UNIVERSITY OF THE BASQUE COUNTRY - UPV/EHU
FACULTY OF ECONOMICS
Social class seg ega ion in
uppe -seconda y school choice in
Canada
A hesis submi ed by Ped o I abu u o he
Mas e in Economics: Empi ical Applica ions and Policies
Supe ised by:
Pe Ma iel
30 h o July, 2020
Bilbo
Abs ac
The aim o his wo k is o analyze how he spoken language a home in luences he
pa en al decision on school choice in Canada. The e is a as li e a u e on he ac o s a -
ec ing school choice, bu e y li le esea ch has been made on bilingual o mul ilingual
coun ies. We ake da a om he 2012 PISA ques ionnai es o es ima e he p obabili y o
choosing a de e mina e ype o school, disc imina ing by ins uc ional language and inanc-
ing. The esul s show clea ly ha he impac o bilingualism dilu es o he wise e y impo an
socio-economic a iables.
Keywo ds: School choice, bilingualism, mul inomial logi .
1
INDEX
1 In oduc ion 4
2 Li e a u e Re iew 6
2.1 The In luence o O icial Mino i y Languages . . . . . . . . . . . . . . . . . 6
2.2 P e ious S udies on School Choice . . . . . . . . . . . . . . . . . . . . . . . 8
2.3 A di e en app oach: bilingualism . . . . . . . . . . . . . . . . . . . . . . . . 10
3 The Da a Analysis 12
3.1 Desc ip ion o he Da abase . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
3.2 Di ision o he p o inces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
4 Me hodology 17
4.1 Theo e ical amewo k.............................. 17
4.2 In e p e a ion o he model: disc e e changes in p obabili ies . . . . . . . . 18
5 Empi ical Resul s 20
5.1 Ou models.................................... 20
5.2 Rele ance o he a iables: Wald es . . . . . . . . . . . . . . . . . . . . . 20
5.3 Disc e e changes in he P obabili ies . . . . . . . . . . . . . . . . . . . . . . 23
5.3.1 Fi s model: G oup I o p o inces .................... 23
5.3.2 Second model: G oup II o p o inces .................. 25
6 Concluding Rema ks & Fu he Resea ch 30
Re e ences 32
Appendix 36
2
Lis o Figu es
1Change in p obabili y wi h espec o he benchma k amily, G oup I . . . . 28
2Change in p obabili y wi h espec o he benchma k amily, G oup II . . . . 29
Lis o Tables
1Popula ion by mo he ongue and geog aphy, 2016 .............. 5
2Numbe o Su eyed S uden s ......................... 12
3Summa y s a is ics o he explana o y a iables ................ 13
4G oup I: Dis ibu ion o public, p i a e, English and F ench school. ..... 15
5G oup II: Dis ibu ion o public, p i a e, English and F ench school. ..... 16
6Wald es s o he signi icance o he explana o y a iables (I). ........ 21
7Wald es s o he signi icance o he explana o y a iables (II). ....... 22
8Mul inomial logi : i s model es ima ion .................... 36
9Bina y logi : second model es ima ion ..................... 37
3
1 In oduc ion
The goal o he hesis is o cas ligh on he ac o s a ec ing he pa en al school choice
in Canada, while aking in o accoun he ole o bilingualism. Canada is di ided in o 10
p o inces and 3 e i o ies. As hese las a e qui e small, hey we e no included in he
P og amme o In e na ional S uden Assessmen (PISA) s udy (OECD, 2012), which we
make use o ex ac he da a necessa y o ou analysis.
The e exis s a wide a ay o languages coexis ing in Canada, including many indige-
nous ones. In spi e o his, only 209,570 indi iduals speak abo iginal languages in p i-
a e (S a is ics Canada, 2016) and hus, we e no conside ed in ou analysis. The wo
main languages in Canada a e English and F ench, al hough he las one is no as widely
sp ead. Acco ding o S a is ics Canada (2020), abou 25% o Canadians use F ench as
hei home language, which is consis en wi h ou indings om he PISA 2012 da a (see
Table 3). Abou 25% o Canadians use F ench as hei home language (S a is ics Canada,
2020), which is consis en wi h ou indings om he PISA 2012 da a (see Table 3).
