreasing age there was a gradual and general drop in reliability with no particular item standing out, in all subsets of the same age group, as an obvious item for exclusion. The items that on the whole did poorer were not the work related ones but those that required self insight, involving such terms as thorough and consistent. As we see below, intelligence, for instance, exhibited as being able to understand the items and instructions, became a factor only at 13 years old. We suggest that at around this age or younger other kinds of measures or informed estimates need to be used. For instance, measures based on skilled observers assessing selected behaviours of children undertaking tasks. Selby (1992) compared KAI results with parent and teacher assessments and found high correlations.
In the results above, Table 2 shows that at the youngest age intelligence is involved. If a comparison is made of the three studies reported for the youngest year, we find distinct improvement in the performance of the KAI with increased intelligence. Selby’s sample of eighth graders (alpha: 0.78) consisted of pupils all engaged in a “gifted” program. Taylor’s sample of the same age group (alpha: 0.67) consisted of a much more heterogeneous group with respect to intelligence. Tefft’s youngest sample (alpha: 0.34), while also intellectually heterogeneous, was from a school known to have lesser academic results than that of Taylor’s. Selby’s sample had the additional benefit of having an experienced administrator.
The KAI has consistently exhibited no correlation with intelligence in adult samples (reviewed in Kirton, 1994; for a general level versus style review, Mudd, 1996, and for school samples, Taylor, 1994a) which supports a key component of cognitive style theory that problem-solving preference and problem-solving ability are distinct. However, at some stage as one collects samples of younger and younger children the language of the measure’s instructions and items are going to become beyond comprehension. Even if not, the judgements the items are attempting to measure may be unreliable. The question is what is the age cut-off point for the average or for the brightest? The evidence here suggests that the cut-off point for most is at 14, given experienced administrators. Bright 13 years olds also managed. The next question is: is this because of a larger or greater self-awareness? The question is still open; the best we can say is that the items vary in word difficulty and some words were certainly a problem but all the items gradually lost integrity, whether they contained difficult words or not. So our tentative conclusion is that both word power and insight are implicated. It is to be stressed that intelligence only played a detectable role with the youngest children.
As the less intelligent respondents mature, however, it is fully expected they will gain the vocabulary and experience necessary to handle the KAI items as ably as those of a higher intelligence. While we do not have definitive evidence to support this hypothesis, we can point to (a) the already noted improvement in the KAI’s performance with increasing age (when unselected for intelligence) and (b) to a small test-retest study conducted by Taylor (1994b) on a sub-sample of her third formers three years on. Taylor’s original third form sample (grade 8) had an internal reliability of 0’67 (n: 72). A sub-sample (n:45) of this original group had an initial internal reliability of 0.81 and, when they were re-tested three years later after entering the sixth form (grade 11), the internal reliability had risen to 0.88. It must be noted that this sub-sample was of a higher academic ability than her whole original sample, as they had each opted for and been accepted into the academic sixth form and this is reflected in their initial KAI’s higher internal reliability. But, even so, the three years of maturity did result in a marked improvement in the internal reliability for this group, which, of course, remained at the same level of potential ability. It is reasonable, therefore, to expect similar significant improvements in the internal reliability of less academically gifted pupils as they mature and particularly after they enter the work force. We suppose’ on this evidence, that maturity is the factor that increasingly allows the respondent to “appreciate” the meaning of items.
In the results above, Table 2 also reveals an effect of the type of administrator who collects KAI data from younger populations. It is apparent, given any particular age group, that using an individual who is experienced in collecting psychometric data in a school environment produces better results with respect to KAI performance. Researchers in the educational field are often constrained to make use of teachers to assist with the collection of data. If teachers are used who do not have experience in the collection of psychometric measures, whose level of interest in the research cannot be uniformly determined, or who have not all been directly selected and briefed by the researcher, then such teachers must be classed as “inexperienced administrators” (in some cases, more accurately if more unkindly, they could be described as “indifferent administrators”) and a much higher wastage rate is to be allowed for in the sample size collected.
This paper has reported studies using the English version of the KAI and the authors wish to express their confidence that similar results will be found when validated versions of KAI in other languages are used in studies of teenagers. A preliminary sample from Slovakia (Kubes, 1994) using the validated Slovak version of KAI supports this expectation. The Slovak study produced Cronbach alpha’s of 0.80, 0.79 and0.74 for pupils of 14 (n: 34), 15 (n: 42) and 16 (n: 38) years of age respectively using two experienced administrators.
