Chapter 1: Theory and Learning

Simon Knight, Simon Buckingham Shum

Connected Intelligence Centre, University of Technology Sydney, Australia DOI: 10.18608/hla17.001

ABSTRACT

7KHFKDOOHQJHRIXQGHUVWDQGLQJKRZWKHRU\DQGDQDO\WLFVUHODWHLVWRPRYHÜIURPFOLFNVWR FRQVWUXFWVÝLQDSULQFLSOHGZD\/HDUQLQJDQDO\WLFVDUHDVSHFLāFLQFDUQDWLRQRIWKHELJJHU shift to an algorithmically pervaded society, and their wider impact on needs careful consideration. In this chapter, we argue that by design — or else by accident — the use of a learning analytics tool is always aligned with assessment regimes, which are in turn grounded in epistemological assumptions and pedagogical practices. Fundamentally WKHQZHDUJXHWKDWGHSOR\LQJDJLYHQOHDUQLQJDQDO\WLFVWRROH[SUHVVHVDFRPPLWPHQWWR a particular educational worldview, designed to nurture particular kinds of learners. We outline some key provocations in the development of learning analytic techniques, key questions to draw out the purpose and assumptions built into learning analytics. We suggest WKDWXVLQJÜFODLPVDQDO\VLVÝØDQDO\VLVRIWKHLPSOLFLWRUH[SOLFLWVWDQFHVWDNHQLQWKHGHVLJQ and deploying of technologies — is a productive human-centred method to address these NH\TXHVWLRQVDQGZHRIIHUVRPHH[DPSOHVRIWKHPHWKRGDSSOLHGWRWKRVHSURYRFDWLRQV Keywords: Theory, assessment regime, claims analysis

In what has become a well-cited, popular article in ZLOODOZD\VEHVLJQLāFDQW">â@,QVXPZKHQ Wired PDJD]LQHLQWKHQHZHUDRISHWDE\WHVFDOH working with , theory is actually more data and analytics, Anderson (2008) envisaged the important, not less, in interpreting results and GHDWKRIWKHRU\PRGHOVDQGWKHVFLHQWLāFPHWKRG1R identifying meaningful, actionable results. longer do we need to create theories about how the )RUWKLVUHDVRQZHKDYHRIIHUHG'DWD*HRORJ\ world works, because the data will tell us directly as 6KDIIHU$UDVWRRSRXUHWDO DQG'DWD we discern, in almost real time, the impacts of probes Archeology (Wise, 2014) as more appropriate and changes we make. PHWDSKRUVWKDQ'DWD0LQLQJIRUWKLQNLQJ 7KLVKLJKSURāOHDUWLFOHDQGVRPHZKDWH[WUHPHFRQFOX - about how we sift through the new masses of VLRQDORQJZLWKRWKHUV VHHIRUH[DPSOH0D\HU6FK·Q - data while attending to underlying conceptual berger & Cukier, 2013), has, not surprisingly, attracted UHODWLRQVKLSVDQGWKHVLWXDWLRQDOFRQWH[W FULWLFLVP ER\G &UDZIRUG3LHWVFK  'DWDLQWHQVLYHPHWKRGVDUHKDYLQJDQGZLOOFRQWLQXH Educational researchers are one community interested WRKDYHDWUDQVIRUPDWLYHLPSDFWRQVFLHQWLāFLQTXLU\ LQWKHDSSOLFDWLRQRIÜELJGDWDÝDSSURDFKHVLQWKHIRUP +H\7DQVOH\ 7ROOH ZLWKIDPLOLDUÜELJVFLHQFHÝ RIOHDUQLQJDQDO\WLFV$FULWLFDOTXHVWLRQWXUQVRQH[ - H[DPSOHVLQFOXGLQJJHQHWLFVDVWURQRP\DQGKLJK BRCA2 gene Red Dwarf actly how theory could, or should shape research in energy physics. The , stars, Higgs bosun this new paradigm. Equally, a critical view is needed and the do not hold strong views on being on how the new tools of the trade enhance/constrain computationally modelled, or who does what with the people WKHRUL]LQJE\YLUWXHRIZKDWWKH\GUDZDWWHQWLRQWR results. However, when become aware that and what they ignore or downplay. Returning to our their behaviour is under surveillance, with potentially opening provocation from Anderson, the opposite important consequences, they may choose to adapt FRQFOXVLRQLVGUDZQE\:LVHDQG6KDIIHU S  RUGLVWRUWWKHLUEHKDYLRXUWRFDPRXĂDJHDFWLYLW\RU to game the system. Learning analytics researchers :KDWFRXQWVDVDPHDQLQJIXOāQGLQJZKHQWKH aiming to study learning using such tools must do so number of data points is so large that something aware that they have adopted a particular set of lenses

