THE MANAGEMENT OF LEARNING: THE IMPACT OF STRATEGY CONTROL AND APPROACH TO LEARNING ON FINAL YEAR NURSING STUDENTS Robert H. Cantwell & Phillip J. Moore Faculty of Education University of Newcastle University Drive Callaghan, NSW, 2308 Paper presented at 1991 Annual National Conference of the Australian Aassociation for Research in Education, Gold Coast, November, 1991. The notion of an approach to learning is well documented in the literature (Entwistle & Ramsden, 1983; Biggs, 1987a), and refers to a learner's relative predispositions towards surface, deep and achieving approaches to learning (Biggs, 1987b). There is also a strong supporting literature suggesting that use of these approaches may have differential effects on learning outcomes (Biggs, 1979, Cantwell & Biggs, 1988; Moore & Telfer, 1990). An approach represents an amalgam of related motivational and strategic states. A deep approach, for example, indicates a bias towards learning as an end in itself, and is associated with a desire to approach tasks with a view to maximising the level of understanding. Strategically, this approach implies the use of a variety of processing strategies aimed at clarifying and integrating meanings across many levels of abstraction (Kirby, 1988; Cantwell & Biggs, 1988). Not surprisingly then, a deep approach is conventionally associated with more complex and more abstract learning outcomes (Marton & Saljo, 1976; Biggs, 1979; Moore & Telfer, 1990; Cantwell, 1990). A surface approach, by contrast, indicates a bias toward learning as a means of attaining of some extrinsic goal: the process of learning itself is not intrinsically valued. The depth and breadth of learning undertaken by those utilising a surface approach is defined by a cost-benefit analysis of the expected learning product. Operationally, this is associated with a focus on the avoidance of failure rather than on mastery learning (Ames & Archer, 1988; Biggs, 1987a). Strategically, learning is likely to revolve around the rote learning of discrete units of information. It is again not surprising to find the literature associating a surface approach with less complex learning outcomes. Both the deep and surface approaches represent intentional states for dealing with content-to-be-learned. Use of a deep approach represents a more complex learning style in the sense used by Kirby (1988) than is the case of a surface approach. It implies a greater breadth and depth in both content analysis and in strategic repertoire. Both approaches may also be associated with learning pathologies based on attentional factors. For a deep learner, the desire to extend and clarify meaning may, under certain circumstances, lead to a failure to buttress ideational learning with lower level elaboration (viz. Reder, 1980). Entwistle and Ramsden (1983) describe such learning as "globetrotting". For a surface learner, the desire to reproduce target content may lead to undue focus at the lower levels of meaning, neglecting the need to integrate content at higher levels (viz. Kirby & Cantwell, 1985). Entwistle and Ramsden (1983) characterise this kind of pathology as "improvidence". It may well be the case that these pathologies are associated with an inability or unwillingness to competently manage data beyond a focal level of analysis. It may also be the case that the concurrent use of an achieving approach, with its focus on maximising learning performance through the imposition of organisation and structure on learning, may act to minimise the occurrence of these pathologies. Learners sensitive to the need to structure and organise their learning may well be more likely to impose a hierarchical structure on content being learned. This would reflect a more efficient mode of content management. The notion of an executive strategy control mechanism influencing the kinds of procedural analyses undertaken by learners provides an important analogue to the approaches to learning paradigm. Proficient content learning has been assumed to involve the construction of complex hierarchical representations of the content being learned. For each level of content analysis, it is assumed that unique sets of processing operations come into play. In other words, the greater the complexity of the content being dealt with, the greater the complexity of the strategic repertoire being utilised. The capacity to manage complex strategic requirements in a manner congruent with complex content analysis may then be taken as an index of the efficiency of the executive strategy control mechanism. The concept of executive strategy control has been operationalised through a Strategic Flexibility Questionnaire (SFQ) (Cantwell, 1990) which purports to measure learner predispositions towards what may be termed Adaptive strategic behaviours, Algorithmic strategic behaviours and Ambivalent strategic behaviours. Strategic Adaptiveness refers to a learner's predisposition toward both planning and monitoring strategic behaviours prior to engaging in content analysis and in situ. Such adaptiveness implies a willingness to construct and maintain some kind of strategy-task congruence. Previous research has indicated that strategic adaptiveness is positively related to complex learning, and, logically, to the use of both a deep and a deep-achieving approach to content management (Cantwell, 1990). The use of a Strategy Algorithm indicates a predisposition towards some kind of fixed strategic style across all learning tasks, regardless of the nature of the content to be learned. The predisposition indicates a clear disinclination towards flexibility in either the planning or monitoring of strategic behaviours. Given the narrowness and shallowness of the surface perspective on content management, it is not surprising that the algorithmic approach to strategy use has a strong association with a surface approach to learning, and a negative relationship with complex learning (Cantwell, 1990). Strategic Ambivalence indicates a loss of strategic control over processing, or a high degree of uncertainty and confusion in generating and implementing strategic decisions. Learners scoring highly on this scale are unlikely to establish congruence between content management and procedural analysis. Previous research points to a positive relationship between ambivalence and the use of strategy algorithms, a negative relationship between ambivalence and learning outcomes at all levels of complexity, and a positive relationship between ambivalence and a surface approach to learning (Cantwell, 1990). Taken as a whole, the two sets of scales provide for complementary perspectives on executive decision-making in learning. The SPQ scales intimate a perspective over content, indicating the adoption