It is now possible, in light of the previous discussions, to design studies that can thoroughly test the validity of the KAI with younger populations, revisiting the more exciting hypotheses of earlier studies and hoping for more useful and informative results. Some information is already to hand as two of the reliability studies contributing to this report also involved validity. Taylor’s (1994a) primary hypothesis that for young people as with adults KAI and IQ were uncorrelated was sustained. However, she also separated the pupils from a number of third form classes (grade 8) of the same school into three groups based upon KAI scores, namely those with distinctly adaptive scores, those with distinctly innovative scores and those with middling A-I scores. Each group was then interviewed as to their respective views on school life and their responses were found to be consistent with A-I theory, in that, for instance, adaptors are more comfortable with more structure than innovators. For example, in response to the question about how they viewed the school rules, the adaptors barely seemed to notice the rules, but said if there were no rules they would have to invent some – “wouldn’t we?” Conversely, innovators gave a long list of rules they found onerous and petty. Those of middling scores responded in-between these two extremes. These experiences mirror reports made by trainers when using KAI in management courses.
Selby (1992), in primarily exploring face and concurrent validity, found with his group of gifted American eighth graders (third formers), that parents were able to estimate the KAI scores produced by their own children. He also found parents’ estimates were closer than those of teachers, leading Selby to make a number of recommendations for the use of the KAI in schools.
The results of these five studies, involving 2,461 secondary school pupils from four English speaking countries and a sixth study of 114 Slovaks, have determined the limits of the KAI’s tolerance in its use in schools. We are confident that the KAI can be successfully used with younger populations given the following guidelines:
• For general samples, the limits of the KAI seem to be reached around the fifth form/grade 10). However, the KAI can perform adequately with bright pupils as young as third form/grade 8).
• For all samples the administrator should be both familiar with the school classroom environment and experienced in the administration of psychometric measures. Such a person will get noticeably more consistent results and far less wastage especially for those below the sixth form/grade
• When designing studies in schools, it is prudent to allow a wastage rate of some 10% for the older pupils with experienced administrators’ samples. For the youngest groups, however, even with best conditions (i.e., intelligent pupils and experienced administrators), wastage may rise to 20%. With less favourable circumstances, wastage can average up to 40% with the occasional sub-group that has to be wholly discarded, generally because of unmotivated teacher intervention or the lack of adequate supervision.
• In general, it is recommended that only the total KAI scores (and not the subscale scores) be used in studies with samples below sixth form/grade 11.
The establishment of the KAI as a reliable measure with young people reopens old territories as well as new, for research into the effects of cognitive style on child development and education. The clear separation between behaviour and preferred style means that each can be measured separately and, so, more effectively. One interesting example of this is the observations of McKenna (1983; 1984) on the notion of Field Dependence (Wikin et al., 1962), which is widely regarded as a cognitive style. He conclude his analysis, after taking Adaption-Innovation Theory position into account (see also Kirton 1978) and applying it to some confusing findings, that Witkin has put forward a style theory and devised for it a level measure – hence the problems with the results.
“Cognitive Gap”, the difference between two individuals (or an individual and a group or groups) on KAI score (or average score) has been explored by MacCarthy (1988). She has shown that women who perceive colleagues to be far from them in style return higher scores on a work pressure questionnaire. We can assume that teachers and their students will feel equal pressures. If this is shown empirically, then attention can be turned not to changing the person’s preferred style, which is probably neither possible nor desirable, but to understanding the phenomenon of cognitive gap better and to improving coping skills. Hammerschmidt (1995) has measured manager performance on a problem solving exercise in a training course, showing that:
(a) when teams have virtually identical KAI scores, team conflicts disappear
(b) that when teams are given jobs that are better fits to their style preference their success rate rises
(c) that teams with similar scores tend to collaborate better
The results are spectacular: randomly selected groups (totalling over 500 managers) yielded an average success rate of 50%; of the two group types selected by KAI score (totalling nearly 500) the groups with worst fit, on all three variables related to KAI, dropped to 42% success rate; those with best fit rose to 85%. These results are food for thought for all who teach, especially as in practice it is impossible, probably undesirable, to try to match teacher and student and, therefore, sophisticated thoughts need to be generated to make best use of these data in teaching in general. We conclude with some thoughts of the kind of research that now might yield interesting results, such as:
- How early in development does cognitive style express itself?
- Does culture influence cognitive style in teenagers differently than in adults?
- In what way does the adaptive-innovative style continuum express itself among young people?
- How does the adaption-innovation concept influence relationships between teacher and student, student and student, student and parent?
- How does the adaption-innovation concept influence interactions between student and curriculum, student and school policy?
The authors should like to express their gratitude to Shell UK for its assistance towards the completion of this work. We would like to thank also those scholars who allowed us access to their research data.
Child, D. (1986). Psychology and the Teacher (4th ed.), London: Holt, Rhinehart &
Clapp, R.G. (1993). Stability of cognitive style in adults and some implications, a longitudinal study of the Kirton Adaption-Innovation Inventory. Psychological Reports, 73, 1235-1245.
Goldsmith, R.E. (1994). Creative style and personality theory. In M.J’ Kirton (Ed.) Adaptors and Innovators: Styles of Creativity and Problem-solving (2nd ed.) (pp. 34-50). London: Routledge.
