Modeling learner satisfaction in an electronic instrumentation and measurement course using structural equation models

190 IEEE TRANSACTIONS ON EDUCATION, VOL. 52, NO. 1, FEBRUARY 2009 Modeling Lea ne Sa is ac ion in an Elec onic Ins umen a ion and Measu emen Cou se Using S uc u al Equa ion Models Se gio L. To al, Fede ico Ba e o, Ma ia R. Ma ínez-To es, Se gio Galla do, and Ma io J. Du án Abs ac —The p e ailing endency in mode n uni e si y e o ms is owa ds “how people lea n,” ollowing a lea ne -cen- e ed app oach in which he lea ne is he main ac o o he eaching-lea ning p ocess. As a consequence, one o he key in- dica o s o he eaching-lea ning p ocess is he measu emen o lea ne sa is ac ion wi hin he class oom. Lea ne sa is ac ion has adi ionally been measu ed using su ey esponses o a s anda d lea ning su ey. Howe e , mo e scien i ic analysis should be pe o med o assess adequa ely no only lea ne sa is ac ion bu also he main dimensions ha ha e a posi i e impac on lea ne sa is ac ion. The pu pose o his pape is o de ine a s uc u al and measu emen model in which causal ela ionships among hese di e en dimensions a e adequa ely es ablished. The me hod- ology is based on a mul i a ia e eg ession model (S uc u al Equa ion Models) o es ablish scien i ically a s uc u al model o lea ne sa is ac ion wi hin a class oom, measu ing i s alidi y and eliabili y. The p oposed app oach has been applied o model lea ne sa is ac ion in an elec onic ins umen a ion cou se a he Uni e si y o Se ille, Spain. The esul s and implica ions o his s udy will con ibu e o imp o e s uden sa is ac ion wi h espec o he dimensions conside ed. Index Te ms—Educa ional echnology, elec onic equipmen , labo a o ies, lea ning sys ems, planning. I. INTRODUCTION THE no ion o lea ne -cen e ed educa ion has been in exis ence o a long ime [1], [2]. Ne e heless, his concep is eeme ging in he Eu opean coun ies due o e o ms ha a e o be implemen ed in 2010 inside he Eu opean Highe Educa ion A ea (EHEA) [3]–[5]. Lea ne -cen e ed p ac ices mo e he ocus om he eache o he s uden , paying mo e a en ion o he lea ning pe o mance a he han he ins uc ion me hodology. Ins uc ion based on a lea ne -cen e ed ame- wo k p o ides oppo uni ies o lea ne s o d aw on hei own expe iences and in e p e a ions o he lea ning p ocess [6]–[9]. These p ac ices ega d lea ning as a li elong p ocess a he han as a p ocess ha akes place only in one’s you h, ollowing he end o he majo i y o cu en highe educa ion e o ms [10]–[12]. Manusc ip ecei ed Ma ch 9, 2007; e ised Feb ua y 27, 2008. Cu en e - sion published Feb ua y 4, 2009. S. L. To al, F. Ba e o, S. Galla do, and M. J. Du án a e wi h he Elec ical Enginee ing Depa men , Uni e si y o Se ille, Se ille 41092, Spain (e-mail: [email p o ec ed]; [email p o ec ed]). M. R. Ma ínez-To es is wi h he Business Adminis a ion and Ma ke ing Depa men , Uni e si y o Se ille, Se ille 41092 Spain. Colo e sions o one o mo e o he igu es in his pape a e a ailable online a h p://ieeexplo e.ieee.o g. Digi al Objec Iden i ie 10.1109/TE.2008.924215 In acco dance wi h his app oach, lea ning is conside ed as a cons uc i e p ocess. Fu he mo e, as lea ning is mo e mean- ing ul and ele an o he s uden , eaching e iciency is also in- c eased. This e ec is pa icula ly impo an in subjec s wi h a high p ac ical wo k con en in which he skills and abili ies o lea ne s need o be imp o ed, especially when s uden s ge in ol ed in he lea ning p ocess, assuming esponsibili y o hei own p og ess [13]. Howe e , he eaching p ocess should be cen e ed no only on he lea ne ’s ac i i ies, bu also on he lea ne ’s sa is ac ion, aking in o accoun wha is ele an o he s uden . The implemen a ion o lea ne -cen e ed me hodologies demands a p io analysis o he subjec , bo h o unde s and wha is ele an o he s uden and o iden i y he dimensions ha ing a highe in luence on lea ne sa is ac ion. Sa is ac ion ela es o pe cep ions o being able o achie e success, and eelings abou he achie ed ou comes [14], [15]. F om his pe spec i e, se e al s udies ha e explo ed s uden sa - is ac ion o imp o e cou se planning [9], [16]. Some imes, hese s udies a e limi ed o one-dimensional pos - aining pe cep ions o lea ne s [17], [18]. Ope a ionally, lea ne sa is ac ion is oo o en measu ed wi h “happy shee s” which ask lea ne s o a e how sa is ied hey we e wi h hei o e all lea ning expe ience. Howe e , he no ion o lea ne sa is ac ion mus be explo ed h ough a mul idimensional analysis ha conside s a wide a- ie y o c i ical dimensions, so as o p o ide e ec i e me ics ha guide imp o emen s in ins uc ional design. Lea ne sa is ac ion scales ha e been used o assess eaching quali y wi hin In o ma ion Sys ems esea ch. Use in o ma ion sa is ac ion (UIS) and