of certain strategic behaviours to ensure congruence between intent and management of content. The SFQ scales are presumed to operationalize such management decisions by focusing on the procedural analyses undertaken by learners in determining when and how changes in strategic behaviour are perceived necessary and undertaken. The present study aims to examine the relationship between nursing students' reported approaches to learning and executive strategy control in predicting performance in the final year of a pre-service nurse education programme. To this date, a search of the literature has produced only a limited amount of studies directly addressing the effects of approaches and performance in nurse education. Heggarty (1990) indirectly addressed this issue in a comparative study of female nursing, education and management students where grade- point averages were related to scores on Schmeck, Ribich & Ramanaiah's (1977) Inventory of Learning Processes. This self- report instrument characterises learners on four sub-scales: deep processing, methodical study, fact retention, and elaborative processing. Heggarty's results showed a significant relationship between deep processing scores and grade point average. None of the remaining three sub-scales influenced GPA. Unfortunately, for those interested in nursing populations, Heggarty did not examine the effects in the different groups she used nor did she consider the possible interaction of approach with complexity of the material to be learned. Chocholowski and Chan (1991), in an examination of diagnostic reasoning among pre-service nursing students, found an indirect link between use of a deep-achieving approach to learning and the complexity of nursing diagnoses. Their study suggested that the effect of the approach to learning was mediated by both the breadth of relevant content knowledge and by the complexity of the underlying reasoning. A report by Cantwell and Moore (1990) suggested that where nurses are required to provide justification for a nursing diagnosis, nurses reporting the use of a deep approach and adaptive strategy management tended to produce higher quality and more systemically relevant reasoning. Specifically then, the study aims to address three major issues: (a) whether scores on Biggs (1987a) approach to learning scales and Cantwell's (1990) strategic flexibility scales significantly predict performance in undergraduate nurse education; (b) whether there are significant interactions between the approaches to learning scales and strategic flexibility scales in predicting performance in undergraduate nurse education; and (c) whether the relationship between the various approach scales, flexibility scales and nurse learning is consistent across subjects of differing degrees of difficulty. METHOD Subjects Subjects for the study included 207 final year nursing students from a New South Wales university. The course is of three years duration, leading to a Diploma of Health Science (Nursing). Mean age of the group was 22.7 years (sd = 5.41), with an age range of 19 to 48 years. The sample included 172 females and 35 males. Materials Study Process Questionnaire The Study Process Questionnaire (SPQ) (Biggs, 1987b) is a 42 item self-report questionnaire requiring subjects to indicate, on a Likert scale of 1 to 5, their strength of agreement with statements about how they approach various aspects of their learning. The responses may be summed to yield three scale scores: a Surface Approach score, a Deep Approach score and an Achieving Approach score. Each scale score consists of a composite of seven items relating to motives for learning, and seven items relating to the associated learning strategies presumed to flow from these motives. The approach scores thus represent an amalgam of related motivational and strategic states. Strategic Flexibility Questionnaire The Strategic Flexibility Questionnaire (SFQ) (Cantwell, 1990) is a 21 item self-report questionnaire requiring subjects to indicate, also on a five-point Likert scale, their strength of agreement with statements about their management of learning strategies. The responses may be summed to yield three scale scores: an Adaptive strategy management score, an Algorithmic strategy management score, and an Ambivalent strategy management score. Each scale consists of seven items. Measures of Academic Learning Learning outcome measures were based on subjects' final grades for their third-year course. Grades were obtained for five modules completed during third-year: Nursing Practice, Health Studies, Anatomy and Physiology, Psychosocial Studies and Professional Studies. Outcome measures were derived by assigning a value 1 to a failure, 2 to a pass, 3 to a credit, 4 to a distinction, and 5 to a high distinction. A final Grade-point average (GPA) was obtained by calculating the mean score for all five modules completed. In order to ascertain the perceived degree of difficulty of each of the five modules, a set of flashcards was prepared. The name of each module was written on a separate card. Instructions accompanying the cards requested Lecturers to rank modules from the easiest to most difficult to learn. Three further measures of academic learning were obtained from within the Nursing Practice module. Two classroom tests, one from a Mental Health Package, and one from a Medical-Surgical Package (Homeostasis) were completed. The Mental Health test was a 30 item multiple choice test assessing subjects detailed knowledge in areas of mental disorders. The Homeostasis test was a short answer test based on a clinical case study of diabetes mellitis. Neither test assessed higher-order learning skills. Both tests were constructed and scored by faculty of the Department of Nursing Studies. Results were available for 196 subjects of the original sample in the case of the Mental Health test, and for 198 subjects in the case of the Homeostasis test. Subjects also completed an essay in an area of Mental Health. The topic requested subjects to discuss the impact of major tranquillisers in the treatment of schizophrenic disorders, with particular reference to Tardive Dyskinesia. As participation in the essay component was voluntary, a sub-sample of 128 subjects only completed the task. The essay was set and scored by the researcher. Lecturing staff involved in the Mental health Package confirmed the face validity of the essay question. Procedures All subjects completed the SFQ and SPQ during normal lecture time. Performance measures relating to academic grades were collected from University records. Package tests were administered in class time by faculty of the Department of Nursing Studies. Results were subsequently made available from Departmental