Even though the belief that assessment influences student learning is widely proclaimed, attempts in field settings to influence learning in desirable ways using assessment have not been very successful e.g., [1
]. One reason may be that even thoughtfully conceived attempts are not informed by a sufficiently theoretically grounded understanding of how assessment influences learning.
While there is much literature relating assessment and learning, there is currently no satisfactory theory or model offering support to the "assessment for learning" endeavour. Calls have been made for a greater role for theory in researching assessment [2
] and in understanding what interventions work under which conditions [3
]. There has only recently been an attempt to formalize existing knowledge by classifying the learning effects of assessment [4
]. A distinction is drawn between pre-, pure and post-assessment learning effects that respectively impact learning before (e.g., study behaviour), during (e.g., portfolios, testing effect) and after (e.g., feedback) assessment.
A validated model explaining (rather than describing) how assessment influences learning could benefit the design of, and research into, assessment for learning. Self-regulation theory has been invoked to explain the effects of assessment [5
]. Other empirical work resulted in the proposal of a "grade point average perspective" [9
], while a synthesis of literature resulted in a model explaining students' study strategies when preparing for classroom tests [10
]. None of these models or frameworks have been further validated, however.
Although not typically designed with learning aforethought, summative assessment strongly influences learning. We recently proposed a model explaining the pre-assessment learning effects of summative assessment [11
]. According to the model (Figure ), task demands and system design influence the quality and regulation of learning. These effects are mediated by a mechanism that involves impact appraisal, response appraisal, perceived agency and interpersonal factors. Thus, when contemplating an upcoming assessment event, students may consider the likelihood that assessment will impact them (positively or negatively) and what the magnitude of that impact is likely to be. They may consider the efficacy of any given learning response in bringing about a desired outcome, the costs of that learning response and how the desired or likely outcome relates to their values. Their perceptions of their ability to bring about a particular outcome may also influence their learning, as may their perceptions of the opinions of referents like lecturers and fellow students and their motivation to comply with those perceptions.
Figure 1 A model of the pre-assessment learning effects of summative assessment .
The relationship between assessment factors, mechanism factors and learning effects is not a simple one-to-one-to-one relationship. In any given assessment context for any given student, one or more assessment factors will influence one or more aspects of learning, acting via one or more facets of the mechanism. Different students can react in different ways to any given assessment event, depending on what factors in their mosaic of academic and personal motivation enjoy prominence at that time. Equally, any one student can react in different ways to different assessment events, as factors enjoying prominence in their motivational mosaic wax and wane. See Additional File 1
: Additional material_Illustrative quotes.pdf for more extensive information on the model.
Whilst this model is grounded in empirical data, it too has yet to be validated. Validation is necessary before any model can meaningfully inform the design of, or research into, the learning effects of assessment. The question now is whether the model can be shown to be valid outside of the immediate context in which it was derived and how to approach this.
To explore this, a clinical setting i.e., a different educational context to that in which the model was derived, held appeal. Furthermore, surprisingly little has been written about the impact of performance assessment in authentic field settings on student learning [12
]. When contemplating the use of assessment to influence learning in desirable ways, this lack is problematic for fields like the health sciences where performance assessment can comprise a substantial component of assessment.
A qualitative rather than a quantitative approach also seemed appropriate. There are too many variables and relationships to attempt validation using traditional quantitative means. Maxwell [15
] argues that qualitative research is well suited to taking account of the integral role of context and mental processes in causal processes and understanding. He further argues that prerequisites for the use of experiments in service of understanding causality include well-developed theory that allows interpretation of the results and a manipulable, fairly simple process free from "temporal and contextual variability" [16
]. Given the as yet tentative nature of the model, its complexity and the temporal and contextual variability of the relationships in the model, we opted for a qualitative approach at this stage in the development of the model.
Our research question was whether the model could be used to explain observations about the learning effects of assessment in a different context to that in which the model was derived. To do this, we undertook a qualitative study based on in-depth interviews with senior medical students about the impact of assessment on their learning in a clinical setting.