Resea ch Pape
Recommended ci a ion: Al izqi, M., & Godwin, A. (2025). Modeling Feedback
Dynamics in Lea ning S udios: A Causal-Loop Analysis o an Unde g adua e
Mechanical Enginee ing Cu iculum. In Kangaslampi, R., Langie, G., Jä inen,
H.-M., & Nagy, B. (Eds.), SEFI 53 d Annual Con e ence. Eu opean Socie y o
Enginee ing Educa ion (SEFI), Tampe e, Finland. DOI:
10.5281/zenodo.17631226.
This Con e ence Pape is b ough o you o open access by he 53 d Annual Con e ence
o he Eu opean Socie y o Enginee ing Educa ion (SEFI) a Tampe e Uni e si y in
Tampe e, Finland. This wo k is licensed unde a C ea i e Commons
A ibu ion-NonComme cial-Sha e Alike 4.0 In e na ional License.
Modeling Feedback Dynamics in Lea ning S udios: A Causal-Loop Analysis o
an Unde g adua e Mechanical Enginee ing Cu iculum
Mohammed A. Al izqi a,1, Allison Godwin b
a Sibley School o Mechanical and Ae ospace Enginee ing, Co nell Uni e si y,
I haca, Uni ed S a es, 0000-0001-6034-8314
b Robe F ede ick Smi h School o Chemical and Biomolecula Enginee ing, Co nell
Uni e si y, I haca, Uni ed S a es, 0000-0002-0741-3356
Keywo ds: Sys ems Thinking, Causal Loop Diag ams, Expe ien ial Lea ning,
Feedback Loops, Collabo a i e Lea ning
ABSTRACT
This s udy adop s a sys ems pe spec i e o examine how a newly in oduced
“Lea ning S udios” (LS) model in a mechanical and ae ospace enginee ing (MAE)
school in luences unde g adua e enginee ing expe iences. LS a e c oss-cou se labs
ha engage mul iple concep s and lea ning objec i es, embedded h oughou all
h ee yea s o he cu iculum. We conduc ed semi-s uc u ed in e iews wi h eigh
mechanical enginee ing s uden s a a U.S. R1 ins i u ion o cap u e hei pe cep ions
o hands-on, eal-wo ld enginee ing asks. D awing on he Expec ancy-Value-Cos
amewo k, we de eloped ou quali a i e s udy and hen employed induc i e
hema ic analysis o iden i y key ac o s, including mo i a ion, pe sonal g ow h,
belonging, and enginee ing iden i y. Subsequen ly, we de eloped causal loop
diag ams (CLDs) o isualize how hese ac o s in e ac and shape eedback loops in
s uden s’ academic and social en i onmen s.
Findings indica e ha LS expe iences os e ein o cing cycles: mas e y o p ac ical
asks boos s sel -con idence, which, in u n, enhances engagemen and s eng hens
pee ne wo ks. Academic and social in eg a ion also os e a sense o belonging ha
u he sus ains mo i a ion and educes he isk o a i ion. Howe e , he diag ams
e eal balancing loops when academic challenges s ain s uden s’ esilience.
Mapping hese dynamics unde sco es he alue o a holis ic sys ems app oach in
enginee ing educa ion. By illumina ing bo h posi i e eedback loops and po en ial
bo lenecks, ou s udy p o ides ac ionable insigh s o designing in e en ions ha
suppo s uden s’ holis ic de elopmen , om echnical compe ence o he o ma ion
o a p o essional iden i y.
1 Co esponding Au ho
M. A. Al izqi
Ma867@co nell.edu
1 INTRODUCTION
1.1 Backg ound
Enginee ing educa ion ope a es wi hin a dynamic sys em whe e mul iple
in e connec ed ac o s shape s uden success. A sys em is de ined as “a g oup o
in e ac ing, in e ela ed, o in e dependen componen s ha o m a complex and
uni ied whole."(Ande son & Johnson, 1997, p. 2). Conside ing hese cha ac e is ics,
enginee ing educa ion can be iewed as a complex sys em (Mona e al., 2022).
Despi e his unde s anding, adi ional esea ch has o en app oached enginee ing
educa ion (EE) in isola ion (Sigahi & Sznelwa , 2022). A sys ems app oach is
essen ial as i allows o a mo e holis ic and in eg a ed unde s anding o complexi ies
and in e connec ed le els (Wal e s & Li ch ield, 2015).
