1_SRI

download 1_SRI

of 25

Transcript of 1_SRI

  • 7/26/2019 1_SRI

    1/25

    A Uniform Measure for Quantifying Student Performance at UNSW

    Geof Whale

    School o Computer Science & Engineering/New South Solutions

    [email protected]

    1 !arch 1"""# re$ised 1% ugust 1"""

    Executive Summary

    Student performance measures

    'he introduction o new administrati$e (' s)stems at *NSW o$erthe ne+t )ear pro$ides a uni,ue opportunit) or the *ni$ersit) to

    ta-e ad$antage o eatures currentl) una$ailale in the Student(normation S)stem S(S0. ne such acilit) is the calculation o auniorm measure or ,uanti)ing and ran-ing student perormancee)ond the mar- awarded or a single course su2ect0.

    NewSouth Student associates with each student record a gradepoint average# or G3. With relati$el) minor modi4cation to thesotware the G3 calculation can ollow a 1556point scale thatrepresents a weighted average markW!0# a measure amiliar inthe *NSW conte+t and much more appropriate or a gradings)stem that has onl) our passing grades. 'he onl) weights applied

    are units o credit.

    Recommendations

    'his paper supports the ollowing proposals7

    that *NSW adopts uniorm# cumulati$e W! as the preerredmeasure o student perormance8

    that 9aculties e strongl) encouraged to e+press their eligiilit)or ran-ing re,uirements or merit6ased awards such as#interalia# :onours in integrated programs# entr) to the :onours

    )ear where separate# ;egree with !erit# and pri

  • 7/26/2019 1_SRI

    2/25

    01_SRIQuantifying Student Performance

    While central calculation is a eature o NewSouth Student#Faculty-specifc ormulas or variations cannot be accommodated.'he sotware can pro$ide onl) one -ind o a$erage.

    dopting a common s)stem would not onl) simpli) the mechanical

    process# it also assists in ensuring e,uit) e$er) student=s W! iscalculated the same wa)0 and transparenc) the W! appears onacademic statements and can e easil) $eri4ed ) the student0.3resent practice in inorming students o their o$erall progress ispatch) at est.

    Would not the adoption o a uniorm measure result in a loss oautonom) ) 9aculties> 'here are two aspects to this concern7

    1. Each academic unit iases its ormula in wa)s it considerssigni4cant8 so uniormit) implies remo$ing this ?e+iilit).Section .% o the paper includes an assessment o the impact

    o one o the a$oured -inds o aritrar) actors# )earweightings. 'he $ast ma2orit) o :onours grade decisions areunafected ) simpli)ing the measure ) counting all )earse,uall). E$en diferential weightings and the practice odiscarding results rom allegedl) irrele$ant courses can eaandoned without loss o integrit). Section A addresses thiscrucial point in detail.

    2. 9aculties ma) percei$e the loss o their ailit) to determinewhere :onours grade thresholds occur. 'he proposal does notre,uire that uniorm thresholds e applied throughout the

    *ni$ersit). Setting and pulicising0 these cutofs remains a9acult) responsiilit).

    3. (n addition# 9aculties must retain the right to accept a studentor an award who does not ,uali) on the asis o W! i thereare special circumstances.

    common measure# centrall) calculated and included on reports#can ree staf or the much more important tas-s o monitoringstandards and dealing with special cases that o$erride theapplication o the ormula.

    Potential usesBesides its application to :onours and !erit grades# the W! isroust enough to use or se$eral purposes7 entr) to an e+tended:onours )ear# pri

  • 7/26/2019 1_SRI

    3/25

    01_SRIQuantifying Student Performance

    Costs

    ;irect costs in sotware modi4cations# orne ) NewSouthSolutions ut actored into a limited udget or customisation.

    3ulicising the measure to staf and students incorporated into

    NewSouth Solutions change6management strategies# alread)re,uired or the new s)stems0

    (mplementing transition arrangements so that students are notdisad$antaged possil) using initiall) lower thresholds or a)ear or two0.

    Benefits

    (ncreased student satisaction with the transparenc) o theprocess.

    ess staf time spent on mundane tas-s.

    (mpro$ed eficiencies in committees dealing with studentran-ing# such as the *ni$ersit) !edal Committee and 3ran-ing committees.

    Con$ersion to e+ternall) reported measures such as *S G3s orthe European EC'S grading s)stem is relati$el) eas).

