Valdés, Manuel T.; Solga, Heike
A icle — Published Ve sion
The STEM leaky pipeline a labo ma ke en y in Spain:
The ole o job compe i ion and social o igin
Social Science Resea ch
P o ided in Coope a ion wi h:
WZB Be lin Social Science Cen e
Sugges ed Ci a ion: Valdés, Manuel T.; Solga, Heike (2024) : The STEM leaky pipeline a labo
ma ke en y in Spain: The ole o job compe i ion and social o igin, Social Science Resea ch, ISSN
1096-0317, Else ie , Ams e dam, Vol. 124, pp. 1-17,
h ps://doi.o g/10.1016/j.ss esea ch.2024.103092
This Ve sion is a ailable a :
h ps://hdl.handle.ne /10419/310005
S anda d-Nu zungsbedingungen:
Die Dokumen e au EconS o dü en zu eigenen wissenscha lichen
Zwecken und zum P i a geb auch gespeiche und kopie we den.
Sie dü en die Dokumen e nich ü ö en liche ode komme zielle
Zwecke e iel äl igen, ö en lich auss ellen, ö en lich zugänglich
machen, e eiben ode ande wei ig nu zen.
So e n die Ve asse die Dokumen e un e Open-Con en -Lizenzen
(insbesonde e CC-Lizenzen) zu Ve ügung ges ell haben soll en,
gel en abweichend on diesen Nu zungsbedingungen die in de do
genann en Lizenz gewäh en Nu zungs ech e.
Te ms o use:
Documen s in EconS o may be sa ed and copied o you pe sonal
and schola ly pu poses.
You a e no o copy documen s o public o comme cial pu poses, o
exhibi he documen s publicly, o make hem publicly a ailable on he
in e ne , o o dis ibu e o o he wise use he documen s in public.
I he documen s ha e been made a ailable unde an Open Con en
Licence (especially C ea i e Commons Licences), you may exe cise
u he usage igh s as speci ied in he indica ed licence.
h p://c ea i ecommons.o g/licenses/by-nc/4.0/
The STEM leaky pipeline a labo ma ke en y in Spain: The ole
o job compe i ion and social o igin
Manuel T. Vald´
es
a,*
, Heike Solga
b
a
Uni e sidad Nacional de Educaci´
on a Dis ancia (UNED), Calle del Obispo T ejo 2, 28040, Mad id, Spain
b
WZB – Be lin Social Science Cen e and F eie Uni e si ¨
a Be lin, Reichpie schu e 50, 10785, Be lin, Ge many
ARTICLE INFO
Keywo ds:
STEM
Gende gap
Leaky pipeline
Uni e si y g adua es
School- o-wo k ansi ion
ABSTRACT
The unde ep esen a ion o women in STEM (Science, Technology, Enginee ing, and Ma he-
ma ics) majo s is well documen ed. Using high-quali y Spanish da a, his s udy examines whe he
emale STEM g adua es a e less likely o pu sue STEM ca ee s han hei male coun e pa s and
conside s he mode a ing ole o labo ma ke condi ions and social o igin. We ind a p onounced
gende e ec in ini ial and subsequen job placemen (4–5 yea s a e g adua ion). No ably, e-
male STEM g adua es a e less likely o wo k in STEM occupa ions, e en i hey s a ed hei
ca ee s in STEM. Exploi ing he signi ican impac o he G ea Recession on he Spanish labo
ma ke , ou s udy e eals a signi ican ly la ge gende e ec among indi iduals who g adua ed
du ing he c isis compa ed o hose who g adua ed du ing he subsequen economic eco e y.
Thus, job compe i ion in luences he magni ude o he gende e ec . Finally, ou in e sec ional
analysis o gende and social o igin sugges s ha he gende di e ence is la ge among STEM
g adua es om low-SES backg ounds.
1. In oduc ion
The adi ional male ad an age in educa ion has been e e sed in ecen decades, wi h women being now mo e likely han men o
pu sue highe educa ion (DiP e e and Buchmann, 2013; Es e e e al., 2016). Howe e , men a e s ill much mo e likely o choose STEM
(Science, Technology, Enginee ing, and Ma hema ics) p og ams (Che yan e al., 2017; Mann and DiP e e, 2013; Mo gan e al., 2013),
which a e mo e p es igious and, despi e conside able in e nal he e ogenei y, be e paid.
1
The p ocess leading o women’s unde ep esen a ion in STEM ields is commonly desc ibed as a leaky pipeline, wi h gi ls and
young women leaking ou o he pa hway o a STEM deg ee a each s age o he p ocess (Be yman, 1983; Lee e al., 2015; Mo gan
e al., 2013; Raabe e al., 2019; Spee , 2023; an de Vleu en e al., 2018): hey a e less likely o engage in science- ela ed ac i i ies
du ing childhood and adolescence (Legewie and DiP e e, 2014), o o m expec a ions abou a ca ee in science (Alm, 2015), o choose
* Co esponding au ho .
E-mail add esses: [email p o ec ed] (M.T. Vald´
es), [email p o ec ed] (H. Solga).
1
STEM- ela ed occupa ions ypically sco e highe on he ISEI scale han non-STEM occupa ions wi h equi alen educa ional equi emen s,
indica ing hei highe socioeconomic s a us (Ganzeboom e al., 1992). In Spain, 84 pe cen o g adua es in “Enginee ing and A chi ec u e” a e
employed ou yea s a e g adua ion, wi h an a e age annual sala y o 31,751
€
; hese igu es d op o 76 pe cen and 27,942
€
o g adua es in
“Social Sciences and Law”, and u he down o 66 pe cen and 26,723
€
o g adua es in “A s and Humani ies” (in o ma ion o 2022, e ie ed
om he Spanish Minis y o Uni e si ies: h ps://www.uni e sidades.gob.es/indicado es-de-a iliacion-a-la-segu idad-social-de-los-eg esados-
uni e si a ios/).
Con en s lis s a ailable a ScienceDi ec
Social Science Resea ch
jou nal homepage: www.else ie .com/loca e/ss esea ch
h ps://doi.o g/10.1016/j.ss esea ch.2024.103092
Recei ed 29 Janua y 2024; Recei ed in e ised o m 26 Sep embe 2024; Accep ed 6 Oc obe 2024
Social Science Resea ch 124 (2024) 103092
0049-089X/© 2024 The Au ho s. Published by Else ie Inc. This is an open access a icle unde he CC BY-NC license
( h p://c ea i ecommons.o g/licenses/by-nc/4.0/ ).
science subjec s in high school (Jacob e al., 2020), o en oll in STEM ields a uni e si y (Mo gan e al., 2013), and o g adua e a e
en olling in STEM p og ams (Spee , 2023). Mo e o mally, e en a e aking a s ep owa ds STEM, he p obabili y o aking he nex
s ep is always lowe o women han o men. This gende ed a i ion con ibu es o he ma ked unde ep esen a ion o women among
STEM g adua es (Che yan e al., 2017), which has led o policy measu es aimed a encou aging women o s udy and g adua e in STEM
ields. Bu how success ul is his in ul ima ely inc easing women’s ep esen a ion in STEM occupa ions?
P e ious esea ch has shown ha he leaky pipeline does no s op a e g adua ion in STEM ields. S udies om he US (Cech and
Blai -Loy, 2019; Spee , 2023; Xu, 2015), he UK (Delaney and De e eux, 2022), Poland (Jasko e al., 2020), o Ge many (Schwe e
and Ilg, 2023) epo ha emale STEM g adua es a e less likely o ind a deg ee- ela ed job a e g adua ion, and ha women’s
a i ion om STEM occupa ions con inues g adually o e hei ca ee s. We join his esea ch e o and examine gende di e ences in
he ansi ion om g adua ing wi h a STEM deg ee o employmen in a STEM occupa ion in Spain.
We use high-quali y Spanish da a on abou 20,000 STEM g adua es. The da a come om wo wa es o he Su ey on Labou Ma ke
Inse ion o Uni e si y G adua es, conduc ed by he Spanish S a is ical O ice. They con ain e ospec i e in o ma ion abou hei ca ee
in he 4/5 yea s a e g adua ion, oge he wi h ele an da a on hei educa ion and backg ound. The 2014 wa e includes indi iduals
who g adua ed om uni e si y a he end o he 2009-10 academic yea (i.e., du ing he wo s o he economic c isis), while he 2019
wa e includes indi iduals who g adua ed a he end o he 2013-14 academic yea (i.e., du ing he economic eco e y).