CHAPTER 1 THEORY AND LEARNING ANALYTICS PG 17 RQÜOHDUQLQJÝWKDWDPSOLI\DQGGLVWRUWLQSDUWLFXODU dence, at scale, the kinds of process-intensive learning ways, and that may unintentionally change the system that educators have long argued for, but have to date being tracked. Researchers should stay alert to the SURYHQLPSUDFWLFDO"([DFWO\ZKDWRQHVHHNVWRGRZLWK emerging critical discourse around big data in soci- analytics is at the heart of this chapter. ety, data-intensive science broadly, as well as within To design analytics-based lenses — with our eyes wide education where the debate is at a nascent stage. RSHQWRWKHULVNVRIGLVWRUWLQJRXUGHāQLWLRQRIÜOHDUQLQJÝ Let us turn now to educators and learners . The potential in our desire to track it computationally — we must RIOHDUQLQJDQDO\WLFVLVDUJXDEO\IDUPRUHVLJQLāFDQW XQSLFNZKDWLVDWVWDNHZKHQFODVVLāFDWLRQVFKHPHV than as an enabler of data-intensive educational , recommendation algorithms, and UHVHDUFKH[FLWLQJDVWKLVLV7KHQHZSRVVLELOLW\LV YLVXDOL]DWLRQVPHGLDWHWKHUHODWLRQVKLSVEHWZHHQ that educators and learners — the stakeholders who educators, learners, policymakers, and researchers. constitute the learning system studied for so long by The challenge of understanding how theory and an- UHVHDUFKHUVØDUHIRUWKHāUVWWLPHDEOHWRVHHWKHLU DO\WLFVUHODWHLVWRPRYHÜIURPFOLFNVWRFRQVWUXFWVÝ own processes and progress rendered in ways that in a principled way. until now were the preserve of researchers outside the /HDUQLQJDQDO\WLFVDUHDVSHFLāFLQFDUQDWLRQRIWKH V\VWHP'DWDJDWKHULQJDQDO\VLVLQWHUSUHWDWLRQDQG bigger shift to an algorithmically pervaded society. The even intervention (in the case of adaptive software) is frame we place around the relationship of theory to no longer the preserve of the researcher, but shifts to learning analytics must therefore be enlarged beyond embedded sociotechnical educational infrastructure. FRQVLGHUDWLRQVRIZKDWLVQRUPDOO\FRQVLGHUHGÜHGX - educators and learners So, for , the interest turns on FDWLRQDOWKHRU\ÝWRHQJDJHZLWKWKHFULWLFDOGLVFRXUVH the ability to gain insight in a timely manner that could around how sociotechnical infrastructures deliver improve outcomes. computational intelligence in society. Thus, with people in the analytic loop, the system The remainder of the chapter argues that by design EHFRPHVUHĂH[LYH SHRSOHFKDQJHLQUHVSRQVHWRWKH — or else by accident — the use of a learning analytics DFWRIREVHUYDWLRQDQGH[SOLFLWIHHGEDFNORRSV DQG tool is always aligned with assessment regimes , which ZHFRQIURQWQHZHWKLFDOGLOHPPDV 3DUGR 6LHPHQV are in turn grounded in epistemological assumptions 3ULQVORR 6ODGH 7KHGHVLJQFKDOOHQJH