Computer adaptive testing,big data and algorithmic approaches to education

This article critically considers the promise of computer adaptive testing (CAT) and digital data to provide better and quicker data that will improve the quality, efficiency and effectiveness of schooling. In particular, it uses the case of the Australian NAPLAN test that will become an online, adaptive test from 2016. The article argues that CATs are specific examples of technological ensembles which are producing, and working through, new subjectivities. In particular, CATs leverage opportunities for big data and algorithmic approaches to education that are symptomatic of what Deleuze saw as the shift from disciplinary to control institutions and societies.     

Datafication,testing events and the outside of thought

Education is undergoing various transformations due to new data-driven educational technologies and the management of educational data through data infrastructures. These technologies are frequently promoted to parents and the profession as being ‘revolutionary’ because they represent a new generation of learning. While computer adaptive tests may arguably improve various efficiencies, the argument that they will revolutionise education requires evaluation. In this paper we draw on the philosophy of Gilles Deleuze and Felix Guattari to theorise (a) the desire for data amongst policymakers and (b) the effects of data infrastructures as systems that coordinate educational thought. We argue that, rather than revolutionising learning as promised, datafication in computer-based modes merely offers more intense expressions of longstanding possibilities for learning. We describe three types of events—breaks, cracks and ruptures—and argue that data-events translate cracks (imperceptible changes that constitute learning) into breaks (information), but either cannot generate rupture (difference) or represent rupture as error. However, the intensification of learning through datafication may, we suggest, rupture educational thought more broadly.     

Introduction

This paper conceptualises think tanks and edu-businesses in relation to education policy work in the Australian polity. It situates the enhanced influence of both in relation to the restructured state, which has lost some key capacities in relation to the generation of research and ideas for policy. This restructuring has been strongly influenced by the techniques of new public management, the auditing of education through national and international testing and new forms of network governance, which have opened up spaces for the increased influence of think tanks and edu-businesses across the policy cycle in education. We see here the workings of a ‘polycentric state’. The paper also considers changing concepts of ‘evidence’, ‘expertise’ and ‘influence’ in respect of the involvement of think tanks and edu-businesses in circulating policy ideas and affecting policy development in Australian education. This introduction to this special issue of The Australian Educational Researcher serves as a provocation to further research on this new policy scenario.     

Auditory delayed discriminations by the dolphin: Nonequivalence with delayed-matching performance

Working memory in a bottlenosed dolphin was tested in both indirect and direct auditory delayed-discrimination tasks in which a correct spatial response was conditional upon the nature of a preceding sound. In the indirect task, either one of two possible sounds was briefly presented. After a prescribed delay, the dolphin was cued to go either to a left-hand or right-hand paddle pair. Responses to the outer paddle of a pair were rewarded following sound A, and responses to the inner paddle of a pair were rewarded after sound B. In the direct delayed-discrimination task, only one paddle pair was used in each session. In both tasks, the delay interval between the discriminative sound stimulus and the opportunity for a spatial response was progressively increased over sessions until the animal failed to meet a specified performance criterion or self-terminated a session. Delay limits of about 30 and 60 sec were obtained in the indirect and direct tasks, respectively. The increase in delay limit in the latter task was attributable to the use of an overt mediational response during the longer delays. In both cases, however, the obtained delay limits fell considerably short of the 2- to 3-min limits obtained in auditory delayed-matching studies using the same test sounds and the same subject. The task differences indicate that working memory functions cannot depend upon memory of the predelay stimulus alone, but must be determined in part by additional processes.     

Models of Decisionmaking Processes for Multiple-Choice Test Items: An Analysis of Spatial Ability

Latent class models of decisionmaking processes related to multiple-choice test items are extremely important and useful in mental test theory. However, building realistic models or studying the robustness of existing models is very difficult. One problem is that there are a limited number of empirical studies that address this issue. The purpose of this paper is to describe and illustrate how latent class models, in conjunction with the answer-until-correct format, can be used to examine the strategies used by examinees for a specific type of task. In particular, suppose an examinee responds to a multiple-choice test item designed to measure spatial ability, and the examinee gets the item wrong. This paper empirically investigates various latent class models of the strategies that might be used to arrive at an incorrect response. The simplest model is a random guessing model, but the results reported here strongly suggest that this model is unsatisfactory. Models for the second attempt of an item, under an answer-until-correct scoring procedure, are proposed and found to give a good fit to data in most situations. Some results on strategies used to arrive at the first choice are also discussed     