records. The essay was administered as a final question for Mental health test. Subjects were told that the essay did not contribute to their grade in the test, but were encouraged by faculty staff to treat the essay as practice for the final exam. The perceived level of difficulty for each module was assessed using a card-sorting technique. Seven lecturing staff involved in the third-year nursing programme were given the cards labelled with the names of the third-year modules. The lecturers were asked to rank the modules from easiest to most difficult to learn (1 being most difficult, 5 being easiest). Higher scores represented perceived easier modules. The respective mean levels of difficulty were as follows: Health Studies, 4.86; Nursing Practice, 3.14; Psychosocial Studies, 2.86; Professional Studies, 2.57; Anatomy & Physiology, 1.57. Scoring The completed essays were scored for structural complexity based on Biggs and Collis' (1982) SOLO Taxonomy. The taxonomy yields a score of between 1 and 5, depending upon the complexity of the data included, and the level of abstraction reached in the interpretation of data. At the lowest level of the Taxonomy, the Pre-structural level, subjects do not provide any relevant dimension to their discussion of the topic. No responses were received at this level. At the Unistructural level, subjects typically incorporate only one relevant dimension in their response. This dimension relates to the level of information: subjects who "list" set benefits and/or risks would be scored at this level. Subject 102 provides an example: The use of major tranquilliser in the treatment of schizophrenics can cause many side effects. Some of these include weight gain, walking with a gait, and swelling of the mouth which causes an inability to speak properly. However the use of these drugs allow for the patient to have better control over himself/herself enabling a more productive life to be led. At the Multistructural level of response, subjects typically provided more integrated accounts of each part of the question, but tended not to view the issue of risk vs. benefit as a single question. The responses go beyond detailed "list- writing", but the sections within the response remain unintegrated. An example is in the response of subject 134: Schizophrenic disorders are extremely frightening both to those that are effected by them and their family/friends. However, with the use of major tranquilliser, the symptoms of schizophrenia may be eliminated, although the side effects of medication can sometimes appear more problematic than the actual illness. So schizophrenia does occur in young people who may be about to enter university or the work force, the availability of major tranquilliser can be of huge benefit. Major tranquilliser allow for the symptoms - hallucinations, delusion, to be reduced and assist the person affected to resume a normal life. Although these major tranquilliser reduce the symptoms of schizophrenia, they also carry with them numerous side-effects such as tardive dyskinesia. This is when the person may experience extreme motor restlessness, an inability to remain still and tongue rolling. This can be most distressing both for the patient and his family and friends. However, when these symptoms are compared with the controlled symptoms of hallucinations and delusions, the benefits are less severe. Also, if these side- effects begin to occur, the patient may be taken from the medication and a new medication tried or their dosage reduced. The side-effects may also be reversed with other medications, such as cogentin. Major tranquillisers allow those affected to experience a normal life, although there will be some side-effects to the medication. At the Relational level of response, subjects are likely to go beyond a discussion of the risks and benefits aspect of the question, and to incorporate some higher-level, unifying dimension. In this case, the relating perspective is likely to involve the question of balance. Case 191 provides an example: Schizophrenia is a mental illness which has no cure, but some of its side-effects can be alleviated by major tranquilliser such as serenade, melleril and largactil. These tranquillisers have basically an antipsychotic effect on users which tend to cause drowsiness, fatigue, dryness of mouth, postural hypotension and so on. Because the majority of major tranquilliser depress the central nervous system, there are many adverse effects and side-effects, one of those being tardive dyskinesia. Tardive dyskinesia is recognised by repetition of movements or activity. these may range from a constant tapping of the foot to a repetitious arm swinging movement. Tardive dyskinesia is also recognised by causing bulging of the eyes, puffiness in the face and neck and the tongue hanging out of the mouth. Therefore it can be summarised that the use of major tranquilliser in treating a patient with schizophrenia, especially paranoid schizophrenia, is essential. They have a calming and quietening effect on the taker. However, the disadvantages associated with these drugs is that they may cause depression, and ultimately, with prolonged use, tardive dyskinesia, which has undesirable traits. In treating a patient with schizophrenia, it is important to monitor the effect of the drug and consider whether its effectiveness is worth the risk of tardive dyskinesia. The patients response to schizophrenia related drugs is very individual, and therefore cannot be generalised. The final level, Extended Abstract, implies a perspective that takes the argument of the essay beyond the immediate context. There were no responses in this category. Essays were scored by the present author. A sub-sample of 30 essays were also scored by a trained colleague. Inter-scorer reliability of 90% was achieved, with differences resolved by discussion. RESULTS Relationship between SFQ and SPQ Scales Means and standard deviations (in parentheses) for the SPQ and SFQ scale scores were as follows: Surface Approach 46.60 (6.59), Deep Approach 44.57 (7.78), Achieving Approach 42.94 (7.69), Adaptive Strategy Management 20.12 (4.66), Ambivalent Strategy Management 20.41 (5.35), Algorithmic Strategy Management 23.04 (5.09). The relationships within and between the two sets of scales are summarised in Table 1. These correlations fall in the directions predicted by the theoretical positions underlying the paper. adaptive strategy management is closely linked with both the deep and achieving approaches to learning, and is negatively linked with both algorithmic and ambivalent strategy management. Algorithmic strategy management is positively linked with a surface approach to learning, and with ambivalent strategy management. It is also