end-use compu ing sa is ac ion (EUCS) ins umen s a e examples o use sa is ac ion scales [19]. Bo h o hem measu e se e al eaching quali y ac o s wi h a a ying numbe o su ey i ems o each ac o [20], [21]. The main d awback o hese me hods is ha hey a e p ima ily ocused on eaching quali y an eceden s o lea ne sa is ac ion, ins ead o conside ing he lea ne as he main an eceden . This consid- e a ion is o pa icula impo ance in lab subjec s o in asyn- ch onous lea ning ac i i ies, whe e he ole o he lec u e essen- ially consis s o encou aging s uden s’ ini ia i e and mo i a ion o ob ain a high lea ning pe o mance. In hese con ex s, now being p omo ed in he EHEA, he con en is no so as impo an as a e he new compe encies (combina ion o knowledge, skills and a i ude) ha he s uden s should de elop [22]. To assess he ex en and speci ic na u e o lea ne sa is ac- ion, di e en dimensions should be heo e ically and ope a- ionally de ined. Ac ually, many s udies ha e been conduc ed on his opic, employing lea ne dimensions as an eceden s. These 0018-9359/$25.00 © 2008 IEEE Au ho ized licensed use limi ed o: IEEE Xplo e. Downloaded on Feb ua y 17, 2009 a 04:29 om IEEE Xplo e. Res ic ions apply. TORAL e al.: MODELING LEARNER SATISFACTION 191 s udies ocus on p io lea ne expe iences in e-lea ning cou ses [23]–[25], on lea ne a i udes owa ds compu e s [26], [27], on lea ne compu e anxie y [28], on lea ne In e ne sel -e icacy [29], and on lea ne ini ial compu e skills [30]. The pu pose o his pape is no only o iden i y he an- eceden s o lea ne sa is ac ion, bu also o de ine a s uc u al and measu emen model in which causal ela ionships among he di e en dimensions a e adequa ely es ablished. The s a ing poin was he cu iculum edesign o an elec onic ins umen a ion and measu emen cou se [25], whe e he an eceden s o lea ne sa is ac ion we e iden i ied. These an- eceden s will be included in a gene al model o highligh he ela ionship among he di e en dimensions p e iously iden i ied. The app op ia e me hodology o pe o m his ask is S uc u al Equa ion Modeling (SEM). SEM g ows ou o , and se es pu poses simila o mul iple eg ession, bu in a mo e powe ul way. SEM allows modeling o he ela ionships be ween mul iple independen and depen- den dimensions simul aneously, while eg ession models can analyze only one laye o linkages be ween independen and de- penden dimensions a a ime. The e o e, SEM may be used as a mo e powe ul al e na i e o mul iple eg ession, pa h anal- ysis, ac o analysis, ime se ies analysis, and analysis o co- a iance. While complex in e ela ionships canno be ully ex- plo ed by hese echniques, SEM has he ad an age o dissec ing hese ela ionships, assessing he o al e ec s o a iables on one ano he . Addi ionally, SEM p o ides he associa ions be ween a iables and es ima es he s eng hs o hese ela ionships in an in eg a ed model [31]. The p inciple adop ed by SEM is based on de e mining model pa ame e s so as o eplica e in he bes possible manne he co a iance ma ix o he dimensions in he model sys em. This app oach acili a es he es ima ion o com- plex model sys ems, de e mining causal ela ionships among a se o dimensions, which may include o dinal esponse a i- ables as well as con inuous measu emen s [32]. SEM has been success ully used in se e al disciplines including, o example, sociology, psychology o ma ke ing [33]. In he ield o edu- ca ional sciences, SEM has been applied o de elop accep ance models [34] o o assess e-lea ning ools [35]. This s udy ex ends he use o SEM o lea ning sa is ac ion modeling ollowing a simila p ocedu e. The pape is s uc u ed as ollows. Sec ion II desc ibes he exis ing me hods o e alua ing lea ne sa is ac ion, pa icula ly in a labo a o y o p ac ical lea ning en i onmen , and gi es he esul s ob ained om a case s udy, an elec onic ins umen a ion and measu emen cou se o e ed by he Uni e si y o Se ille. Sec ion III illus a es he p oposed SEM-based me hodology o alida ing he lea ne sa is ac ion model. Sec ion IV discusses esul s and implica ions and, inally, conclusions a e d awn in Sec ion IV. II. LEARNER SATISFACTION MEASUREMENT IN LABORATORY TEACHING:ACASE STUDY T adi ionally, measu emen o lea ne sa is ac ion has been applied o assess in o ma ion and managemen sys ems, in- cluding class oom eaching in adi ional educa ional con ex s [20], [32], [36]. This measu emen should no be e alua ed using a single-i em scale, such as global sa is ac ion, and should inco po a e di e en aspec s o lea ne sa is ac ion, i i is o become a use ul diagnos ic ins umen . Mo eo e , a lea ne sa is ac ion measu emen de eloped o adi ional educa ional con ex s is no app op ia e o a labo a o y o p ac ical lea ning en i onmen , whe e he ole o he s uden is