    GDW1% ugust 1"""

    1 Introduction

    !an) rewards are estowed on students at *NSW whoseacademic perormance is meritorious in some wa). !ost degreescan e awarded with :onours or with !erit# scholarships areofered to well ,uali4ed students# or pri

  • 7/26/2019 1_SRI

    4/25

    01_SRIQuantifying Student Performance

    de$ised their own measures# perormed their calculations usingpulished or re,uentl) unpulished ormulas# and reported theresults to the administration and the student $ia assessmentcommittees.

    conse,uence o this highl) de$ol$ed responsiilit) or themechanical aspects o determining awards as well as theiracademic ownership is that the methods emplo)ed in eachdiscipline ha$e di$erged o$er man) )ears. s discussed in Section# algorithms ha$e ecome more comple+# with diferentialweights and special rules ma-ing it almost impossile or an)oneapart rom the program authorit) to reproduce the result withaccurac). (n man) cases students recei$e little eedac- on theirprogress and are 2usti4al) anno)ed at the lac- o transparenc) inthe ran-ing process. 'he) should e+pect rather etter ser$ice.

    2 Purpose'he intention o this paper is to propose the adoption o a uniormmeasure or student perormance throughout *NSW. 'his reormstems rom two initiati$es7

    'he simpli4cation o academic program structures appro$ed )the cademic Board in -eeping with the report o the Wor-ing3art) on (ntegrating !easures.

    'he introduction o a new student administration s)stem or the%555 academic )ear. 3eopleSot Student dministration

    NewSouth Student0 is ale to calculate measures o studentprogress or the current enrolment period and cumulati$el)o$er the whole program.

    new measure would need to satis) se$eral re,uirements7

    (t must e simple# so that calculations can e easil) chec-edwithout resorting to complicated algorithms.

    (t must e transparent# so that e$er) student -nows how the)are going to e ran-ed against other students= perormances.

    (t must e equitable# with the same rules used to calculate a$alue or each student.

    (t must e robust enough to e used or most purposes oran-ing meritorious perormance and# potentiall)# orunsatisactor) progress0.

    (t should e reasonal) compatible with e+isting measuresthough there are a large numer o current s)stems0 to a$oidserious discontinuit) in the $alues reported to students undere+isting local schemes and a later uniorm *NSW measure.

    'he measure that is deli$ered with NewSouth Student is grade

    point averageG30# where a single numeric $alue associated witheach 4nalised grade is a$eraged# weighted ) units o credit.

    4

  • 7/26/2019 1_SRI

    5/25

    01_SRIQuantifying Student Performance

    modest modi4cation# such as a$eraging the mar-s rather than agrade point# is possile. :owe$er# $ariations to suit a speci4c9acult)=s presumed need cannote accommodated# oth on thegrounds o sotware modi4cation cost and ecause the measurewould no longer e uniorm and e,uitale.

    lthough a standard measure might e adopted# awards such as:onours Class 1 need notha$e the same ,uali)ing $alue acrossthe *ni$ersit). 9acult) education committees should retain theright to $ar) the cutofs within a small range gi$en -nowledge othe cohort# and to award :onours to indi$iduals on grounds otherthan the pulished ran-ing in special circumstances.

    Section descries the range o present practices in determining:onours and degree !erit awards and summarises the situation inother ma2or ustralian uni$ersities. Section identi4es thepreerred option or a *NSW measure# and Sections A and estalish how well the proposed measure mirrors the wa) :onourshas een awarded in the past ) appl)ing se$eral measures to thegraduating class o 1""H.

    Current !onours and "erit #radin# practices

    lthough ran-ings are re,uired or other purposes such asscholarships and pri

  • 7/26/2019 1_SRI

    6/25

    01_SRIQuantifying Student Performance

    lthough :onours is graded internall) or e+tended programs anadditional re,uirement or student ran-ing is selection or the:onours )ear itsel. Selection criteria or entr) to :onours is ) aormula whose purpose is similar to that used to grade integratedprograms.

    3.2 Handbooks

    9ew descriptions o the method used to award :onours appear in*NSW :andoo-s. Some e+tended programs or e+ample# thosein the College o 9ine rts0 include a section on :onourscontaining statements such as

    A prerequisite [to entry to onours in Fine Arts! is normally a"istinction #"$% average in the studio core sub&ects in years' and () 1""H C9 :andoo-# p%

    :andoo- descriptions o the integrated programs are oten,ualitati$e rather than ,uantitati$e. 'his is rom the 1""HEngineering :andoo- p %A07

    onours will be awarded or meritorious perormance overthe course* special attention is paid to a candidate+sperormance in the fnal year sub&ects and thesis pro&ect)

    Speci4c program descriptions ma) re4ne this8 or e+ample# Ci$ilEngineering speci4es oth )ear weights 1# %# and A0 and theweighted a$erage mar- range associated with each grade.

    good e+ample o a airl) complete speci4cation can e ound inthe !edicine :andoo- # which pulishes e$er) course weightour diferent $alues0 and )ear weight %# # # # # 0. lso,uoted is the percentage o students awarded each grade o:onours# and the corresponding ,uali)ing mar-# or a recentgraduating class.