Ou s udy ex ends exis ing knowledge in ou espec s. Fi s , we examine he gende e ec o ansi ions immedia ely a e
g adua ion and 4/5 yea s in o he labo ma ke , which allows us o de e mine whe he he e is u he a i ion e en a e s a ing a
ca ee in a STEM occupa ion. Second, p e ious s udies ha e in es iga ed whe he economic condi ions a g adua ion a ec in-
di iduals’ ca ee s. We ollow his line o esea ch o examine whe he changes in job compe i ion induced by economic down u ns
mode a e he a i ion o women om STEM a e g adua ion. Thi d, we inco po a e an in e sec ional pe spec i e and conside he
in e play o gende and social backg ound (SES) by explo ing whe he he gende e ec on STEM pe sis ence di e s by SES. Finally, we
conduc ou analysis o he Spanish case, an in e es ing ye unexplo ed con ex o s udy (see below).
We ind a ema kable gende di e ence among STEM g adua es in he ansi ion o hei i s job, wi h women being no ably less
likely o wo k in a STEM occupa ion. We also obse e a sizeable gende e ec o he cu en job condi ional on s a ing he ca ee in
he STEM sec o . Mo eo e , he gende di e ence in he p obabili y o being employed in he STEM sec o is subs an ially la ge o
g adua es du ing he economic c isis and o indi iduals om low-SES backg ounds.
The pape is s uc u ed as ollows. We begin by p esen ing some in o ma ion on he Spanish con ex , including in o ma ion on
o he coun ies o con ex ualize he Spanish case and unde s and why i is pa icula ly in e es ing o ou esea ch ques ions. Ou
heo e ical conside a ions ollow, alongside he desc ip ion o he da a sou ce, he main a iables, and he me hod used. We hen
p esen he esul s, hei discussion, and he main conclusions o ou s udy.
2. The Spanish case
Akin o he gene al end, women in Spain a e much mo e likely han men o ob ain a uni e si y deg ee: 60 pe cen o all Bachelo
g adua es a e women. Howe e , hey a e signi ican ly unde ep esen ed among STEM g adua es. Women accoun o only 14 pe cen
o all g adua es in Compu e Science, 28 pe cen in Enginee ing, and 40 pe cen in Ma hema ics and S a is ics. None heless, no all
STEM ields a e male domina ed. Fo example, a ound wo- hi ds o g adua es in biology- ela ed deg ees a e women.
2
Compa ed o o he coun ies, he emale sha e among e ia y STEM g adua es in Spain in 2013–14 was lowe han in he US, he
UK, o Poland, bu highe han in Finland, No way, Ge many, o he Ne he lands (see Table 1). Fu he mo e, while he emale sha e
dec eased in Spain (simila o No way and Finland) be ween 2009–10 and 2013–14, i inc eased in Poland, he UK, and he
Ne he lands.
Impo an ly o ou wo k, Spain no only has one o he lowes e ili y a es in Eu ope bu also a compa a i ely high mean age a
i s bi h o women, which inc eased om 29.8 in 2010 o 31.1 in 2019.
3
Spanish women’s ep oduc i e beha io also a ies
conside ably by educa ional le el, wi h uni e si y-educa ed women being signi ican ly mo e likely o emain childless and o become
mo he s la e in li e (Compans e al., 2023; Lozano e al., 2024).
Ne e heless, he employmen a e o Spanish women (a ound 42% in 2009 and 40% in 2014)
4
is he lowes o all he coun ies
p esen ed in Table 1. In con as , and his ime consis en wi h he low e ili y a e, he sha e o pa - ime wo k among Spanish women
is compa a i ely low (41% in 2009 and 42% in 2014) – compa ed o he Ne he lands (78% in bo h yea s), No way (61%–60%), he UK
(58%–59%) o Ge many (56%–58%). Only in Poland (30%–32%) and he US (39%–35%), pa - ime wo k is lowe han in Spain.
5
Finally, Spain was hi e y ha d by he G ea Recession (2008–2014) caused by he 2007–2008 global inancial c isis. Fig. 1 shows
ha economic ac i i y collapsed in2008 and did no eco e un il 2014. The unemploymen a e sky ocke ed du ing he c isis,
especially among young people, and hen g adually declined du ing he economic eco e y.
O e all, he Spanish case p o ides a pa icula ly in e es ing con ex o add essing ou esea ch ques ions. On he one hand, he isk
2
Da a o he 2021–2022 academic yea coming om he Minis y o Uni e si y s a is ics da abase (h ps://www.uni e sidades.gob.es/
es adis ica-de-es udian es/).
3
Da a om UNECE (h ps://w3.unece.o g/PXWeb/en/Table?Indica o Code=34). As compa ison, he mean age in 2019 was 29.8 in Ge many,
27.6 in Poland, 27.0 in he US, and 29.0 in he UK (in 2018).
4
Wo ld Bank, Gende S a is ics, e ie ed 12/15/2023.
5
Wo ld Bank, Gende S a is ics, e ie ed 12/15/2023.
M.T. Vald´
es and H. Solga
Social Science Resea ch 124 (2024) 103092
2
o emale uni e si y g adua es being disad an aged due o i s bi h and child en a labo ma ke en y (as shown by Cech and
Blai -Loy 2019; Xu 2015 o he US) is compa a i ely low in Spain, gi en hei low and la e e ili y a es. As a esul , gende di e ences
in STEM pe sis ence will be less a ec ed by selec i e pa en hood. On he o he hand, job compe i ion in Spain was e y high du ing he
c isis and dec eased signi ican ly a e wa d. As p e ious esea ch has es ablished he consequen ial ca ee impac o g adua ing in
imes o economic down u n (Al onji e al., 2016; Del Bono and Mo ando, 2022; Kahn, 2010; O eopoulos e al., 2012), his pe iod o
ex eme economic s uggle in he Spanish his o y p o ides a highly ele an con ex o es whe he he economic condi ions a
g adua ion (and hus di e ences in job compe i ion) mode a e he gende e ec in STEM pe sis ence.
3. Theo e ical conside a ions
As wi h any labo ma ke placemen , gende di e ences in STEM employmen among STEM g adua es may esul om supply- and
demand-side decision p ocesses. On he supply side, women wi h STEM deg ees may op ou o STEM jobs mo e o en han hei male
coun e pa s. On he demand side, STEM employe s may p e e o hi e men o he male-domina ed wo k en i onmen o STEM jobs.
Like o he s udies based on obse a ional da a (Cech and Blai -Loy, 2019; Delaney and De e eux, 2022; Jasko e al., 2020; Schwe e
and Ilg, 2023; Spee , 2023; Xu, 2015), we a e only able o examine he join ou come o bo h o ces. Howe e , we will p o ide some
indi ec e idence abou he unde lying mechanisms.
3.1. Gende e ec in he ansi ion om STEM deg ees o STEM occupa ions
P e ious esea ch has shown ha women a e less likely o be employed in STEM occupa ions, con ibu ing signi ican ly o he
gende pay gap (B eda e al., 2023; Delaney and De e eux, 2022). The ques ion, howe e , is whe he women a e s ill less likely o be
employed in STEM occupa ions e en when hey g adua e wi h a STEM deg ee.
Table 1
A e age sha e o emale g adua es in e ia y STEM p og ams, Spain and selec ed coun ies (%).
Coun y 2009–2010 2013–2014
Ne he lands 20.9 25.0
No way 31.3 25.7
Ge many –26.8
Finland 28.7 27.4
Spain 31.1 29.3
Uni ed S a es (US) 31.4 31.7
Uni ed Kingdom (UK) 30.6 37.6
Poland 39.2 42.2
No es: Coun ies selec ed o he pu pose o compa ison. Yea s a e chosen o ma ch ou s udy
yea s. Fo Ge many, da a a e only a ailable o 2015.
Sou ce: Wo ld Bank, Gende S a is ics, e ie ed 12/15/2023.
Fig. 1. Unemploymen a e and GDP in Spain, 2006–2019.