comple ely di e en o ha o a s uden in a lec u e [19]. Fo ins ance, he deg ee o he s uden pa icipa ion and ini ia i e, he way in which he educa ional ma e ial is deli e ed, and he possibili ies o eedback and in e ac ion wi h physical ins umen a ion a e qui e di e en in a labo a o y en i onmen . As lea ne sa is ac ion is an an eceden o use in en ion [37], he e is a need o de elop a comp ehensi e ins umen o measu ing lea ne sa is ac ion wi hin a p ac ical o labo a o y con ex [19], [25]. In [25], a ecen ly de eloped comp ehensi e ins umen o measu ing lea ne sa is ac ion wi hin a p ac ical o labo a o y con ex is desc ibed. A cou se o e ed du ing he inal semes e o he Telecommunica ion Enginee ing deg ee a he Uni e si y o Se ille, “Elec onic Ins umen a ion and Measu emen Lab,” is used as a case s udy. This op ional cou se consis s o 7.5 Spanish-c edi s o 75 h, one Spanish-c edi being equi alen o 10 h o lessons. The Telecommunica ion Enginee ing deg ee is o ganized in i e academic yea s wi h e e y yea being di ided in o wo semes e s. The cou ses usually las one semes e and mos o hem consis o six c edi s on a e age. The lab cou se analyzed he e is augh in he second semes e o he inal yea , and e e y yea abou 60 o 90 s uden s en oll in he cou se. The main goal o he cou se is o p o ide s uden s wi h an unde - s anding o he ope a ing p inciples and applica ions o a se- lec ed ange o basic and ad anced ins umen s, such as logic analyze s, oscilloscopes, spec um analyze s, LCR me e s, e c., while imp o ing s uden s’skills h ough labo a o y wo k ex- pe ience. The lab wo k is based on a “hands on”ins uc ion ocus on enginee ing opics such as modula ion echniques, mi- c op ocesso s sys ems analysis, e lec ome y p inciples, ixed elephone basics, ins umen a ion buses, and so on. The s uden s lea n abou ins umen s and lab equipmen by using hem o he analysis o in e es ing elec onic enginee ing sys ems and p in- ciples. The cou se also enables s uden s o s eng hen o he im- po an abili ies, in a eas such as collabo a i e wo k, inno a ion and esea ch skills. The cou se is o ganized in wo sepa a e lab g oups, wi h 30 o 45 s uden s pe g oup. G oup “A” akes place on Thu sdays, om 4 PM o 9 PM. G oup “B” akes place on F idays, om 9 AM o 2 PM. The s uden s, wo king in g oups o wo o h ee, ha e o a end o wel e o hese 5-h lab sessions. The e o e, du ing he cou se each s uden a ends 12 sessions, o 5 h each, and will wo k a wel e di e en wo kbenches. Each wo kbench is composed o di e en elec onic ins umen s and p o o ype boa ds ha s uden s ha e o use and es , espec i ely. The 12 lab sessions composing he cou se a e o ganized as ollows. •Session 1: Design and analysis o elemen a y dc me e s and mul ime e s as measu emen ins umen s. •Session 2: Design and analysis o ac me e s and mul ime- e s, including he equency esponse analysis o he me- e s and he implemen a ion o elemen a y hal / ull b idge ac ol me e s. Au ho ized licensed use limi ed o: IEEE Xplo e. Downloaded on Feb ua y 17, 2009 a 04:29 om IEEE Xplo e. Res ic ions apply. 192 IEEE TRANSACTIONS ON EDUCATION, VOL. 52, NO. 1, FEBRUARY 2009 •Session 3: Use o analog and digi al oscilloscopes, and a- milia iza ion wi h hei basic cha ac e is ics. This lab in- cludes he analysis o p obes, hei cha ac e is ics, pe o - mance, applica ions, and limi a ions. •Session 4: In oduc ion o spec um analyze s and hei p inciples o ope a ion. This lab wo k is used o desc ibe he ope a ion p inciples o spec um analyze s. •Session 5: Measu emen and cha ac e iza ion o passi e impedances and il e s using Whea s one B idges, LCR, and Spec um Analyze s wi h acking gene a o s. The limi a ions o p obes in high equency applica ions a e also s udied o unde s and hei in luence on measu emen e o s. •Session 6: In oduc ion o he use o logic analyze s and o he analysis o complex digi al elec onic and mic o- p ocesso s sys ems, including sys ems and mic ocon olle in e acing. •Session 7: Analysis o Gene al Pu pose In e ace Buses (GPIB) and hei applica ion o con olling elec onic equipmen . GPIB p o ocol is used o implemen i ual and emo e ins umen s. •Session 8: Analysis o complex analog elec onic sys ems. The lab wo k is based on he s udy o audio p inci- ples, audio ampli ie s classes and hei cha ac e iza ion me hods. •Session 9: A de ailed desc ip ion o he elec ical ime do- main e lec ome y (ETDR) mechanism, i s applica ions, and a me hod o in e p e ing ETDR signal wa e o ms a e p esen ed. •Session 10: In oduc ion o basic elephony concep s and undamen als. In his lab wo k, lea ne s s udy line ele- phone communica ions p inciples and analyze a eal ele- phone p o o ype. •Session 11: Desc ip ion o o he digi al modula ion ech- niques like FM and FSK, and