As a esul , he e has been inc easing in e es in he pas se e al yea s o a
sys ems app oach o ackle he challenges o complex sys ems in enginee ing
educa ion esea ch (EER). Fo example, Sigahi e al. (2022) explained how
complexi y heo y and sys ems hinking can ans o m EER by p o iding amewo ks
o add ess in e disciplina y challenges and in e dependen ac o s in enginee ing
cu icula (Sigahi & Sznelwa , 2022). Dhuka am e al., (2016) a gued ha highe
educa ion in e ac s wi h a complex ecosys em o poli ical, economic, echnological,
and o ganiza ional ac o s, necessi a ing sys ems hinking o modeling and policy
analysis.
This unde s anding is key in educa ion. A nold (2018) a gued ha academic and
social ac o s in college expe iences a e closely linked and should be add essed
oge he o e ec i e enhancemen . Failing o add ess eedback on hese aspec s
can lead o ine ec i e in e en ions. Fu he mo e, when uni e si ies implemen
changes like new cu icula wi hou conside ing such sys em dynamics, hey may
encoun e limi ed success.
To add ess hese challenges, a ew esea che s ha e applied a sys ems dynamics
app oach—speci ically, causal loop diag ams (CLDs)—in EER. CLDs, which unc ion
simila ly o concep maps, isually depic how a change in one ac o , like mo i a ion,
can in luence ano he , like in e es in a majo (S e man, 2000). Once a de ailed
causal loop diag am is c ea ed, sys emic issues can be modeled and s a egies o
change iden i ied (Vanasupa e al., 2008). CLDs allow esea che s o isualize key
a iables and hei ela ionships by linking hem wi h a ows ha indica e causali y
and designa ing loops as ei he ein o cing o balancing (Zhang e al., 2012). This
isualiza ion helps unco e hidden eedback s uc u es, o e ing a mo e holis ic
pe spec i e on complex educa ional challenges (Dhuka am e al., 2016).
Fo example, a s uden ’s iden i y can a ec hei mo i a ion, which in u n in luences
hei engagemen and pe o mance, and his eedback can impac hei sense o
belonging (Godwin e al., 2016; Godwin & Ki n, 2020). As discussed by se e al
esea che s, CLDs acili a e he isualiza ion o how eedback loops can gene a e
complex beha io s in social sys ems (Ande son & Johnson, 1997; Wal e s &
Li ch ield, 2015). These beha io s include sel - ein o cing imp o emen s in i uous
cycles (e.g., pee collabo a ion ➔ imp o ed p oblem-sol ing ➔ s onge
collabo a ion) o balancing mechanisms (e.g., academic p essu e ➔ s ess ➔
educed pe o mance).
Fo educa ion s akeholde s, mapping hese ela ionships be ween ac o s can
highligh new a eas o esea ch and complica e he cu en EER landscape, he eby
acili a ing e ec i e change.
Al hough s ill eme ging, CLDs and sys em mapping a e gaining ac ion in EER. Fo
example, one s udy demons a ed how CLDs cap u e in e dependencies in social-
ecological sys ems, ein o cing hei alue in modeling educa ion dynamics (G ay e
al., 2019). Ano he s udy applied CLDs o acul y mo i a ion, inding 13 eedback
loops ha a ec eaching inno a ion and no ing ha many change ini ia i es ail due
o igno ed sys em-wide in e connec ions (C uz-Boho quez e al., 2024). These
s udies illus a e how CLDs o e a holis ic lens on educa ional complexi y, guiding
mo e e ec i e in e en ions.
Despi e hese ad ancemen s, a signi ican gap emains in he holis ic modeling o
he unde g adua es' expe ience using a sys ems app oach. In his s udy, we b idge
he gap by e alua ing enginee ing s uden s' expe iences in a newly designed lab,
LS, h ough a dynamic sys em app oach, CLDs. Ou aim in his pape is o discuss
how we used CLDs o explo e s uden expe iences and o e his app oach as a
po en ial way o conside complex, in e ela ed da a in enginee ing educa ion.