As Table 1 shows, he popula ion in Canada is pola ized, wi h many p o inces spa sely
popula ed and sha ing a low pe cen age o he o al. Mos popula ion o Canada is concen-
a ed in On a io and Quebec, each one being he main ep esen a i e o hei espec i e
o icial language: English and F ench. Quebec concen a es he majo i y o ancophones
along wi h New B unswick, being i he only e i o y ha has bo h languages as o icial.
Acco ding o he Sec ion Six een o he Canadian Cha e o Righ s and F eedoms
(Cons i u ion Ac , 1982), English and F ench a e he wo o icial languages o he S a e,
meaning ha bo h ha e equali y o s a us in he Pa liamen , cou s and Go e nmen o
Canada. A he p o ince le el his is di e en , as usually only one is conside ed he
o icial, al hough some o he p o ided public se ices a e a ailable in bo h o hem.
4
Gi en all o his, all schooling ac oss Canada is implemen ed in bo h languages, bu in
di e en p opo ions depending on he p o ince. The majo i y o i is publicly unded (see
Table 2). The educa ional sys em is di ided in o p ima y, seconda y and pos seconda y
educa ion, al hough i a ies on he e i o y, as i is unde exclusi e p o incial ju isdic-
ion (CICIC, 2020). Fo example, educa ion is compulso y up o he age o 17 yea s all
h oughou he coun y, excep o On a io, Mani oba and New B unswick, whe e i is 18
yea s.
In ega d o he 2012 PISA esul s, Canada s ands as one o he coun ies wi h he
highes sco es, being in he op 10 o he bes pe o mances in ma hema ics, eading
and sciences (B ochu, Deussing, Houme & Chuy, 2013). Among all he e i o ies in
Canada, Quebec was placed in he op-pe o ming pa icipan s globally in pape -based
ma hema ics. In sciences, s uden s in B i ish Columbia and Albe a pe o med be e han
he coun y’s a e age.
Table 1: Popula ion by mo he ongue and geog aphy, 2016
P o ince Popula ion O icial Lang. Speake s by Mo he ongue (%)
English F ench
On a io 13,312,870 English 68.2 % 4.0 %
Quebec 8,066,555 F ench 8.1 % 78.0 %
B i ish Columbia 4,598,415 English 70.0 % 1.4 %
Albe a 4,026,650 English 75.4 % 2.0 %
Mani oba 1,261,620 English 72.6 % 3.4 %
Saska chewan 1,083,240 English 83.2 % 1.5 %
No a Sco ia 912,300 English 91.4 % 3.4 %
New B unswick 736,285 Bo h 64.8 % 31.9 %
New F. and L. 515,680 English 97.1 % 0.5 %
P ince Edwa d Is. 141,020 English 91.1 % 3.6 %
Sou ce: S a is ics Canada.
5
2 Li e a u e Re iew
Du ing he las cen u y, many au ho s ha e s udied he a iables and ac o s a ec ing he
school choice, bu only ew ha e conside ed he coexis ence o mo e han one language
in hei models. Bilingualism in Canada is he key issue o unde s and he cu en si ua-
ion and how i has a ec ed i s educa ion sys em h oughou he cen u ies (Vaillancou ,
2012). In Canada, language and eligion ha e been always in insically ela ed, as 70% o
he F ench speake s in 1871 we e Ca holic. This ac led o a his o ical indi ec p o ec ion
o he language h ough eligion, which s a ed wi h he es ablishmen o he Cons i u-
ion Ac o 1867, whose sec ion 93, was in oduced as a measu e o p o ec ing hose
Ca holic Communi ies. This s a ed ha he cu en and u u e school sys ems o he eli-
gious mino i y (which could be ei he p o es an o ca holic), exis ing in a p o ince, should
be p o ec ed om any po en ial ha m om i s p o incial go e nmen ; and ga e au ho i y
o he Go e nmen o in e ene o p o ec such mino i y.
2.1 The In luence o O icial Mino i y Languages
The in oduc ion o he Cons i u ion Ac led o si ua ions such as he one in New B unswick,
which had a e y la ge pe cen age o Ca holics in 1871, and which nowadays sha e bo h
languages as o icial ones. In On a io, al hough ha ing a e y small linguis ic mino i y,
he e exis s an abo e a e age cons i u ional p o ec ion, as hal o he popula ion spoke
F ench a he ime. This p o ec ion was necessa y, as he e ha e been always a unwill-
ingness o English-Canadians o collabo a e in such pu pose, gi ing he ancophones in
Canada an “added incen i e o pu sue na ionalis ic language policies, wi h he possibili y
o some consequences ha English Canadians may no always app ecia e”, as B e on
s a ed in his 1978 pape , “Na ionalism and Language Policies”.