negatively linked with adaptive strategy management. Ambivalent strategy management was also positively linked with a surface approach and algorithmic management, and negatively linked with adaptive strategy management. Methods of Data Analysis Data for academic performance were analysed in three stages. Initial analysis sought to examine the individual relationships between the sub-scales for both instruments and performance on the various measures of academic learning. These analyses were conducted using correlational techniques. The second stage of analysis was concerned with assessing the relative impact of each sub-scale within the SFQ and SPQ instruments respectively on performance on the measures of academic learning. These analyses were conducted using multiple regression techniques. The final stage of analysis was concerned with extracting possible interactions between the SFQ and SPQ sub-scales in explaining performance on the measures of academic learning. These analyses were conducted utilising analysis of variance techniques. All analyses were conducting using the CSS/PC package (Statsoft, 1987) Correlational Analyses The correlations between the SFQ Scales and performance measures was most clearly marked by the significant negative link between ambivalent strategic management and performance across all modules and GPA. Neither adaptive strategy management nor algorithmic strategy management achieved significant associations with performance, although the correlations for both scales were in the predicted directions. For the SPQ scales, the results for overall performance, as reflected in the GPA, revealed significant positive associations with the use of both a deep and an achieving approach, and a near significant negative association with the use of a surface approach. For the individual modules, a surface approach was consistently, but non-significantly, associated with poorer performance in all areas, while both the deep and achieving approaches had significant positive links with two modules (Anatomy & Physiology and Psycho-social studies), and non significant positive links with all other modules. The correlations for both sets of scales with academic grades are presented in Table 2. The results of the correlational analyses shed only limited light on the question of perceived difficulty of courses and the effect of predispositions on either set of scales. Both deep and achieving were associated with better performance on two of the three more difficult modules (Anatomy & Physiology and Psychosocial Studies), but failed to influence the perceived second most difficult module, Professional Studies. All other scales were uniform in impact, with only the consistently negative effect of ambivalent strategy management attaining significance. The results of the analyses for the package tests and essay from Nursing Studies confirmed the general patterns found for the overall grades. A surface approach was generally harmful to performance for both knowledge tests (near significantly for Mental Health test, and significantly for the Homeostasis test) and non-significantly for the essay. A deep approach was only helpful in relation to the more complex essay task. Adaptive strategy management had no effect on lower level knowledge tests, but was positively linked to performance on the essay. Ambivalent strategy management was generally harmful to performance, but was significantly so only in relation to the Mental Health package test. These results are summarised in Table 3. The results relating to the Mental Health Essay give additional support to the possibility that for these subjects, a deep approach to learning is helpful where explicitly complex learning tasks are to be undertaken. Logically, also, where this complexity is inherent in the task, it is clearly advantageous to adaptively manage the learning strategies brought to the task, particularly when the strategy management is associated with the need to co-ordinate a broader array of strategies associated with the use of a deep approach. The possibility exists in relation to this finding that conscious use of adaptive management may only occur where (1) the presidposition exists, (2) the task is sufficiently complex to warrant conscious co-ordination, and (3) the strategy repertoire available to learner is of sufficient breadth to productively allow for adaptive change. Multiple Regression Analyses In order to investigate the independent contribution of the SPQ scales and the SFQ scales toward explaining performance in the final year modules, two sets of multiple regression analyses were conducted. The results of these analyses are illustrated in Figure 1. Beta coefficients, probability levels and variance explained are reported for each equation in Table 4 and Table 5. The analysis for the SFQ scales produced significant negative paths between ambivalent strategy management and all individual modules as well as overall performance. Neither adaptive strategy management nor algorithmic strategy management significantly predicted any dimension of performance. Beta coefficients, probability levels and variance explained are reported for each equation in Table 6 and Table 7. The analyses for the approaches to learning scales produced significant paths for both a surface approach and an achieving approach. The analyses yielded significant negative paths between a surface approach and all individual modules (with the exception of Professional Studies) and overall performance. An achieving approach yielded significant positive paths to all modules (Professional Studies p < .1) as well as to overall GPA. No impact for the use of a deep approach was evident for any individual module, nor for overall performance. The regression analyses for final grades yield two important conclusions: how a learner approaches the task of learning appears to have some impact on the quality of the learning outcome, and how coherently a learner operationalises strategic decisions seems also to impact on the quality of learning outcome. In relation to the approaches to learning, there appears to be a strong negative link between learning outcome and the use of a surface approach, independently of the apparent difficulty level of the module (with the exception of Professional Studies). Additionally, there appears to be a positive link between a commitment to institutional goals and organisation of learning (as reflected in the use of an achieving approach) and successful performance. This finding, in conjunction with the failure of a deep approach to influence learning outcomes, may be suggestive of an institutional effect on learning in nursing: that once students perceive the goals and structure of the course, learning