and pedagogical practices 0RUHRYHUDVZHVKDOOH[ - moves from that of modelling closed, deterministic plain, a long history of design thinking demonstrates V\VWHPVLQWRWKHVSDFHRIÜZLFNHGSUREOHPVÝ 5LWWHO that designed artifacts unavoidably embody implicit  DQGFRPSOH[DGDSWLYHV\VWHPV 'HDNLQ&ULFN values and claims. Fundamentally then, we argue that 0DFIDG\HQ'DZVRQ3DUGR *DÇHYLİ $V GHSOR\LQJDJLYHQOHDUQLQJDQDO\WLFVWRROH[SUHVVHVD we hope to clarify, for someone trying to get a robust commitment to a particular educational worldview, PHDVXUHRIÜOHDUQLQJÝIURPGDWDWUDFHVVXFKUHĂH[LYLW\ designed to nurture particular kinds of learners. will be either a curse or a blessing, depending on how important learner agency and creativity are deemed THEORY INTO PRACTICE WREHKRZā[HGWKHLQWHQGHGOHDUQLQJRXWFRPHVDUH whether analytical feedback loops are designed as interventions to shape learner cognition/interaction, In an earlier paper (Knight, Buckingham Shum, & Lit- and so forth. tleton, 2014) we put forward a triadic depiction of the relationship between elements of theory and practice Our view is that it is indeed likely that education, as in the development of learning analytic techniques, ERWKDUHVHDUFKāHOGDQGDVDSURIHVVLRQDOSUDFWLFH as depicted in Figure 1.1 (we refer the reader to this is on the threshold of a data-intensive revolution paper for further discussion of the depicted relation- DQDORJRXVWRWKDWH[SHULHQFHGE\RWKHUāHOGV$VWKH ships). Our intention was to illustrate the tensions and site of political and commercial interests, education inter-relations among the more or less theoretically LVGULYHQE\SROLF\LPSHUDWLYHVIRUÜLPSDFWHYLGHQFHÝ grounded stances we take through our pedagogic and and software products shipping with analytics dash- assessment practices and policies, and their underlying boards. While such drivers are typically viewed with epistemological implications and assumptions. suspicion by educational practitioners and researchers, the opportunity is to be welcomed if we can learn how The use of a triangle highlights these tensions: that to harness and drive the new horsepower offered by assessment can be the driving force in how we un- analytics engines, in order to accelerate innovation and GHUVWDQGZKDWÜNQRZOHGJHÝLVWKDWDVVXPSWLRQVDERXW improve evidence-based decision-making. Systemic SHGDJRJ\ IRUH[DPSOHDNLQGRIIRONSV\FKRORJ\2OVRQ educational shifts are of course tough to effect, but %UXQHU LQĂXHQFHZKRZHDVVHVVDQGKRZWKDW could it be that analytics tools offer new ways to evi- assessment and are sometimes in tension, where the desire for summative assessment overrides