Sou ce: Unemploymen da a om he Labou Fo ce Su ey collec ed by he Spanish S a is ical O ice. GDP da a om Na ional Accoun s p o ided by
he Spanish S a is ical O ice. Au ho s’ own elabo a ion.
M.T. Vald´
es and H. Solga
Social Science Resea ch 124 (2024) 103092
3
Two scena ios a e possible. On he one hand, he gende e ec may be e e sed in he ansi ion om STEM deg ees o STEM
occupa ions. Due o he g ea e a i ion among women on he pa h o a STEM deg ee, emale STEM g adua es may be mo e selec ed
han hei male coun e pa s and, on a e age, s onge candida es o STEM jobs (“su i o bias”; c . Auspu g e al., 2017, p. 284).
Thus, p opo ionally mo e emale STEM g adua es should ge STEM jobs, e e sing he gende e ec in a o o women.
On he o he hand, mos exis ing esea ch examining he labo ma ke ansi ion and subsequen ca ee ou comes o STEM
g adua es has ound e idence ha emale STEM g adua es a e less likely o wo k in STEM occupa ions han hei male coun e pa s
(Cech and Blai -Loy, 2019; Delaney and De e eux, 2022; Jasko e al., 2020; Schwe e and Ilg, 2023; Spee , 2023; Xu, 2015).
Supply-side explana ions o his lowe pe sis ence o emale STEM g adua es a e ha hey may op ou o STEM ca ee s mo e o en
han men. In he same way ha women end o d i away om STEM ields du ing hei educa ional jou ney, hey may also be mo e
likely o p e e o he occupa ional des ina ions a e comple ing hei STEM educa ion (sel -selec ion mechanism) (Co ell, 2001;
F iedmann, 2018). On he demand side, and gi en he ac ha he STEM wo k o ce is hea ily male-domina ed, STEM employe s
(mos ly men hemsel es) may p e e o hi e male candida es (disc imina ion mechanism), ei he because employe s p e e o p ese e
his male-domina ed wo k en i onmen o because hey expec men o i be e and as e in o hese en i onmen s (F iedmann and
E a -T eis e , 2023; Funk and Pa ke , 2018). The combina ion o hese p ocesses leads o he ollowing hypo hesis:
Hypo hesis 1. Female STEM g adua es a e less likely o wo k in a STEM occupa ion han hei male coun e pa s.
Indeed, many schola s ha e epo ed ha emale STEM g adua es a e subs an ially less likely o wo k in he STEM sec o , bo h
immedia ely a e g adua ion (Jasko e al., 2020; Schwe e and Ilg, 2023) and la e in hei ca ee s (Delaney and De e eux, 2022;
Spee , 2023). I any hing, he ini ial gende e ec a labo ma ke en y inc eases o e he li e cou se (Delaney and De e eux, 2022).
In e es ingly, pa en hood has been ound o play a majo ole in women’s pe sis ence in he STEM sec o , which may pa ly explain he
epo ed gende di e ence in STEM pe sis ence (Cech and Blai -Loy, 2019; Xu, 2015). Ex ending his esea ch, we examine whe he
he gende e ec pe sis s also among STEM g adua es who ha e s a ed hei ca ee s in he STEM sec o . Gi en he low e ili y a e and
la e ansi ion o i s mo he hood in Spain, his would be mainly due o women’s expe iences wi h he wo k cul u e, p omo ion
s uc u es, wo k- amily no ms, and so on in he male-domina ed wo k en i onmen s o STEM occupa ions. Female STEM g adua es
who s a ed hei ca ee s in STEM may hen decide o change hei ca ee pa hs and op o employmen in o he sec o s. Mo eo e ,
emale STEM g adua es who y o change jobs wi hin he STEM sec o may again ace disc imina o y hi ing p ac ices by STEM
employe s and end up in non-STEM occupa ions o unemployed. The e o e, we hypo hesize ha :
Hypo hesis 2. Female STEM g adua es a e less likely o be employed in a STEM occupa ion 4–5 yea s a e g adua ion, e en when
hey s a hei ca ee in a STEM occupa ion.
3.2. Economic condi ions as a mode a o o he gende e ec in STEM
Resea ch has shown ha occupa ional misma ch (such as STEM g adua es wo king in non-STEM jobs) is highe he g ea e he job
compe i ion (Bo gna e al., 2019). Acco ding o he job compe i ion model, labo ma ke ou comes a e de e mined by wo ac o s: he
indi idual’s posi ion in he labo queue and he leng h o he queue o job acancies (Thu ow, 1979). S onge candida es (mo e
educa ed and skilled) a e a he highe anks o he labo queue, so hey a e he i s o seize he oppo uni ies o e ed by he economy.
In u n, lowe -skilled candida es ace g ea e di icul ies in inding jobs, especially du ing economic down u ns when job oppo uni ies
become sca ce, and he supply-demand a io inc eases.
Impo an ly, due o he a o emen ioned disc imina ion mechanism, he labo queue is likely o be gende ed (Reskin and Ross,
1990), wi h men placed ahead o equally educa ed and skilled women. Empi ical e idence sugges s ha employe s p ac ice such
gende ed queuing, e en o iden ically quali ied male and emale candida es (Ba on e al., 2022). In ha case, he gende di e ence in
STEM pe sis ence would be la ge du ing economic down u ns, as he economic c isis sho ens he queue o job acancies, and he
cu -o poin o he las applican in he labo queue o ge a STEM job mo es o highe anks. I women a e indeed unde ep esen ed a
hese highe anks, a la ge numbe o emale STEM g adua es would be o ced o look ou side he STEM sec o compa ed o imes o
economic p ospe i y and lowe job compe i ion, inc easing hei p esence in non-STEM occupa ions and unemploymen . In suppo o
hese conside a ions, we expec o ind ha :
Hypo hesis 3a. The gende e ec in he ansi ion om STEM deg ees o STEM occupa ions is la ge du ing economic down u ns.
Howe e , we migh also a gue ha he p e ailing mechanism in gene a ing he gende di e ence in STEM pe sis ence is he sel -
selec ion mechanism desc ibed abo e. In his case, ha gende di e ence would be smalle du ing economic down u ns, as he c isis
may educe he oppo uni ies o emale STEM g adua es o wo k in (and o compe e o ) non-STEM jobs. This may o ce hem o make
he mos o hei educa ion and o ake ull ad an age o hei s udies, ha is, o seek o s ay in STEM occupa ions (as much as possible)
o a oid unemploymen . I so, we expec o obse e ha :
Hypo hesis 3b. The gende e ec is smalle du ing economic down u ns.
No e ha while Hypo hesis 3a emphasizes he impac o demand-side o ces and he disc imina ion mechanism discussed ea lie ,
Hypo hesis 3b s esses he ole o supply-side o ces and women’s wi hd awal om he STEM sec o . Thus, suppo o one o hese wo
compe ing hypo heses p o ides indi ec e idence o he impo ance o ei he mechanism in gene a ing he gende e ec in he
ansi ion om STEM deg ees o STEM occupa ions.
M.T. Vald´
es and H. Solga
Social Science Resea ch 124 (2024) 103092
4
3.3. In e play be ween social o igin and gende
The concep o in e sec ionali y e e s o he ac ha indi idual cha ac e is ics may in e ac o p oduce speci ic o ms o dis/
ad an age, such as being a black low-SES woman (C enshaw, 1991; McCall, 2005). The in e ac ion be ween social o igin and gende is
commonly conside ed when analyzing educa ional ou comes, mos ly epo ing ha he nega i e e ec o being male on educa ional
expec a ions (O iz-Ge asi, 2020) and a ainmen (Combe and Oesch, 2021) is la ge among s uden s om low-SES amilies. This
in e ac ion has also been in es iga ed when s udying he ansi ion om educa ion o he labo ma ke (Zimme mann and Seile ,
2019). Applied o ou esea ch opic, gende may in e ac wi h social o igin in STEM pe sis ence, such ha being emale is a (mo e)
salien ac o depending on women’s social backg ound. Speci ically, we a gue ha he disad an age o emale STEM g adua es in
inding STEM jobs may be g ea e o women om low-SES amilies.