hei applica ions. FM and FSK modula ion echniques a e analyzed in he ime and equency domains. •Session 12: Analysis o ad anced elec onic digi al sys- ems (like digi al signal p ocesso s o DSP) and p o ocols (like in e nal Pe sonal Compu e s a chi ec u e) using logic analyze s. P e ious eading is equi ed o unde s and he p inciples o ope a ion o he equipmen and ins umen s o be used in each session. Be o e beginning he labo a o y, ins umen handbooks and a de ailed handou o each ask a e a ailable o he s uden s. No o mal epo s a e equi ed o s uden s o e alua e hei wo k, bu hey ha e o answe on- he-spo ques ions du ing each lab- o a o y session. To assess he ex en and he speci ic na u e o lea ne sa is- ac ion, he di e en dimensions, and a s uc u al and measu e- men model aking in o accoun hese dimensions, mus be de- ined. Up o 10 dimensions, shown in Table I, we e conside ed in o de o assess lea ne sa is ac ion in he p io s udy [25]. The selec ion o dimensions was based on he echnological accep- ance model [38], because sa is ac ion is gene ally conside ed a cen al media o o lea ne beha io . Mos beha io esea che s would ag ee ha sa is ac ion in luences u u e usage in en ion and complaining beha io . S uden s wi h high le els o sa is- TABLE I DIMENSIONS OF LEARNER SATISFACTION Fig. 1. Lea ne sa is ac ion model. ac ion a e expec ed o ha e highe le els o euse in en ion and make ewe complain s [37]. A su ey based on hese dimensions was applied o he cou se in o de o imp o e i s o ganiza ion in acco dance wi h lea ne sa is ac ion measu emen s. The esul s ob ained highligh ed hose dimensions wi h a highe in luence on lea ne sa is ac ion, showing ha con en , use in e ace, ease o use, and mo i a ion we e he mos app op ia e o be ein o ced. Acco ding o his analysis, he cou se was edesigned, and pos implemen a ion esul s we e ob ained o show he imp o e- men s in he s uden s’de elopmen . Al hough he expe imen al esul s ob ained clea ly showed such an imp o emen , he esul ing model was e y simple. All he dimensions we e co ela ed wi h lea ne sa is ac ion, and he co ela ion alue was used in he analysis. Ne e heless, no in e ela ionships be ween he dimensions conside ed we e analyzed. These in- e ela ionships should be conside ed, as he co ela ion ma ix shows ha he e a e s ong co ela ions be ween se e al o he dimensions, and hese co ela ions can modi y he eal impac on lea ne sa is ac ion. As a consequence, he me hodology had o be modi ied om ha he one used in [25]. Speci ically, SEM is applied in his case in o de o alida e a inal model conside ing he dimensions lis ed in Table I. III. LEARNING SATISFACTION MODEL VALIDATION USING SEM The model o be alida ed is shown in Fig. 1. The dimension on he igh side is sa is ac ion, di ec ly d i en by he use in- e ace, ease o use, en husiasm and mo i a ion, and indi ec ly d i en by he es o dimensions. The dimensions on he le side a e he pu e independen a iables, while he in e media e dimensions a e dependen a iables ha may ac as an indepen- Au ho ized licensed use limi ed o: IEEE Xplo e. Downloaded on Feb ua y 17, 2009 a 04:29 om IEEE Xplo e. Res ic ions apply. TORAL e al.: MODELING LEARNER SATISFACTION 193 TABLE II VALIDATED SURVEY BASED ON THE LEARNER SATISFACTION DIMENSIONS den a iable wi h espec o one o he . In con as wi h classical eg ession models, SEM hypo hesizes ha dimensions such as hose a o emen ioned a e no di ec ly obse able, and a e be e modeled as la en a he han obse able a iables. As a conse- quence, hey a e indi ec ly measu ed h ough a se o indica o s. In his way, SEM makes i possible o dis inguish wo di e en ypes o e o s: e o s in equa ions, as shown by he pa h model, and e o s in he obse a ion o a iables [39]. The alida ed ques ionnai e p esen ed in [25] and Table II, was also employed he e, and was dis ibu ed o and comple ed by 284 s uden s en olled in he Elec onic Ins umen a ion and Measu emen Lab cou se. Each dimension is measu ed using Au ho ized licensed use limi ed o: IEEE Xplo e. Downloaded on Feb ua y 17, 2009 a 04:29 om IEEE Xplo e. Res ic ions apply. 