1.2 Lea ning S udios Model
In he Fall o 2022, he mechanical and ae ospace enginee ing (MAE) depa men a
a U.S. No heas e n R1 Uni e si y in oduced Lea ning S udios (LS) o enhance
hands-on, eal-wo ld lea ning ac oss mul iple cou ses. Unlike adi ional p ojec -
based lea ning, LS in eg a es ully ope a ional enginee ing sys ems (e.g.,
combus ion engines, d ones, o kli s) in o he co e cu iculum, ensu ing all s uden s
gain p og essi e, hands-on expe ience a he han limi ing i o elec i es. These LS
a e sp ead ac oss he en i e h ee yea s o he MAE cu iculum and in eg a e c oss-
cu ing concep s ac oss cou ses in hese applica ions. This app oach is unique—
mos p og ams ha in eg a e hands-on labs do so wi h a single cou se o sys em.
The LS app oach b idges heo y and p ac ice using ou key componen s: eal-wo ld
enginee ing sys ems, ad anced analysis ools, simpli ied models, and s uc u ed
lea ning modules.
1.3 Objec i e and Resea ch Ques ions
In his pape , we use a pilo o s uden in e iew da a ansla ed in o CLDs o cap u e
he mul i ace ed impac o LS on s uden s’ expe iences. This wo k can enhance ou
unde s anding o how and why in e en ions like LS a e bene icial, p o iding a
sys em-le el s udy ha complemen s adi ional educa ional assessmen s.
Speci ically, we ask wo ques ions:
1. Wha a e he causal links be ween s uden engagemen wi h LS, shi s in sel -
pe cep ion, and communi y eelings, and how do hese a ec hei lea ning
jou ney?
2. Wha a e he main d i ing ac o s o he main hemes eme ging om he
s udy?
2 METHODOLOGY
Ou esea ch employs an in e p e i e, quali a i e esea ch design combined wi h a
sys ems- hinking analy ical app oach.
2.1 Da a Sou ce
The in e iew p o ocol u ilized he Expec ancy-Value-Cos (EVC) model (Al izqi e
al., 2025; Ba on & Hulleman, 2015). This amewo k sugges s ha s uden s' choices
in achie emen a e shaped by hei expec a ions o success and he pe sonal alue
hey associa e wi h asks, encompassing in insic, a ainmen , and u ili y alues, as
well as associa ed cos s.
The semi-s uc u ed in e iew p o ocol was c a ed in adhe ence o ecognized
quali a i e esea ch me hodologies (Al izqi e al., 2025; Bo ego e al., 2009;
C idland e al., 2015; Kallio e al., 2016). Such in e iews o e bene i s by ha ing a
balance be ween s uc u ed ques ions and he lexibili y o di e in o newly eme ging
opics (Kallio e al., 2016). This app oach also mi o s he ecommenda ions o a
p e ious s udy, which ound ha in e iews a e an e ec i e way o elici ich
quali a i e da a o sys ems dynamics models (Luna-Reyes & Ande sen, 2003). Also,
his was employed in simila sys em dynamics s udies (e.g., El Halabi e al., 2012) o
ga he insigh s on key a iables and ela ionships.
2.2 Pa icipan Selec ion and Sampling S a egy
A pu posi e sampling me hod was used o selec pa icipan s wi h i s hand
expe ience in LS. Selec ion c i e ia included en olmen in he MAE deg ee p og am,
engagemen in a leas one LS, and s a i ica ion by class s anding, gende , and ace
o ensu e di e se pe spec i es. Eigh mechanical enginee ing s uden s me hese
c i e ia and olun a ily pa icipa ed; his sample size is app op ia e o a quali a i e
p elimina y explo a ion. Appendix 1 p esen s he demog aphics o he pa icipan s.
Pa icipan s chose hei pseudonyms; howe e , i hey did no p o ide one, he
esea che assigned one o hem.
2.3 Da a Collec ion and Analysis
We pilo ed he in e iew p o ocol wi h one pa icipan o ensu e da a quali y and
make e inemen s based on eedback. Each semi-s uc u ed in e iew was hen
conduc ed and las ed app oxima ely 45 minu es. Each in e iew ook place in a
p i a e se ing o ensu e con iden iali y and was audio- eco ded wi h he pa icipan 's
consen . The Ins i u ional Re iew Boa d (IRB) app o ed he s udy, ensu ing
compliance wi h e hical esea ch s anda ds (Kallio e al., 2016).