PG 18 HANDBOOK OF LEARNING ANALYTICS pedagogically motivated formative feedback; and that centred on the triad of epistemology, pedagogy, and drawing alignment between one’s epistemological assessment. view (of the nature of knowledge) and assessment We use these provocations to illustrate how the im- or pedagogy practices is challenging — relationships SOLFLWÜFODLPVÝPDGHE\DOHDUQLQJDQDO\WLFVWRROFDQ between the three may be implied, but they are not EHGHFRQVWUXFWHG7KHDSSURDFKLQYLWHVUHĂHFWLRQ entailed  'DYLV :LOOLDPV 2IFRXUVHRWKHU on the affordances of the tool’s design at different YLVXDOL]DWLRQVPLJKWEHLPDJLQHGDQGWKHWKHRUHWLFDO OHYHOV LQFOXGLQJGDWDPRGHOOHDUQHUH[SHULHQFHDQG and practical purposes for which such heuristics are OHDUQLQJDQDO\WLFVYLVXDOL]DWLRQ  GHYLVHGLVLPSRUWDQWWRFRQVLGHU7RJLYHWZRH[DP - ples, we have considered versions of the depiction in Computer-supported learning — individual or collabo- which: 1) assessment and pedagogy are built on the UDWLYHØFRYHUVDKXJHDUUD\RIOHDUQLQJFRQWH[WV6XFK foundation of epistemology (in a hierarchical structure), tools support many forms of rich learner interaction DQG DUHEURXJKWLQWRDOLJQPHQWLQD9HQQGLDJUDP with peers and resources, which are implicit claims structure, with greater overlap implying a greater about learning. However, the emergence of computa- FRPSOHPHQWDULW\RIWKHWKHRUL]HGSRVLWLRQ WLRQDODQDO\WLFVHQDEOHVGHVLJQHUVØDQGE\H[WHQVLRQ the artifacts — to value certain behaviours above RWKHUVQDPHO\WKRVHORJJHGDQDO\]HGDQGUHQGHUHG visible to some stakeholder group. The implicit claim is that these are particularly important behaviours. We measure what we value. :HSURYLGHDVHWRIÜVL[:ÝTXHVWLRQVWREHFRQVLGHUHG in the development of learning analytics. Of course, across these questions, there is overlap, and any one TXHVWLRQPLJKWEHH[SUHVVHGLQPXOWLSOHZD\V7KH Figure 1.1. 7KH(SLVWHPRORJ\×$VVHVVPHQW×3HGDJRJ\ intention is neither to prescribe these as the only (3$ WULDG .QLJKWHWDOS  questions to be asked, nor that within each element of the triad only particular questions should apply. Learning analytics, as a new form of assessment As the descriptions of the provocations make clear, instrument, have potential to support current edu- within each facet of the triad, multiple theoretical cational practices, or to challenge them and reshape questions can and should be asked. Rather, we hope education; considering their theoretic-positioning to provide heuristic guidance to readers in developing is important in understanding the kind of education their own analytics. V\VWHPVZHDLPIRU)RUH[DPSOHOHDUQLQJDQDO\WLFV Epistemology — What Are We FRXOGKDYHWKHSRWHQWLDOWR PDUJLQDOL]HOHDUQHUV DQG Measuring? educators) through the transformation of education 7KHāUVWSURYRFDWLRQLQYLWHVWKHDQDO\WLFGHVLJQHU into a technocratic system; 2) limit what we talk about WRFRQVLGHUZKDWÜNQRZOHGJHÝORRNVOLNHZLWKLQWKH DVÜOHDUQLQJÝWRZKDWZHFDQFUHDWHDQDO\WLFVIRUDQG analytic approach being developed, asking, What  H[FOXGHDOWHUQDWLYHZD\VRIHQJDJLQJLQDFWLYLWLHV are we trying to measure? We pose this question to (that may be hard to track computationally), to the prompt consideration of the connection between a detriment of learners. Algorithms may both ignore, conceptual account of the object of measurement (the and mask some key elements of the learning process. knowledge being assessed), and a practical account of 7KHH[WHQWWRZKLFKDQDO\WLFVFDQXVHIXOO\VXSSRUW the methods and measures used to quantify activity educators and learners is an important question. These and outputs within particular tasks. Asking What are pressing issues given the rise of learning analytics, are we trying to measure? encourages us to consider and increasing interest in mass online education at our learning design, the skills and facts we want our both the pre-university and university levels (e.g., the students to learn, and what it means for students to growing interest in MOOCs). ÜFRPHWRNQRZÝ7KLVLVDTXHVWLRQRIHSLVWHPRORJ\ it concerns the nature of the constructs, why they EPA PROVOCATIONS ÜFRXQWÝDVNQRZOHGJHWKHHYLGHQWLDU\VWDQGDUGDQG kind required for a claim of knowledge to be made. ([SDQGLQJRQWKLVSULRUZRUNWKHUHVWRIWKHFKDSWHU aims to illustrate the application of our approach, with This knowledge might be of a more broadly proposi- the aim of providing actionable guidance for those WLRQDONLQG VRPHWLPHVFKDUDFWHUL]HGDVÜNQRZOHGJH developing learning analytics approaches and tools. WKDWÝFKDUDFWHUL]HGDVUHFDOORIIDFWV DPRUHEURDGVHW To do this, we have developed a set of provocations RIVNLOOVDQGFKDUDFWHULVWLFV VRPHWLPHVFKDUDFWHUL]HG DVÜNQRZOHGJHKRZÝIRUH[DPSOHWKHDELOLW\WRZULWH