methods can be efficiently adjusted to maximise performance without the need to generate personalised representations of meaning. This interpretation may also offer some explanation of the failure of the adaptive and algorithmic scales to impact on performance. Provided the learner has some clear perception of the requirements of a course, and provided the course does not impose unduly high levels of complexity in learning objectives, it may well be sufficient to just avoid confusion in strategy management. An additional explanation of the failure of either a deep approach or of adaptive and algorithmic strategy management to impact on performance may relate to the the possibility of interactive rather than independent effects. This possibility is explored in the final sectionof the analyses. The regression analyses relating to the two package tests and essay from the Nursing Practice module are summarised in Tables 6 and 7. The most salient findings from these analyses are the positive effects of a deep approach and adaptive strategy management, and the negative influence of an achieving approach in the equations relating to the essay. the results also indicate a negative influence of a surface approach and for ambivalent management on both tests. These findings provide some qualification to the conclusions discussed in relation to the final grades. The finding that a deep approach and adaptive management can be helpful in those learning tasks explicitly requiring the use of higher-order processing skills may indicate the importance of assessment modality and task requirement in drawing out the use of these approaches and strategy management styles (viz. Thomas, npd; Thomas & Bain, 1984). It seems also important that in contrast to the findings for general grades, the use of an achieving approach in dealing with the demands of essay writing was damaging to performance. This finding may add weight to the suggestion that successful perception of institutional goals may not have been consonant with higher quality learning outcomes. Analyses of Variance (ANOVA) In order to investigate possible interactive effects between the three approaches to learning scales and the three strategic flexibility scales, a series of 2 (Strategy Management) x 2 (Approach) ANOVAs were conducted for each individual module, overall GPA, as well as the two Nursing Practice tests and the essay. Subjects were assigned as High or Low on each scale on the basis of median splits on the respective scale scores. It is noted that due to the large number of ANOVA calculations being conducted the possibility of a Type II error is magnified. Consequently, interpretations of significant findings will be based on general patterns, with less emphasis being placed on isolated significant findings. Additionally, following Hays (1981), the interpretation of significant interactions is based on planned comparisons rather than post hoc techniques. Hays suggests this to be important in minimising the possibility of Type II errors, particularly in situations where the nature of the interaction may be of theoretical importance. The results of all ANOVA's relating to final grades are summarised in Tables 8, 9 and 10. Major interactions are illustrated in Figures 3,4 and 5. The results of ANOVAs relating to package tests and the essay are presented in Tables 11, 12 and 13, with the interactions illustrated in Figures 6, 7, and 8. Means and standard deviations for all ANOVAs are presented in Tables 14 to 19. For convenience, the Figures and Tables relating to the Anova's are included at the end of the paper. Adaptive Strategy Management and Approaches to Learning Adaptive x Surface: The 2 x 2 Anova's for Final Grades revealed no significant effects for either a surface approach nor for adaptive strategy management. There were however, significant interactions between surface and adaptive for Psychosocial Studies (F (1, 203) = 4.78, p < .05), Professional Studies (F (1, 203) = 4.89. p < .05) and for overall GPA (F (1, 203) = 4.03, p < .05). On each occasion, there was a significant effect for the level of adaptiveness in conjunction with a low surface approach to learning, and a significant effect for the degree of "surfaceness" in association with a high level of adaptiveness. It would seem that adaptive strategy management harms performance when associated with a high surface approach, and helps performance when associated with a low surface approach. The ANOVAs relating to Package tests and essay revealed main effects for a surface approach for both the Mental Health test (F (1, 192) = 4.62, p <.05) and the Homeostasis test (F (1, 194) = 3.85, p <.05), and a main effect for adaptive management in relation to the Mental Health Essay (F (1, 124) = 3.57, p = .05). Use of a surface approach was harmful to performance on both tests, while adaptive strategy management enhanced performance on the essay. There were no other main effects or interactions Adaptive x Deep: The analyses for Final Grades revealed an inconsistent positive effect for the use of a deep approach, significantly in relation to Psychosocial Studies (F (1, 203) = 8.19, p < .01), and non-significantly in relation to Anatomy and Physiology, Professional Studies and overall GPA. It was noteworthy that these three modules represented the most difficult modules according to Lecturer rating. While these findings qualify the possibility of an institutional effect on the depth of learning, the failure of the deep approach to have a significant effect on GPA does raise the possibility of a ceiling effect on course difficulty. The failure of a deep approach to impact significantly may well explain the failure of adaptive strategy management to affect performance, either as a main effect or in interaction with a deep approach. Without the use of complex strategy repertoires associated with deep processing, the need for high levels of adaptiveness in management is reduced. Neither a deep approach nor adaptive management produced significant main effects nor interactions in relation to the Package tests nor the Mental Health essay, although there were weak positive effects for both in relation to the essay. The lack of strength of the main effects may be attributable to the shared variance between the two dimensions of learning. It should also be noted that the overall performance of the subjects on the package essay was poor - midway between unistructural and multistructural (Mean = 2.53, sd .61). The difference between subjects low on both scales and those high on both scales was only a half standard deviation unit. Adaptive x Achieving: The impact of both adaptiveness and achieving was weak across all subject areas, with no main effects for either dimension