    3.3 Formulas

    9rom limited handoo- descriptions# sumissions to the *ni$ersit)!edals Committee and personal -nowledge# it is clear that man):onours ormulas share general eatures )et each difers

    consideral) in detail. ten the diferences seem aritrar) oro$erl) in?uenced ) su2ecti$e $iews aout the rele$ance ocertain courses. ')pical eatures are as ollows personalcomments are in s,uare rac-etsJ K07

    Weighting courses initiall) ) credit points or contact hours or:ECS stud) load# t)picall) gi$ing three slightl) diferentrelati$e $alues. J9rom %555# units o credit will supersede allthree.K

    Weighting courses rom one or two areas the ma2or discipline0

    higher than other courses. ;iscounting some courses such as General Education entirel).

    6

  • 7/26/2019 1_SRI

    7/25

    01_SRIQuantifying Student Performance

    Weighting later6)ear courses ao$e earlier )ears# or discounting)ear 1 entirel).

    (gnoring repeat attempts at a course# or# alternati$el)# countingall attempts. J(n m) $iew# all attempts should e included7 a

    second result is unli-el) to ull) counteralance the negati$eefect o the initial ailure on the a$erage# and students shoulde encouraged to e+cel e$en in areas where the) ha$epre$iousl) lapsed.K

    Counting 9 and/or ;9 as

  • 7/26/2019 1_SRI

    8/25

    01_SRIQuantifying Student Performance

    aculties. :owe$er# it is e+pected that a candidate=s perormancethroughout their career e ta-en into account# not 2ust one or two)ears o stud).

    The University of Melbourne uses course grades ased on

    :onours categories with the highest grade awarded at H5 and apass on A50. No central measures o perormance are pulished toidenti) meritorious or unsatisactor) perormance.

    The University of Queensland has a M6point grading s)stem I3ass# I 3ass conceded# % or 1 I 9ail0 and e+press o$erallperormance as a grade point a$erage with ew special rules.

    $ Simplifyin# the measures

    Whether a common ran-ing indicator such as W! or G3 couldreplace locall)6applied ormulas depends on se$eral actors# most

    notal) how similar the group o students nominated or each:onours grade would e under current and simpli4ed ormulas.Section o this document descries the results o an anal)sis othis issue.

    s with reorm to academic structure# re$ision o :onoursormulas comes aout ) ,uestioning the need or comple+it).9eatures should not e retained purel) or historical reasons# utonl) when a strong case is made or their eing essential to theintegrit) o the measure.

    4.1 Minimalist model

    'he simplest $iale model is one that remo$es )ear and disciplineweightings# lea$ing onl) a wor-load weighting units associatedwith a course0. ll graded attempts at a course count. Whether 9and/or ;9 should e counted is a matter o uture polic)de$elopment see Section . or discussion0.

    4.2 mpact o! removing year "eightings and counting repeat attempts

    s a 4rst pass at ,uanti)ing the impact o simpli)ing ormulas theweighted a$erages or all students enrolled in A ,omputerngineering in session %# 1""H were anal)sed. 'his program

    normall) uses )ear weightings o 1# %# and or )ears 1 to respecti$el). nl) the 4rst attempt at a course is counted. Some$ariation o this scheme is applied to each program in the 9acult)o Engineering.

    9igure .1 tallies the diference etween each student=s W!calculated using a uniorm weighting mar- weighted ) creditpoints onl)# all non6mar-s ignored# all graded attempts at a coursecount0 and the con$entional W! )ears weighted as descriedao$e0. 'he diference was rounded to the nearest multiple o 5.mar-s or displa) purposes. 'he mean o the "% scores is 5.A#

    the standard de$iation is 1.A.

    8

  • 7/26/2019 1_SRI

    9/25

    01_SRIQuantifying Student Performance

    E+treme cases were as ollows7

    ne stage student=s W! dropped ) .H mar-s. 'hestudent=s results impro$ed rom a pass a$erage in )ear 1 tocredits in )ear % and distinctions in )ear . 'his monotonicpro4le e+actl) matches the )ear weightings# and thus isdisad$antaged ) uniorm weightings.

    9our students gained more than si+ W! points through ha$ingrepeated results count. Each a$erage is elow A5# howe$er# sothe diference is thus o little importance.

    ne student=s stage6% a$erage dropped rom M% in )ear 1 to A1in )ear %. 'he uniorm W! is higher ) .% points.