Fu he mo e, an induc i e hema ic analysis was conduc ed ollowing B aun and
Cla ke's B aun & Cla ke (2006) six-phase amewo k. This app oach was chosen
because i allows hemes o eme ge di ec ly om he da a, which is essen ial when
explo ing unde - esea ched a eas (Pa on, 2014; Saldaña, 2013). These six s eps
a e (1) Familia iza ion wi h he ansc ip s, (2) gene a ing ini ial codes (using
MAXQDA so wa e), (3) sea ching o po en ial hemes, (4) e ining hemes o ensu e
accu acy and o ensu e code consis ency, (5) de ining and naming hemes, and (6)
o ganizing in o a cohe en na a i e.
2.4 Causal Loop Diag am De elopmen
A e we coded he da a, we buil he CLDs. This s ep in ol ed ans o ming he
quali a i e da a in o a isual sys ems map illus a ing eedback s uc u es.
2.4.1 Iden i ying Va iables
Fi s , we iden i ied he main a iables ha would appea as nodes in he CLDs.
These co esponded closely o he hemes, sub hemes, and ac o s iden i ied
h ough he hema ic analysis. Fo example:
• A heme like “Belonging and Suppo i e Communi y” is ep esen ed as a
a iable “Sense o Belonging.”
• A heme like “Enhanced Mo i a ion and Con idence” was spli in o wo ela ed
a iables: “Mo i a ion” and “Sel -Con idence,” since hose appea ed as
dis inc concep s in he da a bu we e pa o one heme.
2.4.2 Mapping Causal Links
Nex , we e isi ed he in e iew da a o map ou causal ela ionships be ween hese
a iables. We ollowed he app oach sugges ed by Newbe y and Ca ha (2023),
which in ol ed a “da a sou ce e e ence able" o aceabili y (Newbe y & Ca ha ,
2023, p. 12). Fo ins ance, Sa ah said, “we would jus alk abou … cou ses we wan
o ake in he u u e, o … jus gi e ad ice o each o he . This made me eel mo e
belong[ing] o mechanical.” F om such a s a emen , we can de i e a causal link:
Pee Suppo → Sense o Belonging→ Mechanical Enginee Iden i y. F om he
li e a u e on enginee ing iden i y, we connec ed Mechanical Enginee ing Iden i y →
Re en ion (Godwin & Ki n, 2020).
We documen ed each such link and no ed i s pola i y (posi i e o nega i e in luence).
A posi i e link (“+”) means ha as one a iable inc eases/dec eases, he o he ends
o inc ease/dec ease. A nega i e link (“–”) means an in e se ela ionship whe e an
inc ease in one leads o a dec ease in he o he , and ice e sa. In he example
abo e, pee suppo had a posi i e impac on belonging, which in u n had a posi i e
e ec on iden i y and e en ion.
2.4.3 Cons uc ing he Diag am
Once we had a lis o a iables and hei pai wise causal links, we s a ed
cons uc ing CLDs using Vensim™. We i s clus e ed ela ed a iables, o en
co esponding o he hema ic g oups om he analysis. To enhance cla i y, we
segmen ed ou o e all model in o h ee sub-diag ams, simila o he me hod used by
Halabi e al. (2012) (El Halabi e al., 2012).
3 RESULTS
3.1 O e iew o Themes
Ou indings highligh h ee key hemes whe e LS posi i ely impac ed s uden s’
expe iences: (1) Mo i a ion, (2) Pe sonal G ow h, (3) Sense o Belonging, and (4)
Iden i y as Mechanical Enginee s. Pa icipan s epo ed pe sonal de elopmen
h ough hei hands-on expe iences. In e iews e ealed ha mas e ing cou ses
os e ed a sense o belonging in mechanical enginee ing, while s uggles led o
aliena ion. Mo eo e , s uden s honed hei skills and expe ienced pe sonal g ow h
and communi y by engaging wi h eal-wo ld applica ions. Findings show ha LS was
i al in building s uden con idence, helping hem o e come eelings o inadequacy,
and enhancing hei sa is ac ion wi h lea ning. Recognizing hese impac s is
essen ial o c ea ing e ec i e educa ional en i onmen s and p ac ices ha enhance
collabo a i e and hands-on lea ning, which a e c i ical o de eloping enginee ing
skills and os e ing p o essional g ow h.