reaching significance. Non- significant interactions between Adaptiveness and Achieving were apparent for Nursing Practice, Psychosocial Studies and GPA (all p's < .1). These interactions all revealed a tendency for learners low on adaptiveness to perform better in association with a high level of achieving. The analyses for the Package tests and essay yielded a significant main effect for adaptive strategy management in relation to the Mental Health essay (F (1, 124) = 4.94, p <.05), with higher levels of adaptiveness associated with better performance. No other main effects or interactions were evident. Algorithmic Strategy Management and Approaches to Learning Algorithmic x Surface: The analyses for Final Grades produced no significant main effects for either algorithmic strategy management or a surface approach to learning, nor did any interactions between algorithmic and surface attain significance. The analyses did produce a significant main effect for a surface approach in relation to the Mental Health test (F (1, 192) = 4.81, p <.05) as well as a near significant main effect for surface in relation to the Homeostasis Test (F <.1). On both occasions, the reported use of a surface approach was associated with poorer performance. there were no other main effects or interactions in these analyses. Algorithmic x Deep: The analyses revealed significant and positive main effects for a deep approach in relation to Psychosocial Studies (F (1, 203) = 7.42, p < .01), and non- significant positive effects in relation to Health Studies, Anatomy and Physiology and GPA. There were no significant effects for algorithmic strategy management, nor any significant interactions between algorithmic management and a deep approach. The analyses relating to both Package tests revealed no significant main effects nor interactions. However, the analyses did yield a significant main affect for a deep approach in relation to the essay (F (1, 124) = 4.14, p <.05), as well as a significant interaction between use of a deep approach and algorithmic strategy management (F (1, 124) = 3.57, p =.05). While the use of a deep approach was associated with better performance on the essay, the reported use of algorithmic strategy management was beneficial to performance only when associated with a deep approach. Subjects reporting an algorithmic approach to strategy management without the concurrent use of a deep approach were significantly disadvantaged in relation to performance on the Mental Health essay. The interaction is illustrated in Figure 6. Algorithmic x Achieving: With the exception of a main effect for an achieving approach for Psychosocial Studies (F (1, 203) = 3.87, p <.05), in which subjects higher on achieving outperformed those lower on achieving, these analyses revealed no significant main effects of interactions. No significant main effects nor interactions were evident in the analyses of either the Package tests or the Mental Health essay. Ambivalent Strategy Management and Approaches to Learning Ambivalent x Surface: The impact of strategic ambivalence was powerful across all modules ( Nursing Practice, F (1, 203) = 14.91, p < .001; Health Studies, F (1, 203) = 3.80, p < .05; Anatomy & Physiology, F (1, 203) = 9.92, p < .01; Psychosocial Studies, F (1, 203) = 4.95, p <.05; Professional Studies, F (1, 203) = 6.40, p < .05) and overall GPA (F (1, 203) = 12.27, p < .001). For each effect, the influence of a high level of ambivalence was damaging. There were no significant main effects for a surface approach, nor were any significant interactions between ambivalence and surface evident. The analyses for the Package tests and Mental Health essay revealed no significant main effects nor interactions. However, there was a non-significant tendency for both a surface approach and ambivalent strategy management to harm performance in both tests, and for surface to harm performance in the essay. Ambivalent x Deep: Ambivalent strategy management again related negatively to performance in all modules (Nursing Practice, F (1, 203) = 15.12, p <.001; Health Studies, F (1, 203) = 4.03, p < .05; Anatomy & Physiology, F (1, 203) = 9.94, p < .01; Psychosocial Studies, F (1, 203) = 4.13, P , <.05; Professional Studies, F (1, 203) = 7.16, p < .05), as well as GPA (F (1, 203) = 12.29, p < .001). There was a main effect for a deep approach on Psychosocial Studies only (F (1, 203) = 7.03, p < .01), with a high deep approach yielding a higher grade. There were no other main effects for deep, nor were there any significant interactions between deep and ambivalence. While there was a general, but non-significant, tendency for ambivalent strategy management to harm performance across both Package tests and the Mental Health essay, the analyses did reveal a significant interaction between ambivalence and a deep approach in relation to the Mental Health test (F (1,192) = 5.61, p <.05). The interaction is primarily explained by the significantly poorer performance of subjects reporting ambivalent strategy management in conjunction with a lower predisposition towards the use of a deep approach. This interaction is illustrated in Figure 7. Ambivalence X Achieving: As was the case with the previous approach scales, ambivalent strategy management provided significant main effects for all modules and GPA (Nursing Practice , F (1, 203) = 14.98, p <.001; Health Studies, F (1, 203) = 4.13, p <.05; Anatomy & Physiology, F (1, 203) = 10.41, p < .01; Psychosocial Studies, F (1, 203) = 4.54, p < .05; Professional Studies, F (1, 203)= 6.52, p < .05; GPA, F (1, 203) = 12.41, p < .001). No other main effects or interactions were evident. A general tendency for ambivalent strategy management to be harmful to performance in both Package tests (significantly in relation to the Mental Health test (F (1,192) = 3.79, p =.05)) and in relation to the Mental Health essay was again apparent. The analyses also yielded a significant interaction between the use of an achieving approach and ambivalent management in relation to performance on the Mental Health test (F (1, 192) = 4.05, p <.05). In this instance, the interaction was explained by the superior performance of subjects low on both scales. The interaction is illustrated in Figure 8. The ANOVA results in relation to final grades are suggestive of two major conclusions: that ambivalent strategy management, regardless of the content perspective used, is extremely damaging across all areas of learning; and that adaptive strategy management is selectively helpful/harmful. In relation to the effects of ambivalence, the results clearly support the hypothesis that some kind of executive strategy control is necessary if any approach to learning is to be successfully