To subs an ia e ou expec a ion, we d aw on England’s (2010) a gumen ha men and women end o choose gende - ypical o
gende -neu al jobs. To engage in jobs whe e hei gende is a mino i y, hey need addi ional incen i es such as he mo i a ion o
secu e hei social posi ion o o igin. As he STEM sec o is hea ily male-domina ed and consis mos ly o high-paying occupa ions –
bo h impo an mo i es o men’s ca ee choices ega dless o social backg ound (Mon ma que e e al., 2002; Za a , 2013) – we expec
a high p obabili y o STEM pe sis ence o bo h low- and high-SES men. We also an icipa e a high p obabili y o STEM pe sis ence o
high-SES emale STEM g adua es because hey may need a job ha ma ches hei STEM deg ee o main ain hei social posi ion o
o igin. In u n, low-SES emale STEM g adua es can be employed ou side he STEM sec o wi hou isking in e gene a ional downwa d
mobili y. The e o e, we p edic o hem a lowe p obabili y o STEM pe sis ence. As suppo o hese heo e ical conside a ions, we
expec o obse e ha :
Hypo hesis 4. The gende e ec in he ansi ion o STEM occupa ions is la ge o low-SES g adua es han o high-SES g adua es.
4. Da a and me hod
4.1. Da a
To es ou hypo heses, we use da a om wo wa es o he Encues a de Inse ci´
on Labo al de i ulados Uni e si a ios (EILU, Su ey on
Labo Ma ke Inse ion o Uni e si y G adua es). The i s EILU wa e was conduc ed in 2014, in which 30,379 indi iduals who
g adua ed wi h a Bachelo ’s deg ee a he end o he 2009-10 academic yea we e in e iewed abou hei educa ional and
employmen ajec o ies in he ou yea s a e g adua ion. We e e o his wa e as he c isis coho because hey made hei labo
ma ke ansi ion du ing he economic c isis. The second EILU wa e – ou pos -c isis coho – was conduc ed in 2019, in which 31,561
Bachelo g adua es om he 2013-14 academic we e in e iewed abou hei educa ional and employmen ajec o ies in he i e yea s
ollowing g adua ion.
Ou analysis is es ic ed o STEM g adua es. As in p e ious esea ch, we de ine STEM as hose ields ela ed o science (such as
biology, chemis y, o physics), echnology (such as compu e science o in o ma ion echnology), enginee ing (such as ci il engi-
nee ing o a chi ec u e), and ma hema ics (such as ma hema ics o s a is ics). Ou analy ical sample consis s o 19,230 indi iduals, o
which 11,640 a e men and 7,590 (39.4%) a e women (see Table 2).
6
While echnology and enginee ing a e clea ly male-domina ed
ields, science and ma hema ics a e a he gende -neu al ields (bu ha e become less a ac i e o women in he pos -c isis coho ). As
he gende dis ibu ion di e s ac oss he STEM sub ields, we es he obus ness o ou indings by e uning he analyses sepa a ely o
each o he ou sub ields and including STEM sub ields as a con ol a iable. Mo eo e , we also examine an addi ional sample o 3,173
STEM Mas e ’s g adua es su eyed only in he second EILU wa e.
4.2. Va iables in he s udy
In bo h EILU wa es, pa icipan s epo ed hei i s occupa ion a e g adua ion and hei occupa ion a he ime o he su ey
(he eina e , cu en occupa ion). Fo he i s occupa ion, we c ea e a a iable indica ing whe he STEM g adua es we e employed in
a STEM occupa ion, employed in a non-STEM occupa ion, o had ne e been employed. Fo he cu en occupa ion, he a iable
dis inguishes be ween employmen in a STEM occupa ion a he ime o he su ey, employmen in a non-STEM occupa ion, o un-
employmen (including inac i e pe sons, i no o he wise speci ied). The de ini ion o STEM occupa ion ollows he same c i e ia as
used o he ield o s udy (see abo e). In o ma ion on i s and cu en jobs allows o examine bo h whe he emale STEM g adua es a e
less likely han hei male coun e pa s o s a hei ca ee s in STEM occupa ions, and whe he emale STEM g adua es whose i s job
was in STEM a e less likely o emain in ha sec o .
To u he quali y ou indings ela ed o Hypo hesis 1 and 2, we use in o ma ion on how long i ook o pa icipan s o ge hei
i s job a e g adua ion, whe he hey changed jobs be ween hei i s and cu en jobs, and he sala y o hei i s job. Sala ies a e
only a ailable o he second EILU wa e, and only as a ca ego ical a iable wi h se en le els o ne mon hly sala y (and no con inuous
hou ly wages).
7
We use he class ma ks o hese wage in e als o es ima e a linea model. We e an he analysis by calcula ing he
6
The p opo ion o women among all g adua es (i.e., including STEM and non-STEM p og ammes) is 58.4 pe cen .
7
In he 2019 wa e o he s udy, indi iduals epo ed hei ne mon hly sala y in in e als: (1) less han 700
€
; (2) 700–999
€
; (3) 1,000–1,499
€
; (4)
1,500–1,999
€
; (5) 2,000–2,499
€
; (6) 2,500–2,999
€
; (6) mo e han 3,000
€
.
M.T. Vald´
es and H. Solga
Social Science Resea ch 124 (2024) 103092
5
loga i hm o hese alues and he esul s emain mos ly he same (no shown).
We use he coho di e ence in labo ma ke en y condi ions o examine whe he he gende e ec a ies wi h job compe i ion
(Hypo heses 3a/3b). Fi s -wa e pa icipan s en e ed he labo ma ke du ing yea s o in ense economic s uggle, while second-wa e
pa icipan s did so in a con ex o economic eco e y (see Fig. 1 abo e).
To es whe he he gende e ec a ies wi h he indi idual’s social o igin (Hypo hesis 4), we use in o ma ion om he second-
wa e pa icipan s on hei pa en s’ educa ional a ainmen ( i s -wa e pa icipan s we e no asked abou his). We dis inguish in-
di iduals wi h no (low-SES), one (in -SES), and wo uni e si y-educa ed pa en s (high-SES).
As we in end o examine he gende e ec be ween equi alen male and emale STEM g adua es, we adjus o a numbe o con ols.
Fi s , we include he indi idual’s age as i a ec s he sea ch ime a ailable and he employabili y o he candida e. No e ha he age
in o ma ion is a he ime o he su ey, so indi iduals we e younge a g adua ion and mos likely also younge a hei i s job.
Mo eo e , di e ences in pe o mance and quali ica ions may in luence STEM employe s’ assessmen o male and emale applican s. As
we do no ha e di ec measu es (such as g ades), we include ou p oxy indica o s: whe he indi iduals ecei ed an excellence awa d a
he end o he deg ee, whe he hey wen ab oad du ing hei s udies, whe he hey wen ab oad a e g adua ion, and whe he hey
had ob ained a Mas e ’s deg ee. Thi d, we es Hypo heses 1–3a/b con olling o indi idual’s social o igin. We use h ee p oxies o
social o igin a ailable o bo h coho s (pa en al educa ion is only p o ided in he second coho , see abo e): whe he indi iduals
ecei ed a g an du ing hei uni e si y s udies, whe he hey a ended a public o p i a e uni e si y, and whe he hey s udied pa o
hei s udies ab oad. The co ela ions be ween hese a iables and pa en al educa ion o he pos -c isis coho jus i y hei inclusion as
p oxies o social o igin (see Online Supplemen , Table S1). To a oid o e con ol bias (G ¨
a z, 2022), we do no include hese a iables
when es ing Hypo hesis 4 on he mode a ing ole o social o igin (as we use he di ec measu e o pa en al educa ion he e, see abo e).
We do no ha e in o ma ion on he p esence o child en o he age a i s bi h, bo h o which may con ibu e o gende di e ences
in STEM employmen (Cech and Blai -Loy, 2019; Xu, 2015). Howe e , as discussed abo e, he ansi ion o mo he hood in Spain is la e,
especially o uni e si y g adua es. The e o e, we do no expec ha mo he hood plays a la ge ole in explaining ou indings on labo
ma ke en y. In he supplemen a y analysis on sala ies in he i s job (see abo e), we include pa - ime wo k as a con ol. This
a iable can be in e p e ed as indica i e o he impac o mo he hood o some ex en .
8
Table 3 shows desc ip i e s a is ics o all a iables by g adua ion coho o he ull sample and o he STEM g adua e subsample.