Guilford, J.P. (1980). Cognitive styles: What are they? Educational and Psychological Measurement, N., 7 15-735.
Hammond, S.M. Some pitfalls in the use of factor scores: the case of the Kirton Adaption-Innovation Inventory. Personality and Individual Differences, 1986, 7, 401-407.
Hammond, S.M. (1987). Reported in M.J. Kirton, Kirton Adaption-Innovation Inventory
Manual (2nd edition). Hatfield, UK: Occupational Research Centre.
Hammerschmidt, P. The Kirton Adaption Innovation Inventory and Group Problem Solving Success Rates. Journal of Creative Behaviour, 1996, 30, 61-75
Hayes, J. & Allinson, C.W. Cognitive Style and its Relevance for Management Practice. British Journal of Management, 1994, 5, 53-71
Head, J, (1995). A New Approach to Cognitive Style. A.E.R.A. Conference: San Francisco.
Kirton, M.J. (1976). Adaptors and Innovators: A description and measure. Journal of Applied Psychology, 61, 622-629.
Kirton, M.J. (1987). Kirton Adaption-Innovation Inventory Manual (2nd ed.). Hatfield, UK: Occupational Research Centre.
Kirton, M.J. (Ed.) (1gg4). Adaptors and Innovators: Styles of Creativity and Problem-Solving (2nd ed.). London, UK: Routledge.
Kirton, M.J., Bailey, A.J. & Glendinning, J.W. (1991). Adaptors and Innovators: Preference for educational procedures. Journal of Psychology, 125 (4), 445-455.
Kirton, M.J. & Kubes, M, (1992). Kirton Adaption-Innovation Inventory – Manual: Slovak Language Supplement. Maxman: Bratislava, Slovakia.
Kubes, M. (1gg4). Unpublished data, Bratislava, Slovakia: Institute of Creativity.
McCarthy, Rosalyn. The Relationship of Cognitive Style with Coping as a Member of a Minority Group at Work. In Proceedings of the International Conference on Psychological Problems of Creative Scientific Work. Bratislava, Czechoslovakia: The Computing Centre of the Slovak Academy of Sciences, 1989
McKenna, F.P. Field Dependence and personality: a re-examination. Social Behaviour and Personality, 1983, 11, 51-5.
Measures of field dependence: cognitive style or cognitive ability? Journal of Personality and Social Psychology, 1984, 47, 593-603.
Martin, W. (1987). Reported in M.J. Kirton, Kirton Adaption-Innovation Inventory Manual (2nd edition). Hatfield, UK: Occupational Research Centre.
Messick, S.M. (Ed.) (1976). Individuality and Learning. San Francisco: Jossey-Bass.
van der Molen, P.P. (1994). Adaption-innovation and changes in social structure: On the anatomy of catastrophe. In M.J. Kirton (Ed.), Adaptors and Innovators: Styles of Creativity and Problem-solving (2nd ed.) (pp. 158-122). London: Routledge.
Mudd, S.A., Analytic review of Research on the Kirton Adaption Innovation Inventory (KAI). Social & Behavioural sciences Documents, 1986, 16, (2), (163 pages)
Kirton Adaption-Innovation Theory: Evidence Bearing on the Style/Level and Factor Composition Issues. British Journal of Psychology, 1996, 87 241-254).
Murdock, M., Isaksen, S.G. & Lauer, K.J. (1993). Creativity training and the stability and internal consistency of the Kirton Adaption-Innovation Inventory. Psychological, 72, 1123-1130.
Paechter, C, and Head, J, (1996). Power and Gender in the Staff Room. British Educational Research Journal, 22, (1) 57-69.
Prato Previde, G. (1991). Is cognitive style underlying culture? Personality and Individual Differences, 12 (1), 1 -10.
Prato Previde, G. & Massimini, F. (1984). Adattori-Innovatori: Una teoria ed una misura della creativity nelle organizzazioni. Ricerche di Psicoloqia, 3, 99-134.
Selby, E.C. (1992). The Kirton Adaption-Innovation Inventory as a tool for assessing problem-solving styles in eighth grade students. Unpublished Ph.D. thesis, Walden University.
Taylor, J.A. (1994a1. Problems and difficulties of high IQ secondary school children in relation to their cognitive style. Unpublished master’s thesis, University of Hertfordshire, UK.
Taylor, J.A. (1994b). The stability of school children’s cognitive style: A longitudinal study of the Kirton Adaption-Innovation Inventory. Psychological Reports ,74, 1008-1010.
Tullett, A.D. & Kirton, M.J. (1995). Further evidence for the independence of Adaptive-Innovative (A-I) cognitive style from national culture. Personality and Individual Differences, 19 (3), 393-396.
Witkin, H.A., Dyk, R.B., Faterson, H.F., Goodenough, D.R. & Karp, S.A., 1962, Psychological Differentiation, New York: Wiley