As his sys em did no conside he language bu he eligious igh s, in 1982 he “Cana-
dian Cha e o Righ s and F eedom” was edac ed, whe e he p o ec ion o mino i y lan-
guages educa ional igh s go in oduced in o he legal sys em. Villancou (2012), anal-
yses he measu emen o he cos s and bene i s o hese o icial languages policies in
6
Canada, and explains ha , al hough he coexis ence o hese wo languages do no ha e
a posi i e di ec e ec , i g ea ly inc eases he pe son wel a e, as he public se ices a e
a ailable in hei mo he ’s ongue.
Apa om his, Canadians may also ha e an inc eased u ili y om o he ypes o
bene i s, such as he ob ious cul u al alue inc emen o ha ing mo e han one o icial
languages, and indi ec mone a y e ec s. Ch is o ides and Swidinsky (2008), ound ha
men ou side o Quebec (whe e he majo i y is ancophone) wi h F ench language skills,
ended o be disp opo ionally ep esen ed in highe paying occupa ions. B e on (1998),
analysed census da a om 1971-1991, concluding ha he e we e signi ican e u ns o
bilingual language skills in he es ima ed wage equa ions. Being a bilingual coun y ha e
also i s de imen s, as he coexis ence o he wo languages can be e y expensi e ega d-
ing bu eauc acy, as a consequence o he mul ilingualism (Fid muc, 2011;Pons-Ridle &
Ridle , 1990).
The school sys em in Canada di e s om o he bilinguals’ coun ies, such as Belgium
o Swi ze land, whe e he ins uc ion language o he school depends on he egion, and
he e is no eal ee choice. Simila o Basque Coun y, in Canada he pa en s can choose
he language o educa ion o hei child en. This concep is explo ed by Pons-Ridle &
Ridle (1990), s a ing ha he es ablishmen o bilingualism in Canada, espec ing he
school choice, ollows he “Pe sonali y P inciple”. Acco ding o ha , he indi idual cus-
ome o Go e nmen se ices will be a ended in he language o his choice. The au ho s
also s a e ha e i o iali y would be a cheape al e na i e, in a simila way o Belgium and
Swi ze land, as i would sa e a lo o educa ion expendi u e.
Languages can be seen bo h as a means o communica ion and as pa o a cul u al
iden i y as es ablished by CASLT (2016). Fo ins ance, English is widely used in he busi-
ness as a communica ion ool, and no cul u al he i age is de i ed om i s use in such
en i onmen . Speci ically, CASTL (2016, pp. 62) s a es ha , “Language, howe e , ans-
mi s no only meaning in a s ic , e minological sense, bu also commonly held mo al
7
alues, judgmen s o o he social o poli ical e e ences”. I is qui e clea ha he anco-
phones in Canada ha e his o ically sha ed a common he i age, gi en how i was ied o
he Ca holic Communi y.
2.2 P e ious S udies on School Choice
Manski and Wise (1983) a e conside ed he i s au ho s o model choice in he educa-
ional en i onmen , bu now he li e a u e abou school choice is abundan . Fo example,
Nakhaie (2000) ound ha he pa en al educa ion and occupa ion ha e signi ican con-
sequences o hei child en’s educa ional a ainmen . The s udy used log odd a ios o
analyse wo na ional ep esen a i e samples o Canadians su eyed in 1985 and 1994.
Likewise, Bo ge s e al (1999), use an ex ended logi model o analyse da a om Veld-
ho en, in he Ne he lands, which has wo Ca holic, one P o es an and wo public schools.
They ound ha he eligion o he household, he size o he class oom, and he dis ance
o he schools, we e he mos ele an a iables. This is p o ed once again by M¨
ulle ,
Haase and Seidel (2012), by analyzing Ge man schools da a wi h a mul i a ia e analysis
o ind ha he e exis s an inc easingly compe i i eness among all schools, as a esul
om he possibili y o ee school choice. Thei app oach explici ly accoun s o spacial
subs i u ion, as he au ho s conside ed ha his pa e ns be ween school loca ions exis ,
se ing a di e en app oach om he p e ious li e a u e, which is usually ocused on acial
mix, ui ion ees, and a el- o-school dis ance.