The wo coho s a e e y simila , excep ha he pos -c isis coho is no ably olde han he c isis coho . This age di e ence is
consis en wi h he decisions o many Spania ds o e u n o educa ion du ing he economic c isis (Vald´
es, 2020). Mo eo e , a highe
p opo ion o g adua es in he pos -c isis coho ecei ed a g an du ing hei s udies, e lec ing he economic ha dship expe ienced by
many amilies du ing he c isis. Rega ding ou ou come a iables, he sha e o STEM g adua es who we e unemployed o ne e
employed a he ime o su ey is no ably lowe in he pos -c isis coho , mainly due o non-STEM employmen . C ucially, he gende
composi ion o he STEM g adua e sample is he same in bo h coho s: a ound 60 pe cen men and 40 pe cen women. This allays
conce ns ha women o men we e disp opo iona ely discou aged/encou aged om choosing STEM ields du ing he c isis. I any-
hing, women we e less likely o g adua e in mo e gende -neu al STEM sub ields (as shown in Table 2 abo e).
4.3. Me hod
Ou wo dependen a iables a e he h ee-ca ego y nominal a iables i s occupa ion and cu en occupa ion. To assess he e ec o
gende on hese wo a iables a e adjus ing o a well-de ined se o con ols, we employ he ollowing mul inomial model:
ln(P(Y=j)
P(Y=b))=
α
b+βjX+γjZj=1,…k and j ∕= b(1)
whe e Y is he dependen a iable, j ep esen s he ca ego ies in Y, b ep esen s he e e ence ca ego y agains which he es o he
al e na i es a e compa ed (STEM-occupa ion),
α
is a cons an , X is he gende o he indi idual, β
j
a e he coe icien s o gende o
al e na i e j, and Z is a ec o o con ols. All models include obus s anda d e o s.
Table 2
Analy ical sample by STEM ield and gende (%).
To al C isis coho Pos -c isis coho
N % women N % women N % women
Science 6,082 54.4 2,975 57.7 3,107 51.1
Technology 3,901 28.4 2,110 29.4 1,791 27.1
Enginee ing 8,455 32.7 4,282 31.1 4,173 34.4
Ma hema ics 792 52.0 381 58.8 411 45.7
To al 19,230 39.4 9,748 39.9 9,482 39.0
8
Acco ding o Eu opean s a is ics om he Labou Fo ce Su ey, Eu opean women aged 25 o 54 a e signi ican ly mo e likely o wo k pa - ime i
hey ha e child en (32%) compa ed o childless women (20%). See: h ps://ec.eu opa.eu/eu os a /da ab owse / iew/l s _hhp e y__cus om_
12992737/de aul / able?lang=en&page= ime:2023).
M.T. Vald´
es and H. Solga
Social Science Resea ch 124 (2024) 103092
6
The model is adjus ed o STEM g adua es om he wo EILU wa es, including he yea o g adua ion as a ixed e ec . To es
Hypo hesis 2, we u he es ic he analy ical sample o STEM g adua es whose i s job was in STEM o who changed jobs be ween
hei i s and cu en occupa ion.
The esul s a e p esen ed in he o m o A e age Ma ginal E ec s (AME) o gende on he p edic ed p obabili y o wo king in a
STEM occupa ion, in a non-STEM occupa ion, o being unemployed. As we will conside in e ac ion e ec s in a non-linea model,
examining he esul s in e ms o p edic ed p obabili ies en ails a signi ican ad an age (Mize, 2019).
5. Resul s
5.1. Gende e ec in he ansi ion o he i s job
We s a by es ing he gende e ec on he ansi ion om g adua ion in a STEM ield o he i s job. The esul s a e shown
g aphically in he le -hand panel o Fig. 2. As is e iden , he gende e ec in his ansi ion is ema kable: emale STEM g adua es a e
11.3 pe cen age poin s (pp) less likely o ha e a i s job in STEM han hei male coun e pa s (p- alue ≤0.001). Con e sely, emale
STEM g adua es a e subs an ially mo e likely han men o ind hei i s job ou side he STEM sec o (AME =0.106; p- alue ≤0.001),
bu hey a e equally likely o ha e ne e been employed a e g adua ion un il he ime o he su ey (AME =0.007; p- alue =0.016).
Table 3
Desc ip i e s a is ics (column pe cen ages).
Full sample STEM g adua es (Bachelo )
G adua ed 2010 (c isis
coho )
G adua ed 2014 (pos -c isis
coho )
G adua ed 2010 (c isis
coho )
G adua ed 2014 (pos -c isis
coho )
Field o s udy (Bachelo )
Non-STEM 67.9% 70.0% – –
STEM 32.1% 30.0% – –
Fi s occupa ion
Non-STEM occupa ion 72.1% 80.3% 33.2% 48.6%
STEM occupa ion 22.0% 17.5% 61.5% 50.1%
Ne e wo ked 5.9% 2.2% 5.3% 1.4%
Cu en occupa ional si ua ion
Employed: Non-STEM 55.6% 67.6% 23.5% 39.3%
Employed: STEM 19.0% 18.1% 53.3% 50.4%
Unemployed 25.4% 14.3% 23.2% 10.3%
Gende
Men 40.3% 43.0% 60.1% 61.0%
Women 59.7% 57.1% 39.9% 39.0%
Age a su ey
Less han 30 59.0% 49.6% 49.1% 39.7%
30-34 25.1% 27.9% 35.2% 33.4%
35 o mo e 15.9% 22.5% 15.7% 27.0%
Type o uni e si y
Public 86.1% 85.2% 85.7% 90.3%
P i a e 14.0% 14.8% 14.3% 9.7%
Excellence awa d
No 97.8% 95.6% 97.3% 95.2%
Yes 2.3% 4.4% 2.7% 4.8%
G an du ing s udies
No 65.4% 61.9% 68.0% 64.7%
Yes 34.6% 38.2% 32.0% 35.3%
Wen ab oad du ing s udies
No 86.5% 83.3% 85.6% 81.6%
Yes 13.5% 16.7% 14.4% 18.4%
Wen ab oad a e g adua ion
No 83.8% 82.5% 81.7% 81.0%
Yes 16.2% 17.6% 18.3% 19.0%
Comple ed a Mas e ’s deg ee
No 69.7% 56.3% 66.7% 55.6%
Yes 30.3% 43.8% 33.3% 44.4%
Pa en al educa ion
No uni e si y-educa ed
pa en
–58.6% –55.9%
One uni e si y-educa ed
pa en
–20.3% –22.0%
Two uni e si y-educa ed
pa en s
–17.8% –19.1%
To al 30,379 31,651 9,748 9,482
Sou ce: EILU (wa es 1 and 2).
M.T. Vald´
es and H. Solga
Social Science Resea ch 124 (2024) 103092
7
Thus, despi e a po en ially s onge selec i i y in ob aining a STEM Bachelo ’s deg ee among women, emale STEM g adua es a e
no ably less likely han hei male coun e pa s o s a hei ca ee s in a STEM occupa ion, consis en wi h Hypo hesis 1.
9
To es he obus ness o his inding, we e un he analysis o he sample o Mas e ’s g adua es (only a ailable o he pos -c isis
coho ). Again, women wi h a STEM Mas e ’s deg ee a e 8.1 pp (p- alue ≤0.001) less likely han men o ind hei i s job in a STEM
occupa ion (see Online Supplemen , Table S2, Model 1).
Impo an ly, women’s pa icipa ion is he e ogeneous ac oss STEM sub ields, and in some speci ic p og ams, such as biology,
women a e ac ually in he majo i y. To assess how his he e ogenei y may a ec ou indings, we s a i y he analysis by STEM sub ield,
dis inguishing g adua es in Science, Technology, Enginee ing, and Ma hema ics (le -hand panel o Fig. 3). The gende e ec on he
p obabili y o s a ing he ca ee in STEM is subs an ial and s a is ically signi ican in all sub ields, bu he e is a no able deg ee o
he e ogenei y. The gende e ec is la ges among Bachelo ’s g adua es in echnology p og ams (AME = − 0.147; p- alue =0.000) and
smalles among g adua es in enginee ing (AME = − 0.037; p- alue =0.001), bo h male-domina ed ields (see Table 2 abo e).
In e es ingly, he e is no co ela ion be ween he size o he gende e ec and he pe cen age o women in he sub ields.
