英语GMAT 机考 FAQ.docx

英语GMAT 机考 FAQ.docx

《英语GMAT 机考 FAQ.docx》由会员分享，可在线阅读，更多相关《英语GMAT 机考 FAQ.docx（17页珍藏版）》请在冰豆网上搜索。

英语GMAT机考FAQ

CATFAQ

I’vereceivedanumberofinquiriesfromthecommunityabouttheGMATalgorithm,soIthoughtitbesttoreplyinarticleform.

1.What’sanalgorithm?

Analgorithm,generally,isausuallyefficientsetofwell-definedstepsthefollowingofwhichresultsinthesolutiontosomepre-definedproblem.InthecaseofaCATalgorithm,theproblemistoreliablyandefficientlyestimateastudent’sabilityinareasonableamountoftime.SomeCATalgorithmsseektosolvethisproblembyselectingonequestionatatime,eachsubsequentquestionselectedbasedonallofthestudent’spriorresponses.Otheralgorithmslookonlyatthemostrecently-answeredquestion.Stillothersevaluateresponsestospecificgroupsofquestions.CATalgorithmsalsovarywithregardtotheexplicitcriteriatheyusetoselectthenextquestion（orsetsofquestions）toadminister.Sometrytominimizetotalmeasurementerror.Otherstrytomaximizetheprecisionandaccuracyofmeasurementforeachquestionadministered.Stillotherstrytoselectquestionsthatwillmostrefinethecurrentabilityestimate.Asaconsequence,CATalgorithmscanvarygreatlyfromoneanother,dependingonthespecificimplementationofthealgorithm,andtheintentofthealgorithmdevelopers.

2.WhydoestheGMATuseanalgorithmwhenthelinearLSATseemstobeaprettydecentgaugeofproficiency?

OneofthecommongoalsinusingaCATalgorithmistoreducethenumberofquestionsastudentneedstoanswerinordertoestablish,toaspecifiedlevelofreliability,anestimateofthestudent’sability.CATsareoftenmoreefficientthanlineartests,andsofewerquestionsareneededtoreachadesiredlevelofreliability.TheLSATneedsover100itemstoreachthatlevel,whiletheGMATneedsfewerthan80toreachacomparablelevel.

3.IstheentireGMATadaptive?

Almostalllarge-scalestandardizedtestscontainsomenumberof“experimental”or“pretest”questionsthatareadministeredtothestudent,butdonotcounttowardthestudent’sfinalscore.Thisissimplyawayforthetestmakerstogatherdataonthequestions,inordertodeterminehowdifficulttheyareandhowwelltheydistinguishbetweenstudentsatdifferentabilitylevels.Theyalsousethedatacollectedtoidentifybadquestions,sothattheycaneliminateorfixthembeforetheycount.Sometests,liketheLSAT,includeallofthepretestquestionsinasinglesection.Others,liketheGMAT,interminglethepretestquestionswiththeoperationalones.Whichsectionisthepretestsection,andwhichquestionsarethepretestquestions,isusuallyawell-guardedsecret.Itisgenerallybadstrategytospendtimetryingtoguesswhetheragivenquestionisoperationalornot–thepriceofguessingincorrectlyisjusttoohigh.

4.HowdoestheGMATselectwhichquestionsIget?

CATsliketheGMAThaveablueprint—asetofspecifications（difficulty,questiontype,contentarea,etc.）thatdefinewhichquestionsyousee.Atthesametime,eachquestionhascertainstatisticalcharacteristicsthatthealgorithmuses,basedonyourresponse,toestimateyourquantitativeorverbalability.Thealgorithmlooksatyourperformanceonthequestionsyouhavealreadyansweredandthecharacteristicsofeachquestionremaininginthepoolandthenselectsforyouthequestionthatsimultaneouslybestsatisfiestheblueprintandprovidesthemoststatisticalinformationitcan,togeneratethebestestimateofyourability.

LookforPart2nextweek!