Following he same s eps, H´
e u (1991) uses a mul i a ia e analysis model o analyse
ga he ed da a om 1,894 F ench speaking s uden s in Quebec, a g ade 5, 10 and 12,
whe e 12.7% assis ed p i a e schools, o ga he di e en conclusions, such as: “[...] pa -
en s’ assessmen o he ‘quali y’ o eaching in public and p i a e ins i u ions p obably co -
esponds o a pa icula eaching s yle, mo al alues and cul u al p e e ences a he han
me ely o school pe o mance measu ed by g ades and ailu es.” (H´
e u, 1991, pp. 496)
This clea ly shows ha he cul u e is a key ac o o pa en s when conside ing di e en
school ypes.
8
hei socio-demog aphic cha ac e is ics, we apply a MNLM in which ou di e en ype o
school ep esen he explained a iable. The codi ica ion o each ou come, as compu ed
in he model, is espec i ely: 1 o he Public English, 2 o he P i a e English, 3 o he
Public F ench, and 4 o he P i a e F ench.
In he second g oup, we disca ded indi iduals ha a ended o p i a e schools, as he
numbe o obse a ions o his ype o schools was insu icien o pe o m he analysis.
Acco dingly, in his second g oup we c ea ed a bina y logi model wi h only wo ou comes:
Public English o Public F ench. In each subse , we can ind a p o ince expec ed o ha e a
highe in luence on he language componen , gi en i s ancophone popula ion: Quebec in
he i s , and New B unswick in he second. In his second subse , he dependen a iable
will ake alue 1 i he ins uc ional language o he school is F ench, and 0 i English.
Table 4: G oup I: Dis ibu ion o public, p i a e, English and F ench school.
Quebec Mani oba Saska chewan B i ish Columbia
Type n % n % n % n %
Public English 1,165 34.1 1,481 80.5 1,588 90.7 1,409 86.4
P i a e English 293 8.6 94 5.1 74 4.2 73 4.5
Public F ench 1,713 50.2 204 11.1 32 1.8 92 5.6
P i a e F ench 243 7.1 60 3.3 56 3.2 57 3.5
To al P i a e 536 15.7 154 8.4 130 7.4 130 8.0
Sou ce: OECD PISA 2012, Canada.
15
Table 5: G oup II: Dis ibu ion o public, p i a e, English and F ench school.
No a Sco ia Albe a New B unswick On a io
Type n % n % n % n %
Public English 1,064 86.6 1,582 89.0 820 59.6 2,069 63.4
P i a e English - - - - 28 1.7 48 1.5
Public F ench 165 13.4 166 9.3 805 48.7 1,147 35.1
P i a e F ench - - 29 1.6 - - - -
To al P i a e - - 29 1.6 28 1.7 48 1.5
Sou ce: OECD PISA 2012, Canada.
16
4 Me hodology
In his sec ion, we will se he heo e ical ounda ion and he desc ip ion o he disc e e
choice models used in ou analysis: he mul inomial and bina y logi .
4.1 Theo e ical amewo k
As ou dependen a iable is a ca ego ical one wi h 2 o 4 ou comes - depending on
he model- we apply a disc e e choice model o analyse he likelihood o an indi idual
picking a speci ic al e na i e. Consequen ly, we model he pa en ’s school choice by a
MNLM in p o inces whe e ou ype o schools a e a ailable (English P i a e, English
Public, F ench P i a e, F ench Public) and bina y logi model in p o inces wi h only wo
al e na i es (English Public, F ench Public).
Fo he i s subse , we ha e es ima ed a MNLM, o he ollowing ou ou comes: Public
English,P i a e English,Public F ench and P i a e F ench, coded in he model as y=1,
2,3,4, espec i ely. Equa ions (1) and (2) ep esen he p obabili y o ou come min a
MNLM gi en socio-demog aphic a iables o he i- h amily (xi).βa e he co esponding
pa ame e s o be es ima ed. Fo he sake o iden i ica ion, in equa ion (1) he pa ame e s
o one ca ego y (1 in ou case) a e se o ze o.
P (yi=1|Xi) = 1
1+∑J
j=2exp(X0
iβj),(1)
P (yi=m|Xi) = exp(X0
iβm)
1+∑J
j=2exp(X0
iβj) o m <1.(2)
This equa ions will be he basis o he Maximum Likelihood es ima o , which will be
he me hod used o he es ima ion o he pa ame e s, o bo h models.