To quali y ou esul s, we examine wo addi ional pieces o in o ma ion: he ime a e g adua ion un il inding he i s job and he
sala y o he i s job (see Appendix, Table A2). Fi s , emale STEM g adua es whose i s job was in a STEM occupa ion ake almos 2
mon hs longe han men o ind his job (p- alue ≤0.001). Howe e , hal o his di e ence is due o he ac ha emale STEM
g adua es a e mo e likely o con inue hei educa ion and comple e hei STEM deg ees a a younge age. A e including all con ols, i
akes 3 weeks longe o emale han male STEM g adua es han male g adua es (p- alue ≤0.001).
10
In ela i e e ms, women ake
16.6 pe cen longe han equi alen men o ind hei i s job i i is a STEM occupa ion.
Second, we also calcula e he gende di e ence in he ne mon hly sala y o he i s job among hose who s a ed in a STEM
occupa ion (only a ailable o he pos -c isis coho ; see Appendix, Table A2). Gi en da a limi a ions (see Sec ion 4.2), his analysis is
me ely explo a o y. Female STEM g adua es whose i s job was in STEM ea n almos 100
€
less han hei male coun e pa s (p- alue
≤0.001). A e adjus ing o all con ols (including wo king hou s, i.e., pa - ime o ull- ime), his di e ence is educed o 80
€
(p- alue
≤0.001). This means ha emale STEM g adua es s a ing hei ca ee s in STEM a e paid a ound 7.8 pe cen less han equi alen
males.
We also es ima ed he e ec o wo king ou side he STEM sec o on he s a ing sala y o emale STEM g adua es (see Appendix,
Table A3). Women wi h a i s job in STEM ea n 120
€
mo e han women wi h a non-STEM job (14.7 % highe sala y).
11
Con olling o
wo king hou s signi ican ly educes his di e ence o only 26
€
, because only 15.5 pe cen o emale STEM g adua es wi h a i s job in
STEM wo ked pa - ime, while his igu e mo e han doubles (35.6 %) among hei coun e pa s wi h a non-STEM job. This di e ence
sugges s ha mo he hood is associa ed wi h wo king ou side he STEM ield.
5.2. Gende e ec in he cu en occupa ion
Nex , we ocus on he occupa ional si ua ion a he ime o he su ey (i.e., cu en occupa ion). To es whe he he leaky pipeline
con inues o leak e en a e STEM g adua es ook hei i s job in he STEM sec o (Hypo hesis 2), we es ic he analysis o STEM
g adua es ha s a ed hei ca ee in a STEM job (Fig. 2 (Model 2). Female STEM g adua es whose i s job was in STEM a e 5.3 pp less
likely han men o be cu en ly employed in a STEM job (p- alue ≤0.001). Thus, no only a e emale STEM g adua es less likely o ind
a i s job in STEM, bu hey a e also less likely o s ay in STEM a e s a ing hei ca ee in a STEM job (consis en wi h Hypo hesis 2).
This holds o all sub ields bu , again, he gende e ec is la ges o echnology and ma hema ics g adua es and smalle o science and
enginee ing g adua es (see Fig. 3 abo e).
To quali y hese indings, we u he condi ion he analysis on ha ing changed jobs (see Online Supplemen , Table S4, Model 1).
The analy ical sample now includes only hose STEM g adua es whose i s job was in a STEM occupa ion, bu who had le ha job by
he ime o he su ey. Impo an ly, women a e no mo e likely han men o lea e hei i s job. Howe e , women who changed jobs
a e 7.6 pp less likely han men o be cu en ly employed in a STEM occupa ion (p- alue ≤0.001) and 3.6 pp mo e likely o be cu en ly
employed in a non-STEM occupa ion (p- alue ≤0.001). Thus, job changes inc ease he gende e ec in he cu en occupa ion con-
di ional on ha ing s a ed in a STEM occupa ion.
Addi ionally, women wi h job changes a e mo e likely han men o be cu en ly unemployed (AME =0.050; p- alue ≤0.001). This
means ha women who ha e le hei i s job in STEM may s ill wan o wo k in he STEM sec o bu a e less likely o ind a STEM job
because employe s p e e o hi e men. To cla i y his poin , we e un he p e ious model bu di ide he unemployed g oup in o wo
ca ego ies: inac i e and looking o a job (see Online Supplemen , Table S4, Model 2). We ind ha emale STEM g adua es who s a ed
hei ca ee in a STEM occupa ion bu hen changed jobs a e signi ican ly mo e likely han men o be looking o a job (AME =0.034; p-
alue ≤0.001) and only sligh ly mo e likely o be inac i e (AME =0.015; p- alue =0.009). So, indeed, women who ha e le hei i s
job may s ill be sea ching o a new STEM job, bu as wi h he i s occupa ion a e g adua ion, i is mo e di icul o hem han o
hei male coun e pa s. This again may sugges disc imina o y hi ing p ac ices by STEM employe s.
Finally, he igh -hand panel o Fig. 2 (Model 3) epo s he gende e ec in he cu en occupa ion wi hou condi ioning on he i s
job, ha is, he gende e ec on he p obabili y o being employed in a STEM job a he ime o he su ey among indi iduals who
9
Desc ip i e analyses show ha he mos common jobs o emale STEM g adua es wo king ou side he STEM sec o we e eaching- ela ed
occupa ions, heal h- ela ed occupa ions, and sales cle ks.
10
Fo STEM g adua es wi h a i s job in STEM, he a e age ime aken o ind hei i s job is 7.5 mon hs.
11
The a e age sala y o emale g adua es in hei i s job is 916
€
.
M.T. Vald´
es and H. Solga
Social Science Resea ch 124 (2024) 103092
8
Table A1 (con inued)
Model 1: Fi s occupa ion (n =19,230)
P(STEM occupa ion) P (non-STEM occupa ion) P (ne e employed)
Unadjus ed Adjus ed Unadjus ed Adjus ed Unadjus ed Adjus ed
Ob ained a mas e ’s deg ee (Re : Did
no )
−0.044
(0.008)**
0.046 (0.008)** −0.002 (0.003)
Model 2: Cu en occupa ion, condi ional on i s job in STEM (n =10,741)
Female (Re : Male) −0.059
(0.009)**
−0.053
(0.009)**
0.020
(0.006)**
0.018 (0.006)* 0.039 (0.007)** 0.035 (0.007)**
G adua ion 2014 (Re : 2010) −0.005
(0.008)**
0.005 (0.008) 0.085
(0.006)**
0.083 (0.006)** −0.081
(0.006)**
−0.088
(0.006)**
Age 30–34 (Re : <30) −0.059
(0.009)**
0.041 (0.007)** 0.018 (0.007)*
Age 35+(Re : <30) −0.040
(0.012)**
0.002 (0.008) 0.038 (0.010)**
P i a e uni e si y (Re : Public) 0.016 (0.012) −0.002 (0.009) −0.015 (0.009)
x
Excellence awa d (Re : No awa d) 0.057 (0.018)** −0.021 (0.013) −0.037
(0.013)*
G an (Re : No g an ) −0.013 (0.009) −0.008 (0.007) 0.021 (0.007)*
Wen ab oad du ing s udies (Re : Did
no )
0.009 (0.010) 0.003 (0.008) −0.012 (0.008)
Wen ab oad a e g adua ion (Re : Did
no )
−0.075
(0.011)**
0.020 (0.008)* 0.055 (0.009)**
Ob ained a mas e ’s deg ee (Re : Did
no )
−0.069
(0.009)**
0.031 (0.007)** 0.038 (0.007)**
Model 3: Cu en occupa ion (n =19,230)
Female (Re : Male) −0.110
(0.007)**
−0.110
(0.007)**
0.061
(0.007)**
0.064 (0.007)** 0.050 (0.006)** 0.045 (0.006)**
G adua ion 2014 (Re : 2010) −0.030
(0.007)**
−0.012 (0.007) 0.158
(0.007)**
0.147 (0.007)** −0.129
(0.005)**
−0.136
(0.005)**
Age 30–34 (Re : <30) −0.045
(0.008)**
0.035 (0.008)** 0.009 (0.006)
Age 35+(Re : <30) −0.106
(0.010)**
0.082 (0.010)** 0.024 (0.008)*
P i a e uni e si y (Re : Public) 0.044 (0.011)** −0.008 (0.010) −0.036
(0.008)**
Excellence awa d (Re : No awa d) 0.098 (0.019)** −0.038
(0.017)*
−0.060
(0.012)**
G an (Re : No g an ) −0.027
(0.008)**
−0.008 (0.007) 0.034 (0.006)**
Wen ab oad du ing s udies (Re : Did
no )
0.061 (0.010)** −0.039
(0.009)**
−0.022
(0.007)*
Wen ab oad a e g adua ion (Re : Did
no )
−0.022
(0.010)*
−0.040
(0.008)**
0.061 (0.008)**
Ob ained a mas e ’s deg ee (Re : Did
no )
−0.072
(0.008)**
0.041 (0.007)** 0.031 (0.006)**
No es: A e age Ma ginal E ec s (AME). Robus s anda d e o s in b acke s. Age a ime o he su ey. **p- alue ≤0.001; *p- alue ≤0.05; †p- alue
≤0.10.