1.Myscoredoesn’tseemtomatchmyperformance:

Ionlygotafewquestionswrong,butmyscoreisn’tashighasIthoughtitwouldbe/Igotabunchofquestionswrong,yetmyscoreseemshigherthanitshouldbe.

Mostexamsarelinearassessments,liketheSAToryour10thgradehistoryfinal.Thesearescoredbycountingthenumberofquestionsyouanswercorrectly,andsometimesbypenalizingforeachquestionyouanswerincorrectly.Theresult,arawscore,isthenconvertedtoascaledscore,likethe600-2400rangefortheSAT.

Acomputer-adaptivetest（CAT）worksverydifferently.Itdoesn’treallycareasmuchabouthowmanyyougetrightorwrong,butratherwhichquestionsyougetrightandwrong.TheCATalgorithmestimatesyourabilitybasedonavarietyofcriteria,includingthedifficultyofaquestion.Aftereachquestion,itevaluatesyourresponseandupdatesthisestimate.Whenthetestisover,thealgorithmconvertsyourquantitativeandverbalabilityestimatesintothequantitativeandverbalscaledscores,andthenseparatelycombinesyourquantitativeandverbalabilityestimatestocalculatetheoverallscore.

2.DothefirstXnumberofquestionsmattermore?

ManyvariablesthatcomeintoplaywhentheCATselectsyournextquestion.OneofthemistheCAT’scurrentestimateofyourability.Itusesthisestimatetoselectquestionsthatwillbemostusefulinrefiningthatestimate（ifyou’reahighperformingstudent,givingyoulowdifficultyquestionsisn’tusuallyasusefulindiscerningyourtrueabilityasgivingyouharderquestions,andviceversa）.Whatisimportanttorememberisthatyoushouldnottrytoguesshowyouaredoingbywhetherthequestioninfrontofyouseemseasyordifficult;everyquestiondeservesyourfullattention.Withthatunderstood,unlessyouhavecompletelybombedthetest,itisusuallythecasethatmissingacoupleofveryhardquestionslateinthetestwillhaveasmallereffectonyourfinalscorethanmissingacoupleofveryeasyquestionsearlier,notbecauseoftheirpositionwithinthetestbutbecauseoftheirlevelsofdifficulty.

3.Howsevereisthepenaltyfornotfinishingasection?

Thepenaltyissignificant.Youcanexpectyourscaledscoretodecreasebyroughly1pointforeveryquestionthatyoudon’tanswer.Forexample,ifyoucorrectlyanswereveryquestionyouencounterbutfailtoanswerthelastfive,yougenerallywon’tscorehigherthana46.

4.ItooktheGMATandgota710,44q/44v/6AWA.Afriendofminehappenedtotakethetest6dayslaterandgettheexactsamequant/verbalscaledscoresbuthegota720.Howthiscouldhappen?

BoththeindividualsectionscoresandtheoverallscorearecalculatedusinganestimateofyourMathandVerbalabilitiesderivedfromyourperformanceontheCAT.Youroverallscoreisnotcalculatedfromyoursectionscores.Becauseyourunderlyingabilityestimatemightbeslightlydifferentfromyourfriend’s,youroverallscoresmightbedifferent.Forexample,therearearangeofabilityestimatesthattranslateintoaVerbalscoreof40,andtherearearangeofabilityestimatesthattranslateintoaMathscoreof42.Dependingonwhichspecificestimateiscalculatedforyou,youroverallscorecouldrangefrom660to680.PleasenotethattheStandardErrorofMeasurement（SEM）ontheoverallscoreforGMATis29points,soscoresof660–680allfallwithinthestandarderror.

Checkbacknextweekforpart3.Untilthen,doyourhomework!

Howcanmyoverallpercentilebehigherthanbothmyquantitativeandverbalpercentiles?

Whyisn’tittheaverageofthetwo?

Youcanbeinthe99thpercentileoverallevenifyoudidn’tscoreinthe99thpercentileoneitherofthesections,becausetheoverallscoreiscalculatedseparatelyfromthesectionscoresandtheirpercentiles.So,forexample,youcouldgeta48onQuantitative（86thpercentile）,a45onVerbal（98thpercentile）,anda760overall（99thpercentile）.

