In-game tutorials explicitly scaffold the designed pathway for players inLakeland, nudging playerstowards self-sustaining strategies. However, the game grants achievements to players in fourcategories, which may engage different types of players. For example, some players may focus ongaining achievements for a large population, many farms, or money earned. On the other hand,some players may follow a path of destruction, progressing primarily along the algae bloom track.This tension between the designed (and instructed) pathway and the varied achievements leads toopportunities for divergent play. In this case study, the objective is to use our method to evaluate thegame’s existing design, by examining the players’ decisions and progress, and how they interact withgame feedback.

A Method for Clustering Styles of Gamepl

through qualitative approaches such as direct observation,think-alouds and structured interviews.

The ability for an educational game designerto understand their audience’s play styles andresulting experience is an essential tool for improving their game’s design. As a game is subjected tolarge-scale player testing, the designers require inexpensive, automated methods for categorizingpatterns of player-game interactions.

There was a negative association between age and academic attainment. Younger studentswere more likely to complete a course. Students with higher prior educational qualifications hada higher chance of completing, passing and achieving excellent grades.

ncon-scious bias in the nature of feedback that 470 BME students and 470 matched White studentsreceived on their assignment from their tutors.

That means the attainment gap can be explained by differences in studentprior qualifications, with more BME students having a lower prior qualification than White stu-dents when they entered universities

white categorisation should be critically debated, as the term ethnicity is socially constructed,and people may have multiple identities.

The latest review by Richardson(2018) which synthesised data from multiple sources over the last 20years showed that the odds ofobtaining a good degree (i.e. first-class or upper second-class) in BME students are about half thosein White students. This under-attainment effect was stronger for Black students compared to Asianstudents.

nterdisciplinary field of learning analytics hasdemonstrated its potential to identify students who may need additional support from an earlystage and provide real-time interventions (

discrimination-free based simply on the fact that the algorithms do not use discriminationcategories.

This means that even if training is done with a quantifiable goal criterion and no protected category variables, algorithms canstillbe quite discriminatory if a “proxy”variable is included that correlates with both the quantifiable goal criterion and one or more protected category variables.

This direct involvement of student voices inshaping a policy dealing with the ethics of learninganalytics offered unique insight into the ways in whichstudents regard their data — as a valuable entity to becarefully protected and even more carefully applied

An increasing awareness of learning analytics as ameans of doing something to the student withoutthat student necessarily knowing triggered furtherH[SORUDWLRQRILVVXHVDURXQGVXUYHLOODQFHVWXGH

HEIs (higher education institutions) to proactivelyidentify and support students at risk of failing or drop-SLQJRXWWKH\GRVRLQDFRQWH[WZKHUHE\UHVRXUFHVare (increasingly) limited.

PLQFUHDVHGGDWDharvesting, we should not ignore the possibilities ofÜGDWDSUR[\LQGXFHGKDUGVKLSâZKHQWKHGHWDLORE-WDLQHGIURPWKHGDWDSUR[\FRPHVWRGLVDGYDQWDJHLWVHPERGLHGUHIHUHQWLQVRPHZD\Ý 6PLWK

Learning analytics as moral practice — focusingnot only on what is effective, but on what is ap-propriate and morally necessary2. Students as agents — to be engaged as collabora-tors and not as mere recipients of interventionsand services3. Student identity and performance as temporalG\QDPLFFRQVWUXFWVØUHFRJQL]LQJWKDWDQDO\WLFVprovides a snapshot view of a learner at a particularWLPHDQGFRQWH[W4. 6WXGHQWVXFFHVVDVDFRPSOH[PXOWLGLPHQVLRQDOphenomenon5. Transparency as important — regarding the pur-poses for which data will be used, under whatconditions, access to data, and the protection ofan individual’s identity That higher education cannot afford not to use dat

increasing surveillance and the (un)warranted col-OHFWLRQDQDO\VLVDQGXVHRISHUVRQDOGDWDÜIHDUVDQ