operationalised. The impact of an adaptive approach to strategy management was selective, particularly in relation to the interaction between adaptiveness and a surface approach. Given the theoretical description of the surface approach in terms of its limited strategic repertoire, it is not surprising that highly surface learners professing high levels of adaptiveness should perform significantly less well than low surface, highly adaptive learners. Where the repertoire is limited, any attempt at adaptiveness should logically associate with an increase in ambivalence. Given the consistent damaging effect of ambivalence across all modules, this interpretation appears appropriate. The results also suggest that algorithmic strategy management has a neutral effect on performance, regardless of approach taken. This again may be suggestive of an institutional effect: the gains from non-ambivalence appear to be greater than consequential gains of adaptive or algorithmic strategy control. The ANOVA results in relation to the Nursing Practice assesments both confirm and qualify aspects of these conclusions. Both a surface approach and ambivalent strategy management continued to be harmful to performance, particularly in relation to the two package tests. The results in relation to the Mental Health essay, however, did provide some qualification to the earlier conclusions. On the one hand, a deep approach was clearly beneficial to performance in this kind of task. This is consistent with the predictions of the study. On the other hand, where simply avoidance of strategic ambivalence appeared to be sufficient in the analyses of final grades and of the two package tests, the results relating to the Mental Health essay suggest that some kind of coherent strategy management is necessary in specific tasks requiring higher-level strategy use. In this instance, adaptive strategy management was clearly helpful to performance. At the same time, the interaction between algorithmic strategy management and use of a deep approach in explaining performance on the essay suggested that the nature of the management algorithm may be also important in explaining performance. Where algorithmic management is not associated with the concurrent use of a deep approach, performance appears to suffer significantly. This may suggest that the management algorithm used may incorporate some dimensions of adaptivity as inherent parts of the algorithm itself. DISCUSSION The study sought to address three questions in relation to learning in nursing: whether the approaches to learning scales and strategic flexibility scales provide significant predictors of performance in nurse education; whether the predictive power of both sets of scales may be enhanced through analysis of their interactions in predicting performance; and whether the predictive value of both sets of scales is influenced by the perceived difficulty level of the subjects being studied. The independent contribution of each scale within the SPQ and SFQ instruments toward explaining performance in nurse education were assessed by correlational analysis. The results were reasonably encouraging in terms of the predictive utility of both sets of scales. In relation to the SPQ, reported surface learning was generally, but not significantly, harmful to performance. This was surprising, as theory had quite strongly predicted a negative association. A reported use of a deep approach was generally helpful, particularly in two of the perceived more difficult units, in overall GPA, and in the Mental Health essay. This was also consistent with theory. The reported use of an achieving approach was mixed in effect: positive in relation to course modules, and negative in relation to the Mental Health essay. The contribution of individual scales within the SFQ instrument was most marked by the consistently damaging influence of strategic ambivalence across all performance measures. Adaptive management was helpful only in relation to the Mental Health essay. These results yield a number of important theoretical implications. While the intercorrelation between deep and adaptive is high (.61), the results suggest that there is not a 1:1 concordance of impact. A deep approach is fundamentally concerned with imposing a certain perspective over content to be learned. In particular, the approach is generally associated with the generation of more structurally complex levels of understanding. At the same time, the degree of complexity in understanding (beyond rote memorisation) is equivocal. There may indeed be a threshold level of difficulty beyond which some kind of "deeper" learning is required. Beyond this, more task specific factors may determine the ultimate level of abstraction required. In the case of the present data relating to Final Grades and Package tests, there are grounds for suggesting that the requisite level of complexity is of sufficient magnitude to enforce a "beyond rote learning" perspective, but not sufficient (c.f. Mental Health essay) to implicate the need for a higher-level perspective, and with this, the orchestration of a more complex strategy repertoire. Further support for this interpretation comes from the results relating to the achieving approach. As the achieving approach imposes no specific content perspective, the utility of the approach in supporting learning in all modules may relate to the perception by learners of institutionally successful learning routines. While such routines may well involve "more than rote learning", the failure of adaptiveness to significantly help performance outside of the Mental Health essay, and the damaging effect of the achieving approach to performance on the essay, may be indicative of the limitations of these routines, and of the need for greater orchestration of strategies when higher-level learning is required. The overall predictive power of both instruments, as opposed to the discrete influence of each scale within the instruments, was assessed by multiple regression techniques. The results lend support to the conclusions argued in the previous paragraph. The failure of the deep approach, in a common equation with the surface and achieving approaches, to significantly predict performance in any module, combined with the significant negative influence of surface and positive influence of achieving, suggests again a level of complexity that goes beyond rote learning, but which is not of sufficient complexity to go beyond institutionally sanctioned routines. Ultimately this suggests, as the neutrality of both the algorithmic and adaptive approaches, and the significant negative effect of the ambivalent approach, indicate, learning in nursing minimally requires some kind of coherent strategy for