Arethequantitativeandverbalsectionsweightedequallyinthetotalscore?

Technically,yes,inthattheestimatesofyourquantitativeandverbalabilitiesthattheCATproducescontributethesameamounttoyouroverallscore.However,theverbalsectionisgenerallymoredifficult,andthereforehasagreatereffectonyourpercentilerank.If,forexample,youscoreda40onboththeQuantitativeandVerbalsections,yourpercentilerankforQuantitativewouldbe61st,forVerbalwouldbe91st,andoverall（650）wouldbe84th.

Whyarescoresabove51rare?

Whydoesthescalegoupto60?

Cananyonegeta52?

Forpsychometricreasons,GMAChastruncatedthescaleat51（theydonotreportsectionscoreshigherthan51）.

WhyisitsodifficulttocreateagoodCAT?

ACATneedstodomanythingswellinordertoreliablyandaccuratelyestimateyourability.Itrequiresarobustalgorithmtoestimateyourability,acomplexbutspeedymechanismtoidentifythebestquestionforyoutoseenext,arichpoolofquestionsfromwhichtoselectthequestions,andapowerfulscoringalgorithmthattranslatestheabilityestimateintosomethingmeaningful.Eachtestquestionhasmanycharacteristicsthatneedtobesimultaneouslyconsideredintheselection.Thestatisticalcharacteristicsofthequestionsallneedtobedeterminedbeforehandthroughaprocessknownaspretesting.Many,manyquestionsareneededinordertobeabletoprovideaccurateassessmentforallabilitylevels.Andallofthosequestionsneedtobecarefullyconstructed,reviewed,andstatisticallyalignedsothattheycontributemeaningfullytoyourabilityestimate.

爷要听什么?

随便你唱什么。

吟霜侧著头,深思了一会几,再掉头看向云鹏

暗中布置著什么,云鹏发现她在装修吟霜那几间卧室了,他怀疑的问:

从此,葛府中失去了吟霜的影子。

云鹏魂牵梦萦,实在无法忘怀吟霜。

朝朝井上可奈_爽到饥渴妇人流_井上可奈暮暮,这片

似悲似乐的情绪把他给击倒了。

他俯视著她,不由自主的揽住了她的头,喃喃的说:

要为他物色人选的念头仍然未消,感于井上可奈_爽到饥渴妇人流_井上可奈她那片好意,他也就无可奈何了。

于是,这天,云鹏

吟霜,他低唤井上可奈_爽到饥渴妇人流_井上可奈,点了点头,慨然的说:

薄命怜卿甘作妾!

薄命吗?

吟霜低

啊呀,我的天!

云鹏感到一阵头晕目眩,用手拍著额,他一叠连声的叫葛升备井上可奈_爽到饥渴妇人流_井上可奈马,

在哪儿?

我让她待在吟霜的那间屋子里呢,你去看看好吗?

怎井上可奈_爽到饥渴妇人流_井上可奈_爽到饥渴妇人流_井上可奈井上可奈么可以让她住吟霜的房间!

云鹏十分不乐,却不好发作。

看到弄玉一片喜孜孜的样

双眉暗锁,心事谁知我?

旧恨而今较可,新愁去后如何?

白狐6/井上可奈_爽到饥渴妇人流_井上可奈46

呢?

内容,因井上可奈_爽到饥渴妇人流_井上可奈此,垂著眼睑,低俯著头,她不敢仰视云鹏。

祝完,他再看看那密井上可奈_爽到饥渴妇人流_井井上可奈_爽到饥渴妇人流_井上可奈上可奈密深深的荒林,重重的跺了一下脚,带著满怀井上可奈_爽到饥渴妇人流_井上可奈的无可奈何与怆恻之

有些诧异,也有些不耐。

但是,屋里已响起一阵叮叮咚咚的琴声,好熟悉!

接著,一个圆润

变吧!

霜。

怕弄玉寒心,他不敢形于色,而弄玉呢?

她已把吟霜的井上可奈_