17
Fo he second subse , we ha e es ima ed a bina y logi , which only has wo ou comes
ins ead: Public English and Public F ench. The dependen a iable ywill ake alue 0 i he
ins uc ional language o he school is English, o 1 i i is F ench. Equa ion (3) p esen s
he o mula o he p obabili y o ou come one (Long, 1997).
P (y=1|xi) = F(x0
iβ) = exp(x0
iβ)
1+exp(x0
iβ),(3)
This equa ion ep esen s, he e o e, he p obabili y o choosing F ench as he ins uc-
ional language gi en socio-demog aphic cha ac e is ics o he i- h amily (xi), and βa e
he co esponding pa ame e s o be es ima ed. The e e ence le el is ou come ze o, ha
is, a Public English school.
4.2 In e p e a ion o he model: disc e e changes in p obabili ies
As Long (Long, 1997, pp. 164) s a es, he e a e many pa ame e s in a MNLM. Al hough
he s a is ical signi icance is impo an , he magni udes and he di ec ion o he e ec s
canno be in e p e ed di ec ly.The e a e se e al me hods o in e p e ing he esul s o he
es ima ion, such as p edic ed p obabili ies and pa ial changes. In his esea ch, we ha e
used disc e e change o measu e he changes in he p obabili ies. i can be applied o
con inuous and dummy a iables.
The p edic ed p obabili y o he dependen a iable y, being equal o he ou come m
gi en he ec o o co a ia es xiis:
c
P (y=m|xi) = exp(x0
iˆ
βm)
∑J
i=1exp(x0
iˆ
βj)(4)
18
By changing he xkpa ame e om xS o xE,ce e is pa ibus, he change in he p e-
dic ed p obabili y is:
∆
c
P (y=m|x)
∆xk
=
c
P (y=m|x,xk=xE)−
c
P (y=m|x,xk=xS).(5)
The magni ude o he change in p obabili ies depends on he amoun o change in xk,
he s a ing alue o xkand he o he a iables (Long, 1997). Mos o he ime, he o he
x−k a iables a e held a hei mean alues, wi h all he dummies a ei he 1 o 0. How
much should xk a y? I i is a dummy, he choice is p e y s aigh o wa d, as i is a disc e e
change om 0 o 1, o ice e sa. I on he con a y, he chosen a iable is con inuous,
he amoun o change depends on you analysis and on he na u e o i . In ou case we
use a s anda d de ia ion change bu his issue is u he discussed in sec ion sec ion 5.3.
Usually he bes a ailable ool o his ype o analysis is a disc e e change plo , as i
quickly summa izes all he in o ma ion. In his ype o g aph, he ho izon al axis ep esen s
he magni ude o he posi i e o nega i e change in he p obabili y o each ou come.
These a e ep esen ed by le e s. The e ical axis includes all he a iables o he model
(see ig. 1 and ig. 2). This analysis ool is sui able o bo h bina y and mul inomial logi s.
19
5 Empi ical Resul s
This sec ion con ains all o he ob ained esul s om ou analysis, om he de ini ion o
ou model o he in e p e a ion o ou es ima ions.
5.1 Ou models
As explained in sec ion 3.2, we ha e spli he da abase as well as ou analysis in o wo, as
he e we e wo egions clea ly de ined by i s p i a e schooling pe cen age. Gi en ha we
include dummy a iables o p o inces in he wo models he se o explana o y a iables
di e sligh ly. Ne e heless, he key a iables o he in e p e a ion amily weal h and he
spoken language in he household a e common in he wo models. Fo he es ima ion
o he model, we adop he Rp og amming language, mo e speci ically, he VGAM and
NNET packages. Mo eo e , as we canno in e p e he coe icien s di ec ly, due o he
non-linea na u e o he model, we in e p e ou esul s using disc e e changes in p oba-
bili ies, as explained in sec ion sec ion 4.2. Pos e io ly, bo h o mulas a e shown.
Hence, o he i s g oup o p o inces, we ha e de ined a MNLM as he ool o anal-
yse he a iables ha d i es Canadian pa en s o choose a speci ic ype o school, in a
bilingual en i onmen . Ou dependen a iable, ype o school, has ou ca ego ies: pub-
lic English, p i a e English, public F ench and p i a e F ench. As s a ed abo e, Table 3
p esen s he summa y s a is ics o he explana o y a iables, di ided in con inuous o
dummy ca ego ies. This a iables we e based on he p e ious esea ch done by Ma iel
and Vega-Bayo (2015,2018a,2019), ega ding he subjec . Fo he second g oup o
a iables we make use o a bina y logi , whe e we es ima e he p obabili y o selec ing a
F ench speaking school compa ed o a English one.