CHAPTER 1 THEORY AND LEARNING ANALYTICS PG 19 an essay), or dispositions to act in particular ways (for to instrumental aims regarding the analytic’s con- H[DPSOHDVWKRVHGLVSRVLWLRQVUHFHQWO\GLVFXVVHGDV tribution to particular skills (perhaps employability epistemic virtues in epistemology). Evidentiary stan- VNLOOV RUXQLYHUVLW\FRPSOLDQFH IRUH[DPSOHUHSRUWLQJ dards and types concern the warrants indicative of requirements). It might also relate to pedagogic aims NQRZOHGJHIRUH[DPSOHZKHWKHUNQRZOHGJHFDQEH such as the support of particular groups of students, FRQFHSWXDOL]HGLQWHUPVRIXQLWDU\SURSRVLWLRQVWKDW and so on. may be recalled more or less appropriately within Pedagogy — Who is the Assessment/ SDUWLFXODUFRQWH[WVZKHWKHUNQRZOHGJHRIVRPHWKLQJ Analytic For? HQWDLOVWKHDELOLW\WRGHSOR\LWLQVRPHFRQWH[WWKHNLQGV ([WHQGLQJWKHFRQFHUQZLWKWKHQDWXUHRIWKHREMHFWRI RIMXVWLāFDWLRQDQGZDUUDQW DQGWKHVNLOOVXQGHUO\LQJ assessment above, is a further concern regarding the these) that cement some claim as knowledge, and so target of the analytic device, provoking the question, RQ7KHVHDUHØLPSOLFLWO\RUH[SOLFLWO\ØWKHWDUJHWV Who is the analytic for? In the development of analytic of our measurement. devices (and assessments more broadly), we should Epistemology — How Are We consider who the target of the device is, whether it Measuring? supports teachers, parents, students, or administrators Closely related to this conceptual question regarding in understanding some aspect of learning. Is the analytic the epistemological status of the object of analysis designed to provide insight at a macro (government, is a question regarding our access — as researchers institutional), meso (school, class), or micro (individual and educators — to that knowledge. This is a question student or activity) level (Buckingham Shum, 2012), regarding the epistemological underpinning of our and are there insights across these levels that can be research and assessment methods. There is a rich effectively made sense of by all stakeholders (Knight, literature on the various epistemological concerns %XFNLQJKDP6KXP /LWWOHWRQ " around quantitative and qualitative research methods, This question regards the desire for analytic insights ZLWKDJURZLQJVSHFLāFLQWHUHVWLQGLJLWDOUHVHDUFK at multiple levels of a system, and the ability of indi- methods. In addition, there is a focused literature in vidual analytic approaches (including their outputs WKHSKLORVRSK\RIDVVHVVPHQWH[SORULQJWKHHSLVWH - in various forms, such as dashboards) to support PRORJLFDOFRQFHUQVLQDVVHVVPHQWPHWKRGV 'DYLV the following: 1) individual students in developing 1999). Across this literature, issues concerning the their learning; 2) educators in developing their own subjectivity of approaches, and the ability of meth- practice and in targeting their support at individual odologies to give insights, are central. The question student needs; and 3) administrators in understanding invites considerations regarding the ways in which KRZFRKRUWVDUHGHYHORSLQJDQGWKHLURUJDQL]DWLRQDO analytic methods imply particular epistemologies. QHHGV$V&ULFNDUJXHVLQWKLVKDQGERRNDFRPSOH[ Note that this is not just a question of the reliability of systems conception of analytics for different levels in our assessment methods, but concerns the ability of the learning system, spanning from private personal DSSURDFKHVWRVSHDNWRDQH[WHUQDOO\NQRZDEOHZRUOG GDWDWKURXJKWRVKDUHGRUJDQL]DWLRQDOGDWDLPSOLHV (and the nature of that world). different rationalities and authorities to interpret at Pedagogy — Why is this Knowledge WKHGLIIHUHQWOHYHOV 'HDNLQ&ULFN  Important to Us? This question raises a parallel concern regarding The development of analytic approaches in learning the ethical implications in developing analytics that FRQWH[WVLQYROYHVPDNLQJGHFLVLRQVDERXWZKDWNQRZO- H[SOLFLWO\RULPSOLFLWO\ WDUJHWSDUWLFXODUJURXSV7KLV edge will, and will not, be focused on; to measure what concern is at two levels. First, analytics that require we value rather than value merely that which is easily particular forms of technology or participation may PHDVXUHG :HOOV &OD[WRQ 7KLVLVRIFRXUVHLQ create new divides between student cohorts, or addition to a conceptual account of the nature of that HQWUHQFKH[LVWLQJGLYLGHV6HFRQGWKHUHLVDQHWK - knowledge. These decisions in part relate to debates ical concern regarding the use of student data by around the kinds of important (or powerful) knowledge LQVWLWXWLRQVSDUWLFXODUO\ZKHUHVSHFLāFFRQVHQWLV LQVRFLHW\ VHHIRUH[DPSOH