194 IEEE TRANSACTIONS ON EDUCATION, VOL. 52, NO. 1, FEBRUARY 2009 se e al indica o s. C onbach’s alpha index (a eliabili y index associa ed wi h he a ia ion accoun ed o by he ue sco e o he “unde lying dimension”) was used o p o e he eliabili y o he ques ionnai e. This coe icien anges om 0 o 1. The highe he sco e, he mo e eliable is he gene a ed scale. A alue abo e 0.7 is an accep able eliabili y coe icien , al hough lowe h esholds a e some imes used in [40]. Table II shows he eliabili y analysis esul s, including he alue o C onbach’s alpha gi en in pa en heses unde he dimension. The in e -i em co ela ion associa ed wi h each i em is also shown in he second column o Table II be ween pa en heses. Thi y- h ee o he ques ions, he majo i y, epo sa is ac o y alues. I ems I5 and I20 we e ejec ed due o he low in e i em co ela ion alue. Once he se o i ems o measu ing he unde lying dimension has been de ined, he hypo hesis o he p oposed model will be alida ed. I he model is analyzed om igh o le (Fig. 1), i can be concluded ha sa is ac ion (S) is p ima ily d i en by he Use In e ace (UI), Ease o Use (EOU), and En husiasm and Mo i a ion (EAM). UI and EOU a e ela ed o he quali y and e ec i eness o he ins uc o and he ins uc ion (cogni i e dimension) while EAM loca es he ins uc o as a acili a o o knowledge and capabili ies ansmission (a ec i e dimension). Consequen ly, sa is ac ion is achie ed as a mix u e o cogni i e and a ec i e dimensions. Al hough ins uc o s end o imp o e only cogni i e dimensions, a ec i e dimensions should also be conside ed because hey p omo e posi i e eelings owa ds he subjec . A clea an eceden o hese h ee dimensions is Use Con- ol and In e ac i i y (UCI). Indeed, UCI mus be acili a ed by he way in which con en is deli e ed (Use In e ace), which should wo k bo h eliably and con enien ly (Ease o Use). A he same ime, Use Con ol and In e ac i i y also encou ages s uden s o pa icipa e ac i ely in he p ac ical wo k. As shown in Fig. 1, his dimension is he cen al elemen in he model. UCI is s imula ed by h ee independen dimensions and d i es he h ee dimensions wi h a di ec in luence o e sa is ac ion. Going om le o igh o he model, i e pu e independen di- mensions can be men ioned. Lea ning communi y (LC), lea ne esponsibili y (LR), and p e ious expe ience (PE) a e he h ee dimensions ha show he s uden ’s p o ile. These h ee ep e- sen an inpu o he cou se and, consequen ly, hey can be con- side ed as independen a iables in he p oposed model. Use Con ol and In e ac i i y is de e mined by he s uden p o ile. Howe e , he ins uc o is esponsible o guiding he s uden ’s ini ia i e and inc easing he use sa is ac ion. The las wo pu e independen a iables a e con en (CON), ela ed o he quali y and e ec i eness o he knowledge ans- mission, and eedback (FED), ela ed o he s eng hening o use s’lea ning ia e i ica ion (a ec i e dimension). Nei he o hese a iables ha e a di ec in luence o e sa is ac ion. The i s is modula ed by he use in e ace, he way in which his con en is deli e ed (cogni i e dimension). The second is modula ed by he en husiasm and mo i a ion o s uden s when pe o ming he p ac ical wo k (a ec i e dimension). O he dimensions could be conside ed, bu he inclusion o oo many dimensions could cause undesi able e ec s, such as model o e i . Once he model has been desc ibed, he nex s ep is o ali- da e i . The con i ma o y s uc u al and measu emen model is ob ained and alida ed using SEM. Acco ding o [40], i SEM is accu a ely applied i can su pass such i s -gene a ion ech- niques as P inciple Componen s Analysis, Fac o Analysis, Dis- c iminan Analysis, o Mul iple Reg essions. Speci ically, SEM p o ides a g ea e lexibili y in es ima ing ela ionships among mul iple p edic o s and c i e ion a iables, and allows modeling wi h unobse able la en a iables. Addi ionally, SEM es ima es he model wi hou con amina ion om measu emen e o s. The wo app oaches o causal modeling which appea in he li e a- u e a e as ollows. •Pa ial Leas Squa e (PLS). The PLS me hod [40] is a S uc u al Model Equa ion modeling echnique widely used in social sciences and business esea ch [41]–[44], and PLS eg ession is an ex ension o he mul iple linea eg ession models. In i s simples o m, a linea model speci ies he (linea ) ela ionship be ween a dependen a iable ( he use o he ool), and a se o p edic o a iables (ex e nal a iables p e iously ob ained). The objec i e in PLS is o maximize he explana ion a iance. Thus, and he signi icance o he ela ionships among a iables o dimensions a e measu es ha indica e how well a model is pe o ming. The concep ual co e o PLS is an i e a i e combina ion o P incipal Componen Analysis ela ing i ems o dimensions, and pa h analysis pe mi ing he cons uc ion o a causal model. The hypo hesizing o ela ionships be ween dimensions and i ems, and among di e en dimensions is guided by he p e ious li e a u e in his ield. The es ima ion o he pa ame e s ep esen ing he measu emen and pa h ela ionships is accomplished using O dina y Leas Squa es (OLS) echniques. •Co a iance s uc u e analysis as implemen ed in he Linea S uc u al Rela ions (LISREL) model. LISREL es ima es model pa ame e s in an a emp o ep oduce he co a i- ance ma ix o he measu es (o obse able a iables), and also inco po a es o e all goodness-o - i measu es o e alua e how well he hypo hesized model “ i s” he da a. Co a iance s uc