3.2 Causal Loop Diag ams (CLDs)
In ou case, he h ee in e connec ed sub-CLDs we e: (1) Lea ning and Pe sonal
G ow h, (2) Social Dynamics and Communi y, and (3) Mo i a ion and Ca ee
Aspi a ions. Subse ing he models made i simple o discuss speci ic loops. Each
sub-CLD cap u ed a subse o a iables and hei co esponding eedback loops,
which pe ained o a speci ic ace o he s uden expe ience.
3.2.1 Expe ien ial Lea ning and Sel -Rein o cemen (Figu e 1)
This CLD illus a es a ein o cemen loop ha examines how hands-on expe ience
leads o a sel - ein o cing cycle. I demons a es ha p ac ical expe iences and a
sense o compe ence ha e a signi ican in luence on he lea ning p ocess, leading o
a deepe unde s anding and inc eased engagemen .
Fo example, Cha lle explained he expe ience by saying, “I was un o ea down an
engine and look a all he pa s. De ini ely helped me unde s and how engine wo ked
a lo be e ,” showing how di ec manipula ion o a physical sys em can b idge
abs ac heo y and p ac ical applica ion. She con inued, " Now I eel com o able
wo king in a lab…I can ackle any hing,” indica ing ha i e a i e successes bols e
sel -e icacy. Ano he s uden , Zahe , desc ibed how applied asks spa ked in insic
mo i a ion, Sel -De e mina ion Theo y’s key elemen (Deci & Ryan, 1985). He said,
“Compa ing heo y o eal O o cycle da a was ascina ing.” These insigh s show ha
LS encou aged epea ed hands-on successes ha cul i a e s uden success and
d i e u he engagemen wi h complex enginee ing challenges
The pa icipan s' desc ip ions illus a e Kolb’s (Kolb e al., 2001)Expe ien ial Lea ning
and Bandu a’s Sel -E icacy (Bandu a, 1978) heo ies. Speci ically, i aligns wi h
Kolb’s ou -s age lea ning cycle (conc e e expe ience, e lec i e obse a ion,
abs ac concep ualiza ion, and ac i e expe imen a ion) by demons a ing how
con inuous, hands-on p ac ice ein o ces lea ning h ough e lec ion and heo y-
building.
Fig 1. This ein o cing loop (R1) illus a es how hands-on asks con ibu e o deepe concep ual
unde s anding, which enhances con idence and engagemen . This, in u n, ul ima ely inc eases
mo i a ion and enjoymen in he lea ning p ocess (
R2), ein o cing s uden s' sense o compe ence.
3.2.2 Resilience Th ough Challenge-Suppo Dynamics (Figu e 2)
The second CLD examines how academic challenges in e ac wi h pee ela ionships
and ins i u ional suppo o os e esilience. I illus a es how pee suppo can help
indi iduals o e come challenges, he eby enhancing hei esilience and, in u n,
hei pe sonal g ow h. This aligns wi h Tin o (1975) emphasis on academic
in eg a ion—while no malizing s uggle in line wi h Dweck (2006) g ow h mindse ,
hus mo i a ing sus ained engagemen in igo ous enginee ing con ex s.
Fo example, s uden s’ s uggles—such as Zahe ’s commen , “S a ics made me
e hink my pa h”— e lec Tin o’s emphasis on academic in eg a ion as essen ial o
e en ion. Lea ning communi ies’ collabo a i e en i onmen s help coun e ac
possible disengagemen by os e ing social in eg a ion. As Zahe obse ed,
“E e yone chee s each o he on—i ’s no cu h oa ,” exempli ying how no malized
s uggle and pee suppo e ame challenges as lea ning oppo uni ies, aligning wi h
Dweck’s g ow h mindse amewo k. These indings highligh how well-s uc u ed lab
spaces such as LS can mi iga e he isk o a i ion by in eg a ing challenge and
suppo .
3.2.3 Collabo a i e Iden i y and Belonging (Figu e 3)
This CLD cen e s on how eamwo k in LS os e s a p o essional iden i y and sense
o belonging. I also emphasizes he social componen o he lea ning en i onmen ,
such as Collabo a i e p ojec s, in shaping he mechanical enginee ing iden i y.