PG 20 HANDBOOK OF LEARNING ANALYTICS ways takes place in a physical location, in response to or not a particular technology provides after-the-fact SDUWLFXODUWDVNGHPDQGVLQDVRFLRFXOWXUDOFRQWH[W or real-time feedback, and whether this feedback is with a particular set of tools. Contrast an individual intended to provide a scaffold or model for current SHQDQGSDSHUH[DPLQDVLOHQWKDOOZLWKSHHUV behaviour, is targeted at future behaviour and learning, with an emergency response simulation on the ward, or is just intended as a feedback mechanism on prior with tackling a problem in a MOOC, with work (which may not be covered again). FRQĂLFWUHVROXWLRQLQUHODWLRQVKLSFRXQVHOOLQJ)RU In our earlier paper, we made a distinction between HDFKFRQWH[WZHPXVWDVNQRWRQO\LIWKHDVVHVVPHQW the metaphors of biofeedback and diagnostic learning LVPHDQLQJIXOEXWDOVRWRZKDWH[WHQWPHDQLQJIXO analytics. The intention here was to draw a distinction computational analytics can be designed to add value. between formative and summative assessments (re- Moreover, we should also consider the ways in which spectively). However, while the analogy can be drawn, the assessment biases particular kinds of response of course systems that provide real-time feedback can ØLQVRPHWLPHVXQLQWHQGHGZD\V)RUH[DPSOHSDU - EHVXPPDWLYHLQQDWXUHDQGFDQWDNHRQDÜPRQLWRULQJÝ ticular groups of students, or kinds of knowledge, role towards some end-point. In addition, diagnosis does might be privileged over others through the design QRWKDYHWKHāQDOLW\SHUKDSVLPSOLHGE\WKHDQDORJ\Ø RIDVVHVVPHQWFRQWH[WVZLWKDYHU\QDUURZGHāQLWLRQ diagnosis provides insight into what is going wrong, of achievement; through requirements for behaviours ZKLFKPD\EHDFWLRQDEOHE\WKHÜSDWLHQWÝ VWXGHQW RU that not all students might engage in; through the use ÜGRFWRUÝ HGXFDWRU ,QVWHDGWKHQWKHIRFXVVKRXOG of technologies that unfairly assume socioeconomic be on whether the analytic device is targeted at a means; or through separating assessment from the summative snapshot perspective on student learning DSSOLHGFRQWH[WLQZKLFKDQH[SHUWLVHFDQEHGLVSOD\HG and monitoring towards that end, or instead, targeted authentically. at development and improvement over time. Across assessments, we should also consider the ways in which the particular systems shape the data CONCLUSION obtained — note this is a practical concern regarding Through the provocations, we have drawn attention the reliability and validity of methods, rather than the to the ways in which analytic approaches and artifacts UHODWHGHSLVWHPRORJLFDOFRQFHUQUDLVHGDERYH)RUH[ - subscribe to particular perspectives on learning: they ample, technologies are mediational tools, which shape implicitly make claims DFURVVWKH(3$WULDGDQGWKH the ways in which people interact with each other and provocations drawn from them. the world around them, and hence, the activity they PHDVXUH7KLVLVWUXHERWKRIWKHVSHFLāFWHFKQRORJLHV Tools can be used in many ways, and should not DQGWKHWDVNGHVLJQXVHGLQDVVHVVPHQWVIRUH[DPSOH EHLVRODWHGIURPWKHLUFRQWH[WRIXVH,QWHUDFWLRQDO WKHXVHRIÜDXWKHQWLFÝDVVHVVPHQWVSURYLGHVDGLIIHUHQW DIIRUGDQFHVOLNHEHDXW\DUHWRVRPHGHJUHHÜLQWKH range of possible responses than more traditional H\HRIWKHEHKROGHUÝ:HRIIHUWKHVHSURYRFDWLRQVDVD pen-and-paper assessments of various kinds. pragmatic tool for thinking, for designers, educators, researchers, and students — whether considering how Assessment — When Does the one currently makes use of analytical tools, how one Assessment, and Feedback, Occur? might do in the future, or indeed when designing new $āQDOFRQVLGHUDWLRQUHODWHVWRWKHWHPSRUDOFRQWH[W WRROVIRUQHZFRQWH[WV:HSURSRVHWKDWLWLVSURGXFWLYH of learning analytics, asking when the assessment and WRFRQVLGHUWKHVHSURYRFDWLRQVLQRUGHUWRUHĂHFWRQ feedback cycle occurs. This provocation is intended WKH(3$FODLPVEHLQJPDGHWKURXJKWKHGHSOR\PHQW to prompt consideration of the formative or/and RIDOHDUQLQJDQDO\WLFWRROZLWKLQDJLYHQFRQWH[W summative nature of the learning analytic; whether