u e analysis is “ heo y-o ien ed, and emphasizes he ansi ion om explo a o y o con i - ma o y analysis”[45]. The basic dis inc ion be ween PLS and LISREL as causal modeling me hodologies es s in hei objec i es. LISREL is bes used o heo y es ing and de elopmen [45]; while PLS is o ien ed owa ds p edic i e applica ions [46]. In compa ison wi h LISREL, he objec i e o PLS is he explana ion o a i- ance in a eg ession sense, and hus and he signi icance o ela ionships among dimensions a e measu es mo e indica i e o how well a model is pe o ming. “PLS is p ima ily in ended o causal-p edic i e analysis in si ua ions o high complexi y, bu low heo e ical in o ma ion”[45]. Fo hese easons, model es ing was examined h ough a PLS amewo k in he p esen case. Following he wo-s ep analy ical p ocedu e [47], he mea- su emen model is i s examined, and hen he s uc u al model. The a ionale o his wo-s ep app oach is o ensu e ha he conclusion on s uc u al ela ionship is d awn om a se o measu emen ins umen s wi h desi able psychome ic Au ho ized licensed use limi ed o: IEEE Xplo e. Downloaded on Feb ua y 17, 2009 a 04:29 om IEEE Xplo e. Res ic ions apply. TORAL e al.: MODELING LEARNER SATISFACTION 195 TABLE III PLS RESULTS p ope ies. The measu emen model is e alua ed in e ms o eliabili y, in e nal consis ency, con e gen alidi y, and dis- c iminan alidi y. Table III summa izes he ac o loadings and a e age a iance ex ac ed om he measu es o he p oposed esea ch model. •Indi idual i em eliabili y. In gene al, one would like o ha e each indica o sha ing mo e a iance wi h he com- ponen sco e han wi h he e o a iance. This condi ion implies ha he squa e o loadings should be g ea e han 0.70. Loadings o 0.5 and 0.6 a e accep able i he e a e addi ional indica o s in he block o compa ison basis [40]. This condi ion was me in his s udy, as shown in Table III. •Con e gen alidi y indica es he ex en o which he i ems o a scale ha a e heo e ically ela ed should ha e a high co ela ion. Con e gen alidi y was e alua ed o he measu emen scales using wo c i e ia sugges ed by [42]: (1) all indica o ac o loadings should be signi ican and exceed 0.70 and (2) A e age Va iance Ex ac ed (AVE) o each dimension should exceed he a iance due o measu emen e o o ha dimension (i.e., should exceed 0.50). All he measu es mee he ecommended le els (see Table III). •Disc iminan alidi y is he ex en o which he measu e is no a e lec ion o ano he a iable. Disc iminan alidi y is indica ed by low co ela ions be ween he measu e o in e es and he measu es o o he dimensions. E idence o disc iminan alidi y o he measu es can be e i ied using he squa ed oo o he A e age Va iance Ex ac ed o each dimension highe han he co ela ions be ween i and all o he dimensions [42]. As summa ized in Table III, he squa e oo o A e age Va iance Ex ac ed o each dimension (on he diagonal) is g ea e han he co ela- ions be ween he dimensions and all o he dimensions. The esul s sugges an adequa e disc iminan alidi y o he measu emen s. Nex , he s uc u al model is examined. The esea ch model was es ed using PLS-G aph .3.0 [48]. The model was es i- ma ed using he maximum likelihood me hod. Fig. 2 depic s i s a is ics, o e all explana o y powe , and es ima ed pa h coe - icien s. To assess he s a is ical signi icance o he pa h coe - icien s, which a e s anda dized be as, a boo s ap analysis was pe o med. Boo s apping p o ides an es ima e o he a iabili y o he pa ame e s in a inal model. The use o boo s apping, as opposed o adi ional - es s, allows he es ing o he signi i- cance o pa ame e es ima es om da a which a e no assumed Au ho ized licensed use limi ed o: IEEE Xplo e. Downloaded on Feb ua y 17, 2009 a 04:29 om IEEE Xplo e. Res ic ions apply. 196 IEEE TRANSACTIONS ON EDUCATION, VOL. 52, NO. 1, FEBRUARY 2009 Fig. 2. PLS esul : de ailed model. TABLE IV PATH COEFFICIENTS TABLE (T-STATISTIC) p < 0 : 05 ; p < 0 : 01 ; p < 0 : 001 , (0 : 05;499) = 1 : 9672 ; (0 : 01;499) = 2 : 5857 ; (0 : 001;499) = 3 : 3101 o be mul i a ia e no mal. Subsamples a e au oma ically gene - a ed om he exis ing da a by emo ing cases om he o al da a se gene a ed by he 284 s uden s. The numbe o andom sub- samples o be gene a ed is se by he analys . Fo his s udy, he numbe o boo s ap subsamples was se a 500. PLS es ima es he pa ame e s o each sub sample and “pseudo alues”a e cal- cula ed by applying he boo s ap o mula. Table IV shows ha mos o he pa hs p o ed o be signi ican a he p- alue 0.001 le el. All he hypo heses abou ela ionships among dimensions we e suppo ed. Finally, he esul s om he analysis show he explana o y powe o he esea ch model, e ealing ha he p oposed model sa is ac o ily accoun ed o 70.6% o he a iance. In summa y, he necessa y s eps o alida e a s uc u al and measu emen model a e nex de ailed. 