    ;espite these e+tremes the data is well clustered around the

  • 7/26/2019 1_SRI

    10/25

    01_SRIQuantifying Student Performance

    % Possi&le Student Ran'in# Indicators

    (n order to place the 4nal recommendation into perspecti$e# arange o possile measures that could e considered or ran-ingstudents is now proposed. (n the remainder o this document the)are called .tudent /anking IndicatorsSD(s0.

    'en SD(s were de$ised or the anal)sis descried in the ne+tsection. 'he) are not intended to e simple alternati$es some arear more dificult to implement under a centralised scheme thanothers0# ut to allow the efect o counting some results anddiscounting others to e measured. Dules or deciding whichresults to discount are ased on oth discipline and le$el# eaturesthat appear in some orm in man) o the current ormulas. 3recisematching o acult) practices is not possile in all cases ecauseo incomplete speci4cation and historical changes in credit points0.Ne$ertheless diferent SD( models capture the essence o currentpolicies.

    'he SD(s orm two related groups o 4$e each. 'he 4rst group usesa weighted a$erage mar-# while the second group is ased on agrade point a$erage# using the *C M6point G3 s)stem :;IM#;NI# CDIA# 3SI# 3CI# 3'I# 9I50. 'he 4$e -inds o SD( ineach group are as ollows7

    $erall a$erage# all results count

    Pear 1 ignored. 'his could e accomplished in NewSouthStudent ) resetting counters.

    Pear 1 to onl)# 4nal )ear ignored this would gi$e the SD( or

    applicants to e+tended :onours programs# otherwise not $er)meaningul0

    10

    igure !"' WAM $ifferences (a)so&ute va&ue*

    0#5 to 1

    25$

    o%er 5

    2$2 to 5

    11$

    1 to 2

    27$

    0#2 to 0#5

    18$

    u& to 0#2 difference17$

  • 7/26/2019 1_SRI

    11/25

    01_SRIQuantifying Student Performance

    !a2or studies onl)# de4ned separatel) or all programs. 'his isunli-el) to e easile ecause o the comple+it) o de4ning andmaintaining the relationships centrall).

    !a2or studies in )ears 1 to onl). Similarl) used or comparisononl).

    'he onl) weighting is the numer o credit points associated with acourse or courses ta-en prior to 1""A and no longer ofered anominal C3 $alue is assigned0. Some multiple courses such asEEC"15/"11 ha$e the credit points or the 4rst parttranserred to the second. 'his will also e possile in NewSouthStudent.

    ( )n analysis of SRI relia&ility

    ( a common measure is adopted it is necessar) or :onours

    awards ater con$ersion to e comparale with i not identical to0those awarded under current schemes. ne measure o the ,ualit)or reliailit) o an) scheme is the proportion o :onours gradesthat are mismatched under the new scheme compared to the oldone. mismatchedresult is one that changes rom one grade toanother or rom 3ass to :onours or $ice $ersa. Whether the newscheme can or should exactly mirror the old is a moot point.:owe$er# 4nding a simpli4ed measure that appro+imatel) matchesthe current outcomes would appear to e an achie$ale goal.0

    #.1 Methodology

    'he S(SWe records o all students who graduated at the 9eruar)to !a) 1""H ceremonies were used as the data source to comparecurrent practice using the grade o :onours or !erit attached toeach student record0 against the ten SD(s descried in Section A.Each SD( was used purel) as a ran-ing measure8 the ,uali)ing$alues or each grade o :onours was allowed to $ar) in order tominimise the numer o mismatches on either side o thethreshold. 'his usuall) resulted in $er) similar numers o gradesmo$ing up as would e mo$ed down

    9or e+ample# the ollowing tale lists the :onours grades o all

    1""H graduates in course 5 whose o$erall uniorm W! liesetween M%.5 and H5.5. lso shown are the numer o ailures andnumer o e+ternal e+emptions granted such students ma) ha$eeen treated speciall) ecause their *NSW record contains ewerresults0.

    WAM +onours ai&ures ,em%tions

    79.0 H1

    79.0 H1 14

    78.6 H1 14

    77.9 H2/1

    76.9 H1

    76.6 H1 12

    11

  • 7/26/2019 1_SRI

    12/25

    01_SRIQuantifying Student Performance

    76.1 H1 13

    75.8 H2/1 2 13

    75.3 H1 14

    74.6 H2/1 13

    74.2 H1

    74.2 H1 1

    73.8 H2/1

    73.4 H1

    72.9 H2/1

    72.6 H1

    72.5 H2/1 14

    72.4 H2/1

    ( the :onours 1 ,uali)ing mar- were set to# sa) MA.5# then :1results would e displaced the) would ecome :%/10 and % :%/1

    results would e displaced upward. t M.%# % :1s and :%/1s aredisplaced. 'his is the preerred $alue arrowed0. t M%. no :1swould e displaced ut A :%/1 results would mo$e up to :1.