Sou ce: EILU (wa es 1 and 2), only STEM g adua es.
Table A2
Reg essions on ime o i s job and ne mon hly sala y o i s job.
Time o he i s job (mon hs)
a
Sala y o he i s job
b
Unadjus ed Adjus ed Unadjus ed Adjus ed
Women (Re : Men) 1.802 (0.22)** 0.765 (0.21)** −98.172 (17.34)** −80.506 (15.76)**
G adua ion 2014 (Re : 2010) 1.097 (0.21)** 1.330 (0.20)** – –
Age 30–34 (Re : <30) −2.758 (0.23)** −54.703 (18.70)*
Age 35+(Re : <30) −6.245 (0.25)** 220.967 (22.37)**
P i a e uni e si y (Re : Public) −0.506 (0.28)†30.686 (25.57)
Excellence awa d (Re : No awa d) −1.653 (0.50)** 125.307 (34.24)**
G an (Re : No g an ) 0.049 (0.22) −60.605 (16.32)**
Wen ab oad du ing s udies (Re : Did no ) −0.917 (0.26)** 115.838 (20.72)**
Wen ab oad a e g adua ion (Re : Did no ) 1.560 (0.27)** 132.099 (21.93)**
Ob ained a mas e ’s deg ee (Re : Did no ) 4.018 (0.23)** −18.627 (17.23)
Full- ime job (Re : Pa - ime job) 513.121 (18.83)**
N 10,723 4,553
M.T. Vald´
es and H. Solga
Social Science Resea ch 124 (2024) 103092
15
No es:**p- alue ≤0.001; *p- alue ≤0.05; †p- alue ≤0.10.
a
OLS model o bo h coho s.
b
OLS model only o he pos -c isis coho . Robus s anda d e o s in b acke s. Age a ime o he su ey.
Sou ce: EILU (wa es 1 and 2), only STEM g adua es whose i s job was in a STEM occupa ion.
Table A3
OLS eg essions: E ec o i s job in a STEM s. non-STEM occupa ion on ne mon hly sala y in i s job among women.
Model 1 Model 2
STEM occupa ion (Re : non-STEM) 120.5 (18.0)** 26.0 (16.9)
Age 30–34 (Re : <30) −73.3 (21.1)** −48.6 (19.2)*
Age 35+(Re : <30) 221.8 (26.3)** 189.9 (23.6)**
P i a e uni e si y (Re : Public) 50.2 (33.0) 19.9 (29.9)
Excellence awa d (Re : No awa d) 82.2 (39.3)* 78.5 (34.5)*
G an (Re : No g an ) −98.6 (18.4)** −89.1 (16.6)**
Wen ab oad du ing s udies (Re : Did no ) 62.7 (24.5)* 62.6 (22.5)*
Wen ab oad a e g adua ion (Re : Did no ) 92.3 (25.2)** 86.5 (23.0)**
Ob ained a mas e ’s deg ee (Re : Did no ) −29.5 (19.5) −11.8 (17.6)
Full- ime job (Re : Pa - ime job) 518.3 (16.5)**
N 3,494 3,494
No e: Only emale STEM g adua es o he pos -c isis coho . Robus s anda d e o s in b acke s. Age a he ime o he su ey. **p- alue ≤0.001; *p-
alue ≤0.05; †p- alue ≤0.10.
Sou ce: EILU (wa e 2), only emale STEM g adua es.
Re e ences
Alm, S., 2015. D eams mee ing eali y? A gende ed pe spec i e on he ela ionship be ween occupa ional p e e ences in ea ly adolescence and ac ual occupa ion in
adul hood. J. You h S ud. 18, 1077–1095. h ps://doi.o g/10.1080/13676261.2015.1020928.
Al onji, J.G., Kahn, L.B., Spee , J.D., 2016. Cashie o consul an ? En y labo ma ke condi ions, ield o s udy, and ca ee success. Jou nal o Labou Economics 34,
361–401. h ps://doi.o g/10.1086/682938.
Auspu g, K., Hinz, T., Schneck, A., 2017. Be u ungs e ah en als Tu nie e: Be u ungschancen on Wissenscha le innen und Wissenscha le n. Z. Soziol. 46, 283–302.
h ps://doi.o g/10.1515/z soz-2017-1016.
Ba on, K., Di lmann, R., Geh ig, S., Schweigho e -Kod i sch, S., 2022. Explici and implici belie -based gende disc imina ion: a hi ing expe imen . CESi o Wo king
Pape s 9731. Munich 1–42. h ps://www.cesi o.o g/node/69404.
Be yman, S.E., 1983. Who Will Do Science? T ends, and Thei Causes in Mino i y and Female Rep esen a ion Among Holde s o Ad anced Deg ees in Science and
Ma hema ics. Rocke elle Founda ion, New Yo k, NY.
Bo gna, C., Solga, H., P o sch, P., 2019. O e educa ion, labou ma ke dynamics, and economic down u n in Eu ope. Eu . Socio Re . 35, 116–132. h ps://doi.o g/
10.1093/ESR/JCY046.
B eda, T., G ene , J., Monne , M., Van E en e e, C., 2023. How e ec i e a e emale ole models in s ee ing gi ls owa ds STEM? E idence om F ench high schools.
Econ. J. 133, 1773–1809. h ps://doi.o g/10.1093/ej/uead019.
Cech, E.A., Blai -Loy, M., 2019. The changing ca ee ajec o ies o new pa en s in STEM. PNAS P oceedings o he Na ional Academy o Sciences o he Uni ed S a es
o Ame ica 116, 4182–4187. h ps://doi.o g/10.1073/pnas.1810862116.
Che yan, S., Ziegle , S.A., Mon oya, A.K., Jiang, L., 2017. Why a e some STEM ields mo e gende balanced han o he s? Psychol. Bull. 143, 1–35. h ps://doi.o g/
10.1037/BUL0000052.
Combe , B., Oesch, D., 2021. The social-o igin gap in uni e si y g adua ion by gende and immig an s a us: a coho analysis o Swi ze land. Longi udinal and Li e
Cou se S udies 12, 119–146. h ps://doi.o g/10.1332/175795920X16034769228656.
Compans, M.-C., Beaujouan, E., Sue o, C., 2023. T ansi ions o second bi h and bi h in e als in F ance and Spain: ime squeeze o social no ms? Compa a i e
Popula ion S udies 48. h ps://doi.o g/10.12765/CPOS-2023-13.
Co ell, S.J., 2001. Gende and he ca ee choice p ocess: he ole o biased sel -assessmen s. Am. J. Sociol. 10, 1691–1730. h ps://doi.o g/10.1086/321299.
C enshaw, K., 1991. Mapping he ma gins: in e sec ionali y, iden i y poli ics, and iolence agains women o colo . S an o d Law Re . 43, 1241–1299. h ps://doi.
o g/10.2307/1229039.
Del Bono, E., Mo ando, G., 2022. Fo some, luck ma e s mo e: he impac o he g ea ecession on he ea ly ca ee s o g adua es om di e en socio-economic
backg ounds. Ox . Econ. Pap. 74, 869–893. h ps://doi.o g/10.1093/OEP/GPAB053.
Delaney, J.M., De e eux, P.J., 2022. Gende di e ences in STEM pe sis ence a e g adua ion. Economica 89, 862–883. h ps://doi.o g/10.1111/ECCA.12437.