the management of learning. The limitation to this, however, is that the coherent management emerges from a need to go beyond rote learning rather than an explicit need to take learning processes to levels conventionally implied by a deep approach. The analyses of variance sought to isolate specific interactions between the two sets of scales that may provide additional predictive information in explaining performance in nurse education. The data suggested two theoretically important sets of interactions: those relating to the interactions between surface and adaptive for Final Grades, and the interaction between deep and algorithmic in relation to the Mental Health essay. In both instances, a question of congruence between intent, strategy selection and strategy management appears to be important. The data did not offer clear cut support for the prediction that performance in more complex modules would be advantaged by the use of a deep approach and adaptive management. While the correlational analyses, and to some extent the main effects emanating from the ANOVAs supported the utility of a deep approach in relation to two of the three reportedly more difficult subjects, the evidence was not compelling. It may be more profitable to ultimately define complexity in more precise terms than was possible in this study. Indeed, it may well be the case that the "range" of difficulty alluded to by the raters was in fact quite small. The evidence from the study does point far more strongly to negative factors relating to course performance than it does to positive factors. Implications for Nurse Education The extent to which this study has implications for nurse education depends to a large extent on the objectives of the nurse education programme. If it is considered, at least at the level of a pre-service Diploma, a minimal objective to produce graduates capable of implementing at least moderately complex nursing processes and decisions, then to a large extent the programme is meeting its objectives. The damaging impact of a surface approach and of ambivalent strategy management clearly testify to this. If, however, it is the aim of nurse education to produce graduates capable of higher- level analysis and reflection, then the data appear less supportive. While there were some associations between a deep approach and some of the perceived more complex modules, as well as overall GPA, the data did not support a general reinforcement of deep and adaptive approaches to learning and strategy management as prime indicators of competence. Two implications flow from this. On the one hand, it would appear important that the nurse education programme pursue assessment modalities more sensitive to higher quality learning outcomes. In the present study, the use of the expository essay with qualitative measurement did appear to tap into this level of processing. The contrasting impact of the achieving approach in relation to general assessment and the Mental Health essay would support this. If nurses are aware, for example, that the submission of formulaic type nursing care plans are acceptable indices of understanding, then it is unlikely that such nurses will focus their learning activities on the development of higher-level and more complex strategic repertoires. On the other hand, the data also suggest that if higher level learning outcomes are to be encouraged, some kind of metacognitive training may need to be considered in order to change the way many nurses both approach the problem of content analysis (deep vs. surface) and approach the problem of the management of learning (controlled (adaptive/ algorithmic vs. ambivalent). Such programmes have significant implications for the way content is delivered and assessed, and for the ways in which nurses are encouraged to treat and to manage the content. REFERENCES Ames, C. & Archer, J. (1988). Achievement goals in the classroom: Student learning strategies and motivational processes. Journal of Educational Psychology, 80, 260-267. Biggs, J. (1988). Approaches to learning and essay writing. In R. Schmeck (Ed.) Learning Styles and Learning Strategies. New York, Plenum Press. Biggs, J. (1987a). Student Approaches to Learning and Studying. Hawthorn, ACER. Biggs, J. (1987b). The Study Process Questionnaire: User's Manual. Hawthorn, ACER. Biggs, J. (1979). Individual differences in study processes and the quality of learning outcomes. Higher Education, 8, 381-394. Biggs, J. & Collis, K. (1982). Evaluating Student Learning: The SOLO Taxonomy. New York, Academic Press. Cantwell, R. (1990). Flexibility and learning. In B. Ross (Ed.) Research and Development in Higher Education, Volume 13: Teaching for Effective Learning. Sydney, HERDSA. Cantwell, R. & Moore, P. (1990). Flexibility and learning in nursing. Paper presented at the AARE Conference, Sydney University, Sydney, November, 1990. Cantwell, R. & Biggs, J. (1988). Effects of bilingualism and approaches to learning on the writing and recall of expository text. In M. Gruneberg, P. Morris, & R. Sykes, (Eds.), Practical Aspects of Memory, Volume 2: Clinical and Educational Implications. London, Wiley. Chocholowski, K. & Chan, L. (1991). Diagnostic reasoning processes in nursing. Paper to be presented at the AARE Conference, Gold Coast, November. Entwistle, N. & Ramsden, P. (1983). Understanding Human Learning. London, Croom Helm. Haggarty (1990). Information processing strategies of women enrolled in traditional and non-traditional college majors. Western Journal of Nursing Research, 72, 525-536. Hays, W. (1981). Statistics. New York, Holt-Saunders. Kirby, J. (1988). Style, strategy and skill in reading. In R. Schmeck (Ed.) Learning Styles and Learning Strategies. New York, Plenum Press. Kirby, J. & Cantwell, R. (1985). Use of advance organisers to facilitate higher-level comprehension. Human Learning, 4, 159-168. Marton, F. & Saljo, R. (1976). On qualitative differences in learning: Outcome and process. British Journal of Educational Psychology, 46, 4-11. Moore, P. & Telfer, R. (1990). Approaches to learning: Relationships with pilot performance. Journal of Aviation/ Aerospace Education and Research, 1, 44-58. Reder, l. (1980). The role of elaborations in the comprehension and retention of prose. Review of Educational Research, 50, 5-53. Schmeck, R. Ribich, F. & Ramanaiah, N. (1977). Development of a self-report inventory for assessing individual differences in learning processes. Applied Psychological Measurement, 1, 413-431. Statsoft, (1988). Complete Statistical System, Rel 2.1. Thomas, P. & Bain, J. (1984). Contextual dependence of learning approaches: The effects of assessments. Human Learning, 3, 227-305. Thomas, P. (npd.). Individual differences in learning for assessments.