    ,omment0lternati$e strategies such as a$ouring upward mo$eso$er downward ones in order to disad$antage ewer students0results in slightl) higher total misplaced grades. 'he cutofs,uoted later in this section represent upper ounds# and could elowered slightl) or reasons o e,uit).

    Comined programs present a dificult). signi4cant numer ostudents graduate rom comined courses etween aw and rts

    or Commerce. 9or these programs all WS su2ect area resultswere ignored in calculating a$erages or the non6law degree.Numers o graduates in other comined courses is too low orthem to ha$e an impact on acult) 4gures# ut the issue o howa$erages or such students are to e calculated in uture is stillunresol$ed.

    !a2ors were de4ned according to the schools0 responsile or thema2orit) o courses in a program. Both su2ect areas were countedor doule ma2ors in Commerce and Economics# such asccounting and 9inance. nce again# slight de4ciencies in the

    model do not in$alidate it7 the point o the e+ercise is to seewhether more hea$il) weighting some areas reall) ma-es much oa diference in the end.

    'he charts that ollow record the mismatch rate or each SD( orthe programs in se$eral aculties. 'he '((trace treats all studentsthe same and uses a common cutof thus producing moremismatches due to program $ariations7 a common threshold or allprograms is not a realistic goal0. 'races or two or three programsthat e+hiit pro4les with some e+treme characteristics are alsoplotted..

    12

  • 7/26/2019 1_SRI

    13/25

    01_SRIQuantifying Student Performance

    #.2 ntegrated programs

    s the most li-el) use or a common measure is :onours gradingin integrated programs it is important that a reasonale

    correlation e oser$ed etween current :onours gradings andthose under a proposed common measure. Better than "5Omatches is achie$ale under good conditions8 less than H5O woulde cause or concern. 'he charts plot the in$erse o this7 "5Ocompliance I 15O mismatches. 'hus the etter results areindicated ) traces lower on the scale.

    13

  • 7/26/2019 1_SRI

    14/25

    01_SRIQuantifying Student Performance

    igure ."# acu&ty of the /ui&t ,nvironment (+onours*

    0

    5

    10

    15

    20

    25

    30

    35

    )'*

    o%era((

    )'*

    yr2+

    )'*

    yr1,3

    )'*

    ma-or

    )'*

    ma-or

    yr1,3

    .P'

    o%era((

    .P'

    yr2+

    .P'

    yr1,3

    .P'

    ma-or

    .P'

    ma-or

    yr1,3

    Student 0anking 1ndicator

    $/

    onour"mi"matc0ed

    '((

    3260

    3331

    3385

    Built Environment. Se$eral small programs# airl) well correlated#suggesting uniorm acult) polic). Shows strong dependence on4nal )ear worst results or Pear 1 to W! or G30. $erallW! or G3 is a good predictor or most programs# with etterthan "5O compliance. Ne$ertheless standards $ar) etweenprograms.

    igure ."' acu&ty of ,ngineering (+onours*

    0

    5

    10

    15

    20

    25

    30

    )'*

    o%era((

    )'*

    yr2+

    )'*

    yr1,3

    )'*

    ma-or

    )'*

    ma-or

    yr1,3

    .P'

    o%era((

    .P'

    yr2+

    .P'

    yr1,3

    .P'

    ma-or

    .P'

    ma-or

    yr1,3

    Student 0anking 1ndicator

    $/

    onour"mi"matc0ed

    '((

    3140

    3620

    3640

    Engineering. (gnoring )ear 1 results impro$es matching in all ut oneprogram# 151ining ngineering ormerl) under the 9acult) opplied Science0. Strong dependence on )ear . W! slightl)

    etter predictor than G3.

    14

  • 7/26/2019 1_SRI

    15/25

    01_SRIQuantifying Student Performance

    igure ."2 acu&ty of Medicine (+onours*

    0

    5

    10

    15

    20

    25

    30

    )'*

    o%era((

    )'*

    yr2+

    )'*

    yr1,3

    )'*

    ma-or

    )'*

    ma-or

    yr1,3

    .P'

    o%era((

    .P'

    yr2+

    .P'

    yr1,3

    .P'

    ma-or

    .P'

    ma-or

    yr1,3

    Student 0anking 1ndicator

    $/

    onour"mi"matc0ed

    '((

    3801

    3821

    Medicine. 'he smaller o the two programs in !edicine# H%1.cience21edicine# had a small graduating class o H. 'heir resultswere so well spread that every SD( was ale to reproduce theawarded grades. 'he larger group H51 1edicine0 shows amar-ed dependence on W! o$er G3.