1) De ine a model o be alida ed. The links o he model should be suppo ed by p e ious s udies o wo ks. 2) Design a ques ionnai e o measu e each o he dimensions o he model. Each dimension mus be measu ed by se e al indica o s ha can be ob ained om a su ey o s uden s. The eliabili y o he ques ionnai e is hen checked using a C onbach’s alpha index, emo ing hose indica o s wi h a low in e i em co ela ion alue. 3) Analyze he s uc u al and measu emen model using SEM. A s uc u al equa ion modeling, such as PLS, al- lows he alida ion o he indica o s o each dimension (measu emen model) as well as he hypo hesized links (s uc u al model). a) The measu emen is es ed using se e al c i e ia like indi idual i em eliabili y, con e gen alidi y and disc iminan alidi y. b) The s uc u al model is es ed using a boo s ap analysis. IV. RESULTS AND IMPLICATIONS Acco ding o Fig. 2, he esul s om he analysis gua an ees ha he p oposed model accoun ed o 70.6% o he a iance in sa is ac ion. Consequen ly, 70.6% o he a iance in s uden s’ Au ho ized licensed use limi ed o: IEEE Xplo e. Downloaded on Feb ua y 17, 2009 a 04:29 om IEEE Xplo e. Res ic ions apply. TORAL e al.: MODELING LEARNER SATISFACTION 197 TABLE V NEW TEACHING ACTIVITIES AND LEARNING ENVIRONMENTS TO IMPROVE STUDENTS’SATISFACTION Fig. 3. Imp o emen s on he cu en cou se o ganiza ion acco ding o lea ne sa is ac ion measu es. sa is ac ion can be explained, which is an excellen esul o his kind o analysis. The ob ained esul co obo a es he p elimina y s udy p e- sen ed in [25]. Topics ela ed o con ollable dimensions like Au ho ized licensed use limi ed o: IEEE Xplo e. Downloaded on Feb ua y 17, 2009 a 04:29 om IEEE Xplo e. Res ic ions apply. 198 IEEE TRANSACTIONS ON EDUCATION, VOL. 52, NO. 1, FEBRUARY 2009 con en , use in e ace, ease o use, and en husiasm and mo i a- ion mus be p omo ed o imp o e s uden s’sa is ac ion. New ac i i ies and lea ning en i onmen s can be p oposed o imp o e he con en , use in e ace, ease o use, and en husiasm and mo- i a ion dimensions. In he case s udy, he Elec onic Ins umen a ion and Mea- su emen Lab cou se, he con en dimension is imp o ed by planning new ac i i ies in he lab, based on indus ial buses and p o ocols, domo ic sys ems, emo e and i ual ins umen a ion con ol and mobile echnologies. The use in e ace dimension is imp o ed h ough he applica ion o a lea ning managemen sys em while he ease o use dimension is enhanced using mul- imedia echnologies and con en s o imp o e he lea ning p o- cesses. The en husiasm and mo i a ion dimension is enhanced using eal wo ld applica ions, wi h p ac ical esul s o he de- eloped wo k. The inclusion o collabo a i e and coope a i e hands-on wo k imp o es compe ences like eamwo k and col- labo a ion skills. Finally, he use con ol and in e ac i i y di- mension is imp o ed h ough s uden s’decision-making du ing he class, and he eedback dimension is p omo ed by doubling he numbe o p o esso s a ending each class. These imp o e- men s a e de ailed in Table V and Fig. 3. V. CONCLUSION Lea ne sa is ac ion has been modeled in an Elec onic In- s umen a ion and Measu emen Lab using S uc u al Equa ion Models. The adop ed app oach is based on he lea ne ’s sa is- ac ion, acco ding o he cu en Eu opean highe educa ion e- o ms, whe e he ocus o a en ion is mo ing om he eache o he lea ne . S uden s’sa is ac ion is posi i ely impac ed when he con en is ansmi ed h ough an adequa e use in e ace, when he in e ac ion wi h ins umen a ion equipmen and ools is easy and adequa e o he s uden s’le el, and when hey eel mo i a ed by he wo k hey a e equi ed o do. The model dis in- guishes i e pu e independen a iables: con en and eedback, ha should mainly be managed by he lec u e , and he h ee el- emen s o he s uden s’p o ile (Lea ning Communi y, Lea ne Responsibili y and P e ious Expe ience dimensions), which de- pend on he a i udes and capabili ies o s uden s who a end he cou se. The managemen o all o hese a iables is he espon- sibili y o he lec u e , who mus also conside a i udes and ca- pabili ies o p omo e use con ol and in e ac i i y. In acco dance wi h he p oposed model, sa is ac ion is he esul o wo ypes o in luences: he cogni i e and he a ec- i e in luence. Cogni i e in luence is essen ially d i en by he uppe -middle pa o he model, ha is, he con en o he cou se and he way i is deli e ed o s uden s, plus he possibili ies o in- e ac ion wi h equipmen and ools. The lowe -middle pa o he model ep esen s he a ec i e in luence o e sa is ac ion, ha is, s uden s’mo i a ion when wo king in he lab. Use con ol and in e ac i i y has also an a ec i e componen in he sense ha i can p omo e en husiasm and mo i a ion. In acco dance wi h