DiP e e, T.A., Buchmann, C., 2013. The Rise o Women: he G owing Gende Gap in Educa ion and wha i Means o Ame ican Schools. Russell Sage Founda ion, New
Yo k, NY.
England, P., 2010. The gende e olu ion: une en and s alled. Gend. Soc. 24, 149–166. h ps://doi.o g/10.1177/0891243210361475.
Es e e, A., Schwa z, C.R., Van Ba el, J., Pe manye , I., Klesmen , M., 2016. The end o hype gamy: global ends and implica ions. Popul. De . Re . 42, 615–625.
h ps://doi.o g/10.1111/pad .12012.
F iedmann, E., 2018. Inc easing women’s pa icipa ion in he STEM indus y. J. Soc. Ma ke . 8, 442–460. h ps://doi.o g/10.1108/JSOCM-12-2017-0086.
F iedmann, E., E a -T eis e , D., 2023. Gende bias in STEM hi ing: implici in-g oup gende a o i ism among men manage s. Gend. Soc. 37, 32–64. h ps://doi.o g/
10.1177/08912432221137910.
Funk, C., Pa ke , K., 2018. Women and Men in STEM O en a Odds o e Wo kplace Equi y. Pew Resea ch Cen e , Washing on, DC. h p://hdl.handle.ne /10919/
92671.
Ganzeboom, H.B.G., De G aa , T eiman, D.J., 1992. A s anda d in e na ional socio-economic index o occupa ional s a us. Social Science Resea ch 21 (1), 1–56.
h ps://doi.o g/10.1016/0049-089X(92)90017-B.
G ¨
a z, M., 2022. When less condi ioning p o ides be e es ima es: o e con ol and endogenous selec ion biases in esea ch on in e gene a ional mobili y. Qual.
Quan i y 56, 3769–3793. h ps://doi.o g/10.1007/s11135-021-01310-8.
Jacob, M., Iannelli, C., Du a, A., Smy h, E., 2020. Seconda y school subjec s and gende ed STEM en ollmen in highe educa ion in Ge many, I eland, and Sco land.
In . J. Comp. Sociol. 61, 59–78. h ps://doi.o g/10.1177/0020715220913043.
Jasko, K., Py kosz-Pacyna, J., Cza nek, G., Dukała, K., Szas ok, M., 2020. The STEM g adua e: immedia ely a e g adua ion, men and women al eady di e in job
ou comes, a ibu ions o success, and desi ed job cha ac e is ics. J. Soc. Issues 76, 512–542. h ps://doi.o g/10.1111/JOSI.12392.
M.T. Vald´
es and H. Solga
Social Science Resea ch 124 (2024) 103092
16
Kahn, L.B., 2010. The long- e m labo ma ke consequences o g adua ing om college in a bad economy. Lab. Econ. 17, 303–316. h ps://doi.o g/10.1016/J.
LABECO.2009.09.002.
Lee, S.W., Min, S., Mame ow, G.P., 2015. Pygmalion in he class oom and he home: expec a ion’s ole in he pipeline o STEMM. Teach. Coll. Rec. 117, 1–40.
Legewie, J., DiP e e, T.A., 2014. The high school en i onmen and he gende gap in science and enginee ing. Sociol. Educ. 87, 259–280. h ps://doi.o g/10.1177/
0038040714547770.
Lozano, M., Es e e, A., Boe ien, D., Mogi, R., Cui, Q., 2024. Lowes low e ili y in Spain: insigh s om he 2018 Spanish e ili y su ey. Demog . Res. 51, 625–636.
h ps://doi.o g/10.4054/DEMRES.2024.51.19.
Mann, A., DiP e e, T.A., 2013. T ends in gende seg ega ion in he choice o science and enginee ing majo s. Soc. Sci. Res. 42, 1519–1541. h ps://doi.o g/10.1016/j.
ss esea ch.2013.07.002.
McCall, L., 2005. The complexi y o in e sec ionali y. Signs: Jou nal o Women in Cul u e and Socie y 30, 1771–1800. h ps://doi.o g/10.1086/426800.
Mize, T.D., 2019. Bes p ac ices o es ima ing, in e p e ing, and p esen ing nonlinea in e ac ion e ec s. Sociological Science 6, 81–117. h ps://doi.o g/10.15195/
V6.A4.
Mon ma que e, C., Cannings, K., Mahse edjian, S., 2002. How do young people choose college majo s? Econ. Educ. Re . 21, 543–556. h ps://doi.o g/10.1016/
S0272-7757(01)00054-1.
Mo gan, S.L., Gelbgise , D., Weeden, K.A., 2013. Feeding he pipeline: gende , occupa ional plans, and college majo selec ion. Soc. Sci. Res. 42, 989–1005. h ps://
doi.o g/10.1016/j.ss esea ch.2013.03.008.
O eopoulos, P., on Wach e , T., Heisz, A., 2012. The sho - and long- e m ca ee e ec s o g adua ing in a ecession. Am. Econ. J. Appl. Econ. 4, 1–29. h ps://doi.
o g/10.1257/app.4.1.1.
O iz-Ge asi, L., 2020. Wha shape g ea expec a ions? Gende and social-o igin e ec s on expec a ion o uni e si y g adua ion. Res. Soc. S a i . Mobil. 69, 100527.
h ps://doi.o g/10.1016/j. ssm.2020.100527.
Raabe, I.J., Boda, Z., S ad eld, C., 2019. The social pipeline: how iend in luence and pee exposu e widen he STEM gende gap. Sociol. Educ. 92, 105–123. h ps://
doi.o g/10.1177/0038040718824095.
Reskin, B.F., Ross, P.A., 1990. Job Queues, Gende Queues. Explaining Women’s In oads in o Male Occupa ions. Temple Uni e si y P ess, Philadelphia , PA.
Schwand , H., on Wach e , T., 2019. Unlucky coho s: es ima ing he long- e m e ec s o en e ing he labo ma ke in a ecession in la ge c oss-sec ional da a se s.
J. Labo Econ. 37, S161–S198. h ps://doi.o g/10.1086/701046.
Schwe e , J., Ilg, L., 2023. Gende di e ences in he labou ma ke en y o STEM g adua es. Eu . J. High Educ. 13, 308–326. h ps://doi.o g/10.1080/
21568235.2021.2010226.
Spee , J.D., 2023. Bye bye Ms. Ame ican Sci: women and he leaky STEM pipeline. Econ. Educ. Re . 93, 102371. h ps://doi.o g/10.1016/J.
ECONEDUREV.2023.102371.
Thu ow, L.C., 1979. A job compe i ion model. In: Pio e, Michael J. (Ed.), Unemploymen and In la ion. Rou ledge, pp. 17–32. h ps://doi.o g/10.4324/
9781315084374-4.
Vald´
es, M.T., 2020. ¿Han cambiado las p e e encias o ma i as del alumnado espa˜
nol? Un an´
alisis de la e oluci´
on de las es adís icas de ma iculaci´
on en CFGM. In:
Vald´
es, Manuel T. (Ed.), Indicado es Comen ados Sob e El Es ado Del Sis ema Educa i o Espa˜
nol, Miguel ´
Angel Sancho Ga gallo, and Me cedes de Es eban Villa .
Fundaci´
on Ram´
on A eces, pp. 71–74.
an de Vleu en, M., S einme z, S., an de We ho s , H.G., 2018. Gende no ms and STEM: he impo ance o iends o s opping leakage om he STEM pipeline.
Educ. Res. E al. 24, 417–436. h ps://doi.o g/10.1080/13803611.2019.1589525.
Xu, Y., 2015. Focusing on women in STEM: a longi udinal examina ion o gende -based ea ning gap o college g adua es. J. High Educ. 86, 489–523. h ps://doi.o g/
10.1080/00221546.2015.11777373.
Za a , B., 2013. College majo choice and he gende gap. J. Hum. Resou . 48, 545–595. h ps://doi.o g/10.3368/JHR.48.3.545.
Zimme mann, B., Seile , S., 2019. The ela ionship be ween educa ional pa hways and occupa ional ou comes a he in e sec ion o gende and social o igin. Soc. Incl.
7, 79–94. h ps://doi.o g/10.17645/SI.V7I3.2035.
M.T. Vald´
es and H. Solga
Social Science Resea ch 124 (2024) 103092
17