    igure ."! Sciences (+onours in integrated %rograms*

    0

    5

    10

    15

    20

    25

    30

    35

    40

    )'*

    o%era((

    )'*

    yr2+

    )'*

    yr1,3

    )'*

    ma-or

    )'*

    ma-or

    yr1,3

    .P'

    o%era((

    .P'

    yr2+

    .P'

    yr1,3

    .P'

    ma-or

    .P'

    ma-or

    yr1,3

    Student 0anking 1ndicator

    $/

    onour"mi"matc0ed

    '((

    3431

    3950

    Sciences9acult) o ie Sciences8 9acult) o Science & 'echnolog)0. 'hetwo programs rom ie Sciences 1 3sychology and "A54ptometry0 produce ,uite diferent results. 1 is one o the ewprograms or which G3 indicators are apparentl) etter thanW! at predicting :onours grade the correlation is surprisingsince )ear is weighted M5O in determining :onours in thisprogram and the course 3SPC55 carries the o$erall a$erage0.9or "A5 the opposite is true# and o$erall W! is in act the estSD( o all.

    15

  • 7/26/2019 1_SRI

    16/25

    01_SRIQuantifying Student Performance

    (ntegrated programs in Science & 'echnolog) are mainl) rom theormer 9acult) o pplied Science. 'heir results lie etween thee+tremes shown. Note7 6)ear programs such as d$anced Scienceare treated as e+tended programs as this is how the) are usuall)graded.

    16

  • 7/26/2019 1_SRI

    17/25

    01_SRIQuantifying Student Performance

    #.3 Merit a"ards

    ;egrees awarded with !erit are not $er) common7 onl) the9acult) o Commerce and Economics and the Board o Studies in'a+ation are -nown to award degrees with !erit# and onl)

    Commerce and Economics had signi4cant numers o graduandsin session 1# 1""H. wards with !erit are selected in a similar wa)to :onours in integrated programs.

    igure ."3 acu&ty of 4ommerce 5 ,conomics (Merit*

    0

    5

    10

    15

    20

    25

    )'*

    o%era((

    )'*

    yr2+

    )'*

    yr1,3

    )'*

    ma-or

    )'*

    ma-or

    yr1,3

    .P'

    o%era((

    .P'

    yr2+

    .P'

    yr1,3

    .P'

    ma-or

    .P'

    ma-or

    yr1,3

    Student 0anking 1ndicator

    6

    studentsmismatc

    hed

    '((

    35021300 'ccounting inance

    3502 om oter

    3543 c

    47 ommercea

    'he BCom is awarded with !erit to H5O o BCom/B graduands#to A%O o BCom graduands ut to onl) HO o BEc graduands. llprograms show a strong dependence on the ma2or studies a$erage#although the ma2or with the largest enrolments# ccounting and9inance# could use an) o the SD( satisactoril).

    #.4 $%tended programsSince the grade o :onours in e+tended programs is largel) asedon the 4nal )ear o stud)# we would not e+pect a$erages ta-en o$erthree or our )ears to e a particularl) good predictor o 4nal:onours grade. 'his is oser$ed in practice. (nterestingl)# the )ear1 to overall a$erage is oten as good or no worse than0 apredictor o :onours grade than the )ear 1 to ma&or-studiesa$erage# suggesting that the lac- o a ma2or6studies W! or G3would not e a disad$antage in selecting students or the e+tended)ear.

    'wo e+amples o aculties with signi4cant numers o e+tended6:onours graduates are rts and Social Science and Co9 ie

    17

  • 7/26/2019 1_SRI

    18/25

    01_SRIQuantifying Student Performance

    Science and Science & 'echnolog) ha$e large numers ut the)are spread $er) thinl) o$er 4t) programs8 while Commerce &Economics has too ew :onours graduates or an) useulconclusions to e drawn0.

    igure .". acu&ty of Arts and Socia& Science (e-tended +onours*

    0

    5

    10

    15

    20

    25

    30

    35

    40

    45

    )'* o%era(( )'* yr2+ )'* yr1,3 .P' o%era(( .P' yr2+ .P' yr1,3

    Student 0anking 1ndicator

    $/

    onour"mi"matc0ed

    '((

    3400 'rt"

    3420 Socia( Science

    Arts and Social Science. 'he acult) trend ollows that o 55 Artssince H%O o the 1""H session 1 graduating class and HO o:onours graduands came rom that program. !a2or6studiesa$erages ha$e een omitted as this acult) does not identi)ma2ors using the program code. $erages are somewhat less

    successul in predicting :onours in %5 .ocial .cience.