REFERENCES

$QGHUVRQ& -XQH 7KHHQGRIWKHRU\7KHGDWDGHOXJHPDNHVWKHVFLHQWLāFPHWKRGREVROHWH Wired Magazine KWWSDUFKLYHZLUHGFRPVFLHQFHGLVFRYHULHVPDJD]LQHSEBWKHRU\

$UDVWRRSRXU*%XFNLQJKDP6KXP6&ROOLHU:.LUVFKQHU3.QLJKW6.6KDIIHU': :LVH$)  $QDO\WLFVIRUOHDUQLQJDQGEHFRPLQJLQSUDFWLFH,Q-/3ROPDQHWDO (GV  Learning and Becoming in Practice: Proceedings of the International Conference of the Learning Sciences  ,&/6Ú ×-XQH %RXOGHU&286$ 9ROSS× %RXOGHU&2,QWHUQDWLRQDO6RFLHW\RIWKH/HDUQLQJ6FLHQFHV boyd, danah, & Crawford, K. (2011, September). Six provocations for big data 3UHVHQWHGDW A Decade in Internet Time: Symposium on the Dynamics of the Internet and Society KWWSVVUQFRPDEVWUDFW 

CHAPTER 1 THEORY AND LEARNING ANALYTICS PG 21 Buckingham Shum, S. (2012). Learning analytics policy brief . UNESCO. http://iite.unesco.org/pics/publica- WLRQVHQāOHVSGI

'DYLV$   The limits of educational assessment . Wiley.

'DYLV$ :LOOLDPV.  (SLVWHPRORJ\DQGFXUULFXOXP,Q1%ODNH36PH\HUV 56PLWK (GV  Black- well guide to the philosophy of education . Blackwell Reference Online.

'HDNLQ&ULFN5  /HDUQLQJDQDO\WLFV/D\HUVORRSVDQGSURFHVVHVLQDYLUWXDOOHDUQLQJLQIUDVWUXFWXUH,Q &/DQJ *6LHPHQV (GV  Handbook on Learning Analytics (this volume).