his analysis, se e al imp o emen s e- ga ding he con en , use in e ace, ease o use, and mo i a ion should be conside ed. Al hough some o he dimensions could be conside ed, he numbe o hese should be kep low o a oid model o e i . The p oposed me hodology can be gene alized o any subjec in a highe educa ion cou se, and demons a es i s use ulness in highligh ing which ele an aspec s should be im- p o ed om a lea ne -cen e ed app oach. REFERENCES [1] M. Al assi, “E ec s o a lea ne -cen ed en i onmen on he academic compe ence and mo i a ion o s uden s a isk,”Lea n. En i on. Res., ol. 7, pp. 1–22, 2004. [2] A. Jones and V. Comp on, “Towa ds a model o eache de elopmen in echnology educa ion: F om esea ch o p ac ice,”In . J. Technol. Design Educ., ol. 8, pp. 51–65, 1998. [3] Join Decla a ion o he Eu opean Minis e s o Educa ion, The Eu o- pean Highe Educa ion A ea—Bologna Decla a ion. Bologna, Jun. 19, 1999. [4] Uni e si y o Deus o and Uni e si y o G oningen. Tuning Educa- ional S uc u es in Eu ope, Final Repo Phase One, J. González and R. Wagenaa , Eds., 2003. [5] S. L. To al, M. R. Ma ínez-To es, F. Ba e o, S. Galla do, E. Va gas, and V. Gónzalez, “Planning a mas e ’s le el cu iculum acco ding o ca ee space ecommenda ions using concep mapping echniques,”In . J. Technol. Design Educ., ol. 16, no. 3, pp. 237–252, 2006. [6] E. D. Wagne and B. L. Mccombs, “Lea ne cen e ed psychological p inciples in p ac ice: Designs o dis ance educa ion,”Educ. Technol., ol. 35, no. 2, pp. 32–35, 1995. [7] M. Du an, S. Galla do, S. To al, M. R. M. To es, and F. Ba e o, “A lea ning me hodology using Ma lab/Simulink o unde g adua ed elec- ical enginee ing cou ses a ending o lea ne sa is ac ion ou comes,” In . J. Technol. Design Educ., ol. 17, no. 1, pp. 55–73, 2007. [8] S. Hadje oui , “Lea ne -cen ed web-based ins uc ion in so wa e en- ginee ing,”IEEE T ans. Educ., ol. 48, pp. 99–104, 2005. [9] J. Milliken and L. P. Ba nes, “Teaching and echnology in highe ed- uca ion: S uden pe cep ions and pe sonal e lec ions,”Comp. Educ., ol. 39, pp. 223–235, 2002. [10] J. González and R. Wagenaa , Tuning Educa ional S uc u es in Eu ope II. Uni e si ies’Con ibu ion o he Bologna P ocess Uni . Deus o and Uni . G oningen, 2005. [11] M. R. Ma ínez-To es, F. Ba e o, S. L. To al, and S. Galla do, “A digi al signal p ocessing eaching me hodology using concep mapping echniques,”IEEE T ans. Educ., ol. 48, no. 3, pp. 422–429, 2005. [12] S. L. To al, F. Ba e o, M. R. Ma ínez-To es, S. Galla do, and J. Lillo, “Implemen a ion o a web-based educa ional ool o digi al signal p ocessing eaching using he echnological accep ance model,” IEEE T ans. Educ., ol. 48, no. 4, pp. 632–641, 2005. [13] N. M. Lambe and B. L. McCombs, How S uden s Lea n: Re o ming Schools Th ough Lea ne Cen ed Educa ion. Wash., DC: Ame . Psycholog. Assoc., 1997. [14] J. M. Kelle , “Ins uc ional-design heo ies and models: An o e iew o hei cu en s a us,”in Mo i a ional Design o Ins uc ion. Hillsdale, NJ: Law ence E lbaum , 1983, pp. 386–434. [15] E. L. Deci and R. M. Ryan, Handbook o Sel -De e mina ion Re- sea ch. Camb idge, U.K.: Boydell and B ewe , 2002. [16] S. D. Johnson, S. R. A agon, N. Shaik, and N. Palma-Ri as, “Compa - a i e analysis o lea ne sa is ac ion and lea ning ou comes in online and ace- o- ace lea ning en i onmen s,”J. In e ac . Lea n. Res., ol. 11, no. 1, pp. 29–49, 2000. [17] I. M. Abdel-Qade , B. J. Bazuin, H. S. Mousa inezhad, and J. K. Pa ick, “Real- ime digi al signal p ocessing in he unde g adua e cu iculum,”IEEE T ans. Educ., ol. 46, no. 1, pp. 95–101, 2003. [18] H. Mousa inezhad and I. Abdel-Qade , “Digi al signal p ocessing in heo y and p ac ice,”in P oc. 2001 F on ie s In Educ. Con ., Reno, NV, Oc . 2001. [19] Y. S. Wang, “Assessmen o lea ne sa is ac ion wi h asynch onous elec onic lea ning sys ems,”In . Manage., ol. 41, pp. 75–86, 2003. [20] W. J. Doll and T. S. Raghuna han, “A con i ma o y ac o analysis o he use in o ma ion sa is ac ion ins umen ,”In . Sys . Res., ol. 6, no. 2, pp. 177–188, 2001. [21] R. McHaney, R. High owe , and J. Pea son, “A alida ion o he end- use compu ing sa is ac ion ins umen in Taiwan,”In . Manage., ol. 39, pp. 503–511, 2002. [22] Commision o he Eu opean Communi ies, No . 2005, P oposal o a Recommenda ion o he Eu opean Pa liamen and o he Council on Key Compe ences o Li elong Lea ning [Online]. A ailable: h p://ec. eu opa.eu/educa ion/policies/2010/doc/key ec_en.pd [23] J. B. A baugh, “How ins uc o immediacy beha iou s a ec s uden sa is ac ion and lea ning in web-based cou ses,”Business Communi- ca ion Qua e ly, ol. 64, no. 4, pp. 42–54, 2001. Au ho ized licensed use limi ed o: IEEE Xplo e. Downloaded on Feb ua y 17, 2009 a 04:29 om IEEE Xplo e. Res ic ions apply.