    igure ."7 4o&&ege of ine Arts (e-tended +onours*

    0

    5

    10

    15

    20

    25

    30

    35

    40

    45

    50

    )'*

    o%era((

    )'*

    yr2+

    )'*

    yr1,3

    )'*

    ma-or

    )'*

    ma-or

    yr1,3

    .P'

    o%era((

    .P'

    yr2+

    .P'

    yr1,3

    .P'

    ma-or

    .P'

    ma-or

    yr1,3

    Student 0anking 1ndicator

    $/

    onour"mi"matc0ed

    '((

    4800 ine 'rt"

    4803 'rt :eory

    o!A*ni,uel) among e+tended programs# the :onours grades o the Hgraduands in H5Art 5heorywere predicted ) e$en the simplest

    18

  • 7/26/2019 1_SRI

    19/25

    01_SRIQuantifying Student Performance

    SD( the de4nition o Fma2or studies= is not particularl) o$ious orCo9 programs# so the pea- in this area ma) e an aerration0.

    19

  • 7/26/2019 1_SRI

    20/25

    01_SRIQuantifying Student Performance

    * SRI selection

    &.1 '( )uality

    'he results o the pre$ious section are summarised in 9igure M.1#which shows the proportion o mismatched :onours grades across

    all ma2or integrated programs at *NSW# weighted ) the numero graduands.

    igure 7"# S01 se&ection

    10#9

    7#6

    17#2

    8#6

    15#3

    12#0

    9#0

    17#2

    9#1

    16#0

    0

    2

    4

    6

    8

    10

    12

    14

    16

    18

    20

    ;%era(( Ignore year 1 Ignore fina( year *a-or "tudie" *a-or "tudie"

    ece&t fina( year

    Student 0anking 1ndicator 8rou%

    $/

    onou

    r"mi"matc0ed

    )'*

    .P'

    'wo o the 4$e groups are o$iousl) unsatisactor)7 those thatdiscount the 4nal )ear. 'here is little to choose etween the

    remaining three SD( groups. 'he ma2or6studies group is not acandidate or reasons o comple+it)# while the remaining our arerelati$el) eas) to maintain centrall). lthough ignoring )ear 1produces the est results# resetting a counter at the end o stage 1is not a completel) reliale method since the cohort does notprogress as a group. (t is also a tas- that must e perormed )the program authorit) or each student7 there is no pro$ision inNewSouth Student or automaticall) resetting the counter.

    *sing an o$erall a$erage appears to produce acceptale results.'he ,uestion is# should it e a weighted a$erage mar- or grade

    a$erage>

    &.2 *+M versus ,P+

    W! and G3 a$erages produce similar results# with the mar-a$erage slightl) superior.

    rguments in a$our o a G3 include7

    1. 'he G3 scale can e the M6point *C s)stem used or transersetween institutions# a de actostandard throughout ustralia.

    %. G3 calculations are easier to automate than W! in

    NewSouth Student.

    20

  • 7/26/2019 1_SRI

    21/25

    01_SRIQuantifying Student Performance

    3. rogue low mar- 9# ;9# or $er) low ail0 has a $er)sustantial afect on W! ut somewhat less in?uence on G3.

    'he ollowing are arguments or adopting W!7

    1. W!s are ased on the mar- scale used throughout *NSW and

    well understood ) staf and students.

    2. 'here are too ew passing grades at *NSW or the G3 to airl)represent student progress. student with consistent mar-saround 5 will oten recei$e a lower G3 than another whoseresults $ar) etween A5 and M5.

    3. 'he ma2orit) o *S uni$ersities ha$e mo$ed to a letter Q/s)stem in order to discriminate etter etween students=results# and their G3 calculations ha$e a similarl) e+pandedrange. *NSW grades cannot easil) e modi4ed this wa).

    4. 'here is a ris- that Fgrade argaining= ) students would nolonger e limited to mar-s 2ust elow a pass i G3 were to eadopted as the *NSW measure. Students recei$ing a resultslightl) elow a credit# distinction or high distinction ma) eencouraged to challenge it ecause o the impact a small mar-change near a grade oundar) can ha$e on G3.

    5. marginal ailure is penalised in the G3 calculation relati$el)more than in a W! calculation. nl) MMO o :onoursrecipients in session 1# 1""H had a totall) ail6ree record#although "AO o :onours Class 1 did so.

    6. W!s could e con$erted to G3s or transer purposes# or orstudents see-ing entr) to o$erseas institutions or graduatestud). lthough the *ni$ersit) o ueensland uses a M6pointG3 or internal purposes# it is not completel) compatile withthe *C M6point G3.