Vulnerability in Research: A Developmental Perspective on Research Risk

Assessing potential risks to children who participate in developmental research is a challenging task because children are a heterogeneous population, varying in developmental competencies and in background characteristics. This essay offers a developmental perspective on research risk, emphasizing that children's vulnerability to research risk changes in complex ways: some risks decrease with increasing age, some increase as the child matures, others change in a curvilinear fashion, while some remain essentially stable with development. Because vulnerability in research does not simply decline linearly with age, assessments of research risk must entail multidimensional considerations that vary over developmental time. In a similar manner, individual characteristics of children at any age (e.g., maltreatment, at-risk status, etc.) may also heighten their vulnerability to certain risks which require special consideration by researchers. Finally, this discussion of developmental vulnerability and the principles underlying research ethics suggests that in addition to the conventional risk/benefit analysis, researchers are in an optimal position to establish and maintain standards of decent treatment of children in research that safeguard their rights as research participants. Suggestions for fostering this process in the research community are outlined.     

Diagnosing the Poor Performance of Self‐worth Protective Students: A product of future outcome uncertainty,evaluative threat,or both?

Self‐worth protective students characteristically perform well on some occasions yet on other occasions they perform poorly. In this study, two accounts of the poor performance of self‐worth protective students are assessed. The first is that their poor performance is an outcome of evaluative threat. The second is that their poor performance is an outcome of future outcome uncertainty: a product of their uncertain global self‐esteem and uncertainty about the causes of achievement outcomes. Students high or low in self‐worth protection were exposed to either noncontingent success (creating future outcome uncertainty), noncontingent failure (evoking evaluative threat), or contingent success. Their ability to solve two tasks that involved a high degree of uncertainty was then assessed. Students high in self‐worth protection performed poorly following both noncontingent failure and noncontingent success, supporting the roles of both evaluative threat and future outcome uncertainty. The implications in terms of enhancing the achievement of students high in self‐worth protection are discussed.     

Sexual victimization and adolescent weight regulation practices: a test across three community based samples

OBJECTIVE: This study had four objectives: 1) to examine the association between sexual victimization and weight regulation across three samples of adolescent girls, 2) to assess whether sexual victimization is associated with more extreme forms of weight regulation in girls, 3) to discern whether sexual victimization has implications for the use of multiple forms of weight regulation, and 4) to address the role of physical victimization in explaining these relationships. METHOD: Data from three separate samples of 9th-12th graders were used to test these hypotheses. Girls from the upper Midwest responded to survey questions related to victimization and weight regulation in a largely urban sample (N = 2,086), a rural sample (N = 2,629), and a statewide sample (N = 966). RESULTS: Logistic regression revealed that sexual victimization was consistently associated with weight regulation in adolescent girls, independent of the effects of physical victimization. In the urban sample, being sexually victimized was associated with an increase in the probability of purging by 18% relative to not being sexually victimized. Sexual victimization was associated more strongly with extreme forms of weight regulation and significantly discriminated whether girls would choose multiple weight regulation forms. CONCLUSIONS: Sexual victimization contributes unique variance to the probability that girls will practice weight regulation techniques. Data from three independent samples confirms that being violated sexually places girls at risk for various health compromising eating behaviors.     

Infants' affective responses in the strange situation: effects of prematurity and of quality of attachment

20 full-term and 20 preterm infants and their mothers were videotaped in the Strange Situation, and the security of their attachment relationships was later determined. Each episode was subsequently divided into consecutive 15-sec intervals, during each of which ratings of facial expressions were performed. From these ratings several summary dimensions of affect were derived (e.g., affective peak and range during all episodes, latency and rise time for onset of distress during separation episodes, and recovery time during reunions). Term and preterm infants did not differ from one another in either the security of attachment or their affective expression and regulation. When groups were combined, patterns of affective expression were significantly different for infants classified as insecure-avoidant, insecure-ambivalent, and securely attached, as well as for group B1 + B2 infants compared to group B3 + B4 babies. The findings indicated that attachment-related affect may reflect an affect continuum that underlies certain mother- and stranger-directed behaviors in the Strange Situation, but that not all aspects of reunion behavior can be predicted by prior separation reactions.     

A faculty team works to create content linkages among various courses to increase meaningful learning of targeted concepts of microbiology

As research faculty with expertise in the area of host–pathogen interactions (HPI), we used a research group model to effect our professional development as scientific educators. We have established a working hypothesis: The implementation of a curriculum that forms bridges between our seven HPI courses allows our students to achieve deep and meaningful learning of HPI concepts. Working collaboratively, we identified common learning goals, and we chose two microorganisms to serve as anchors for student learning. We instituted variations of published active-learning methods to engage students in research-oriented learning. In parallel, we are developing an assessment tool. The value of this work is in the development of a teaching model that successfully allowed faculty who already work collaboratively in the research area of HPI to apply a “research group approach” to further scientific teaching initiatives at a research university. We achieved results that could not be accomplished by even the most dedicated instructor working in isolation.