Hey, T., Tansley, S., & Tolle, K. (Eds.). (2009). 7KHIRXUWKSDUDGLJP'DWDLQWHQVLYHVFLHQWLāFGLVFRYHU\ . Red- mond, WA: Microsoft Research. http://research.microsoft.com/en-us/collaboration/fourthparadigm/

Knight, S., Buckingham Shum, S., & Littleton, K. (2013). Collaborative sensemaking in learning analytics. CSCW and Education Workshop (2013): Viewing education as a site of work practice, co-located with the 16th ACM Conference on Computer Support Cooperative Work and Social Computing (CSCW 2013), 23 February 2013, 6DQ$QWRQLR7H[DV86$KWWSRURRSHQDFXN

Knight, S., Buckingham Shum, S., & Littleton, K. (2014). Epistemology, assessment, pedagogy: Where learning meets analytics in the middle space. Journal of Learning Analytics, 1 (2). http://epress.lib.uts.edu.au/jour- QDOVLQGH[SKS-/$DUWLFOHYLHZ

0DFIDG\HQ/3'DZVRQ63DUGR$ *DÇHYLİ'  (PEUDFLQJELJGDWDLQFRPSOH[HGXFDWLRQDO systems: The learning analytics imperative and the policy challenge. Research & Practice in Assessment, 9 , ×

0D\HU6FK·QEHUJHU9 &XNLHU.   Big data: A revolution that will transform how we live work and think /RQGRQ-RKQ0XUUD\3XEOLVKHU

2OVRQ'5 %UXQHU-6  )RONSV\FKRORJ\DQGIRONSHGDJRJ\,Q'52OVRQ 17RUUDQFH (GV  The Handbook of Education and Human Development  SS× +RERNHQ1-:LOH\%ODFNZHOO

3DUGR$ 6LHPHQV*  (WKLFDODQGSULYDF\SULQFLSOHVIRUOHDUQLQJDQDO\WLFV British Journal of Educa- tional Technology, 45  ×KWWSGRLRUJEMHW

3LHWVFK:  $VSHFWVRIWKHRU\ODGHQQHVVLQGDWDLQWHQVLYHVFLHQFH3DSHUSUHVHQWHGDWWKHWK%LHQQLDO 0HHWLQJRIWKH3KLORVRSK\RI6FLHQFH$VVRFLDWLRQ 36$ ×1RYHPEHU&KLFDJR,/KWWSSKLOV - ci-archive.pitt.edu/10777/

3ULQVORR3 6ODGH6  6WXGHQWSULYDF\VHOIPDQDJHPHQW,PSOLFDWLRQVIRUOHDUQLQJDQDO\WLFV Proceed- ings of the 5 th International Conference on Learning Analytics and Knowledge  /$.Ú ×0DUFK3RXJK - NHHSVLH1<86$ SS× 1HZ

Rittel, H. (1984). Second generation design methods. In N. Cross (Ed.), Developments in design methodology (pp. × &KLFKHVWHU8.-RKQ:LOH\ 6RQV

6KDIIHU': 'HFHPEHU (SLVWHPLFQHWZRUNDVVHVVPHQW3UHVHQWDWLRQDWWKH1DWLRQDO$FDGHP\RI(GX - FDWLRQ6XPPLW:DVKLQJWRQ'&

:HOOV* &OD[WRQ*   Learning for life in the 21st century: Sociocultural perspectives on the future of education +RERNHQ1-:LOH\%ODFNZHOO

:LVH$)  'DWDDUFKHRORJ\$WKHRU\LQIRUPHGDSSURDFKWRDQDO\]LQJGDWDWUDFHVRIVRFLDOLQWHUDFWLRQ in large scale learning environments. Proceedings of the Workshop on Modeling Large Scale Social Interac- tion in Massively Open Online Courses at the 2014 Conference on Empirical Methods in Natural Language Processing  (01/3 2FWREHU'RKD4DWDU SS× $VVRFLDWLRQIRU&RPSXWDWLRQDO/LQJXLV - tics. https://www.aclweb.org/anthology/W/W14/W14-4100.pdf

:LVH$) 6KDIIHU':  :K\WKHRU\PDWWHUVPRUHWKDQHYHULQWKHDJHRIELJGDWD Journal of Learn- ing Analytics, 2  ×

PG 22 HANDBOOK OF LEARNING ANALYTICS