    &.3 (ecommendations

    (n $iew o the ao$e discussion# the ollowing recommendationsare made7

    that *NSW adopts uniorm# cumulati$e W! as the preerred

    measure o student perormance8 that 9aculties e strongl) encouraged to e+press their eligiilit)

    or ran-ing re,uirements or merit6ased awards such as#interalia# :onours in integrated programs# entr) to the :onours)ear where separate# ;egree with !erit# and pri

  • 7/26/2019 1_SRI

    22/25

    01_SRIQuantifying Student Performance

    + ,utstandin# Issues

    'he ollowing are put orward as matters that need to e resol$edi the ao$e recommendations are adopted. 9aculties ma) wish tocomment on these or related matters to assist the cademic Board

    in estalishing the est polic) or *NSW.1. :ow would the W! e calculated or e+isting students>

    Should consideration e gi$en to an e+6gratia top6up to theW! to compensate or changing scales and standards> rshould aculties agree to appl) more lenient cutofs in coming)ears.

    2. 'he W! will e calculated or students in e+tended :onoursprograms as a matter o course. While the grade o :onoursawarded ma) still e ased on the :onours )ear alone# there isa potential or conusion i the cumulati$e W! ears little

    relationship to the 4nal grade.

    3. (s there a case or using o$erall W! or pre6:onours W! incomparing students or the award o postgraduate scholarshipse+tended6:onours )ear mar-s are calirated diferentl) roma$erages in integrated programs0> ;o we need anotherrecaliration scheme>

    4. Will courses granted transer credit e included in the W!>

    5. :ow will comined programs e handled NewSouth Studentsupports calculating a single# comprehensi$e $alue onl)0> 9or

    e+ample# would the BSS! accept a BSc/B student as acandidate or the :onours )ear in !athematics on the asis oa credit a$erage in all courses# including aw>

    6. What ran-ing inormation should e orwarded to the*ni$ersit) !edals Committee with acult) recommendations>Can this e a relati$el) standard sumission>

    )ppendix- .ualifyin# /alues

    9or each program in each acult)# the optimum cutof or ,uali)ing$alue was estalished or e$er) grade o :onours# using the

    strateg) descried in susection .1 p. "0. 'he cutofs are listed inthe tale at the end o this document. 'he range o cutofs or thesame :onours grade across all programs ma) e a measure o theconsistenc) o assessment within the acult) other interpretationsare possile0. 'he range is presented in the three 4gures elow.

    Since the aim was to e,ualise mismatches# the cutofs arerelati$el) pessimistic. 'hat is# aculties ma) wish to adopt slightl)lower ,uali)ing $alues in order to disad$antage ewer students.

    9igure .1 shows :onours 1 cutofs in terms o W!8 9igure .%e+presses :onours %/1 cutofs in G3 units or $ariet). 9igure .

    shows oth :onours %/% and !erit 9acult) o Commerce andEconomics onl)0 in W! units. 'he ars lin- the minimum and

    22

  • 7/26/2019 1_SRI

    23/25

    01_SRIQuantifying Student Performance

    ma+imum $alues across the programs o the indicated acult). 9ore+ample# the lowest ,uali)ing W! or :onours Class 1 or an)program in the 9acult) o Engineering was M%.M and the highestwas M..

    "onours #. 'he highest standard is set ) .cience21edicine and3sychology8 the lowest )Industrial "esign in Built En$ironmentand the straight 1edicine program. No acult) grouping all

    students treated the same0 ,uite re,uires a ;istinction MA.50cutof.

    "onours $%#.!ost acult) a$erages lie around the A.5 G3 all6credit

    a$erage# roughl) e,ui$alent to a W! o M50. Engineering displa)sthe largest spread.

    23

    igure A"# Qua&ifying WAM for +onours 4&ass #9

    integrated degrees

    70

    71

    72

    73

    74

    75

    76

    77

    78

    79

    80

    minimum maimum entire facu(ty

  • 7/26/2019 1_SRI

    24/25

    01_SRIQuantifying Student Performance

    "onours $%$ and Merit.arge ranges are e+hiited ) Engineering andthe ew programs in the Sciences aculties. !edicine and BuiltEn$ironment oth re,uire o$er a A a$erage. !erit awards areclearl) the 6)ear e,ui$alent to second class :onours in 6)earprograms.

    W! cutofs or all rele$ant programs are taulated elow.

    24

    igure A"2 Qua&ifying WAM for +onours 4&ass ' $ivision '9 (integrated

    degrees*9 and Merit (4ommerce 5 ,conomics*

    60

    61

    62

    63

    64

    65

    66

    67

    68

    69

    70

  • 7/26/2019 1_SRI

    25/25

    01_SRIQuantifying Student Performance

    :a)&e A"# Qua&ifying WAMs for integrated %rograms

    acu&ty ProgramNum)er of8raduands

    6 +onours

    Qua&ifying WAM for +onours

    4&ass # 4&ass '$ivision #

    4&ass '$ivision '