G oup asks, such as oubleshoo ing luid sys ems, e lec legi ima e pe iphe al
pa icipa ion, whe e lea ne s gain compe ence h ough sha ed p ac ice.
Sa ah’s s a emen , “Helping o he s wi h hei p ojec s made me eel like pa o he
eam,” and Cha lle’s e lec ion, “Seeing pee s succeed made me p oud o be
aMechE,” bo h unde sco e how sha ed p ac ice and mu ual suppo os e
mechanical enginee ing iden i y. I also demons a es how ecogni ion by o he s
ein o ces an indi idual's enginee ing iden i y (Godwin e al., 2016).
Exclusion om eams, howe e , can in oduce a balancing loop, as no ed by Kwami:
“I was discou aged when I didn’ ge in o a p ojec eam.” This highligh s he
impo ance o os e ing ins i u ional belonging o p e en po en ial disengagemen .
Howe e , Kwami’s e en ual esilience, “Tha de ini ely ein o ced my posi i e a i ude
Fig 2. exhibi s how academic challenges in e ac wi h pee /ins uc o suppo in a balancing loop
(B) ha mi iga es educed belonging. A ein o cing loop (R1) shows how o e coming challenges
builds esilience, leading o pe sonal g ow h and s uden pe sis ence (R2), ei
n o cing engagemen
owa ds he mechanical enginee ing,” illus a es how si ua ed lea ning expe iences
p omo e he co-cons uc ion o enginee ing iden i y.
O e all, hese obse ed dynamics align wi h heo ies such as Communi ies o
P ac ice (La e & Wenge , 1991) Vygo sky’s Social Cons uc i ism (VYGOTSKY,
1978), and Social Iden i y Theo y (Taj el & Tu ne , 1979), which sugges s ha LS
se e as communi ies o p ac ice ha con inually ein o ce a sense o membe ship,
sha ed pu pose, and p o essional p ide.
4 DISCUSSION
Ac oss he h ee models, se e al sys emic insigh s become appa en . Two o hose
CLDs' loops a e ein o cing, indica ing ha LS can c ea e i uous cycles—posi i e
eedback loops in which each gain uels he nex . S uden s equen ly desc ibe
upwa d ends as s a ing ou unce ain, hen gaining skills, o ming iendships,
g owing in con idence, and ending highly mo i a ed. This esul does no mean he
p ocess is uni e sal o au oma ic; balancing loops (e.g., academic challenges) can
impede he posi i e cycles. Howe e , he ac ha we iden i ied common ein o cing
loops sugges s ha well-designed in e en ions can push a s uden ’s expe ience in o
a posi i e, sel -sus aining ajec o y—wha sys ems heo y calls an “a ac o s a e,” a
s able pa e n ha he sys em na u ally ends o main ain.
Ou indings align s ongly wi h es ablished educa ional heo ies and p io esea ch.
Fo ins ance, he impo ance o communi y and belonging in d i ing engagemen
aligns wi h Tin o’s model o s uden e en ion, as demons a ed in Model 2 (sec ion
3.2.2). Recen s udies (Smi h e al., 2021; Wal on & Cohen, 2011) Addi ionally,
in e en ions a ge ing a sense o belonging can enhance academic ou comes. Ou
quali a i e da a p o ide a causal na a i e o how hese in e en ions may wo k
(belonging → engagemen → mo i a ion → achie emen ). Simila ly, he ole o
hands-on lea ning in boos ing mo i a ion esona es wi h ac i e lea ning li e a u e
(F eeman e al., 2014). Ou CLDs ou line he mechanism; hands-on successes lead
o enjoymen and mo i a ion, which in u n lead o deepe engagemen . The social
dimension is also cap u ed, e ealing how pee suppo uels a sense o belonging,
which in u n ein o ces engagemen and enhances he abili y o ace academic
challenges.
A key me hodological ad an age o CLD-based quali a i e analysis is i s abili y o
make indi ec causal pa hways and empo al dynamics explici . In ou model, he
p ima y in e en ion—hands-on s udio expe ience—has a di ec , posi i e impac on
Fig 3.This ein o cing loop (R) shows how collabo a i e p ojec s enhance pee ecogni ion,
s eng hening enginee ing iden i y, belonging, and mo i a ion, which d i es u he engagemen .