5.2 Rele ance o he a iables: Wald es
The es ima ion ou comes o bina y logi and MNLM a e p esen ed in Tables 8 and 9in
he Appendix. In o de o es he signi icance o all he pa ame e s associa ed wi h he
20
ou comes o bo h models, we ha e chosen he Wald es o analyze he signi icance o he
coe icien s. The null hypo hesis o his es s, is ha a speci ic explana o y a iable does
no ha e an e ec on he explained a iable. I is, he e o e, a join es ha in ol es h ee
pa ame e s in he MNLM. Table 6 and Table 7 show he esul s o his es . Mos o hem
a e signi ican a he 1% le el, al hough h ee a iables in he bina y logi a e no ele an
a 5% signi icance le el. We a e keeping hem as hey a e ele an o he i s model, o
compa ison pu poses: Siblings, he educa ional le el o he a he and wo-pa en amily.
Table 6: Wald es s o he signi icance o he explana o y a iables (I).
Va iable χ2s a is ic p- alue
Mo he no wo king 42.3 <0.01 ***
Two-pa en amily 13.3 <0.01 ***
Siblings 31.1 <0.01 ***
G andpa en s li ing wi h he amily 15.3 <0.01 ***
Cul u al possessions 128.9 <0.01 ***
Educ. le el mo he 21.7 <0.01 ***
Educ. le el a he 17.3 <0.01 ***
Family Weal h 108.4 <0.01 ***
Highes pa en al occup. s a us 122.7 <0.01 ***
Home language no English 1242.2 <0.01 ***
Quebec 284.3 <0.01 ***
Saske chwan 63.2 <0.01 ***
B i ish Columbia 58.7 <0.01 ***
***, **, * deno es signi icance a he 1%, 5% and 10% le el espec i ely.
21
Table 7: Wald es s o he signi icance o he explana o y a iables (II).
Va iable χ2s a is ic p- alue
Mo he no wo king 26.8 <0.01 ***
Two-pa en amily 13.3 <0.952
Siblings 0.1 0.335
G andpa en s li ing wi h he amily 35.9 <0.01 ***
Cul u al possessions 116.5 <0.01 ***
Educ. le el mo he 9.8 <0.01 ***
Educ. le el a he 1.9 0.167
Family Weal h 7.3 <0.01 ***
Highes pa en al occupp. s a us 36.8 <0.01 ***
Home language no English 1247.3 <0.01 ***
On a io 66.8 <0.01 ***
New B unswick 114.4 <0.01 ***
Albe a 41.4 <0.01 ***
***, **, * deno es signi icance a he 1%, 5% and 10% le el espec i ely.
22
5.3 Disc e e changes in he P obabili ies
This las sec ion o he empi ical esul s includes he o e iew o he disc e e changes in
he p obabili ies and hus, i ep esen s he key esul o he in e p e a ion o ou model.
Resul s include, o each g oup o p o inces ( he e a e wo, as seen in sec ion 3.2), an
analysis o each explana o y a iable. The heo y behind his me hodology is explained
in sec ion 4.2.
As p e iously men ioned, he e a e bo h dummy and con inuous a iables, and each
ype ollows a dis inc echnique when s udying i s in luence. In he i s case, he change
is in ui i e: we will swi ch he ini ial alue o he pa ame e om 0 o 1o ice e sa, de-
pending on he benchma k. Fo he la e , we will add and sub ac he s anda d de ia ion
o he mean. This is no a bi a y, he logic behind his is ha his shi ep esen s no a
uni , bu a ele an change o he a iable.
G oup I is he i s g oup o p o inces o which he in e p e a ion will be p esen ed, ha
is, hose ha showcased enough da a o p i a e schooling analysis. A e wa ds comes
G oup II, whe e we only disc imina ed he schools by language. The ou comes will be
illus a ed in ig. 1 and ig. 2. The ables con aining he coe icien s and he s d. e o s o
each g oup can be ound in he appendices.
5.3.1 Fi s model: G oup I o p o inces
This g oup is ep esen ed by Quebec, Mani oba, Saska chewan and B i ish Columbia.
Quebec is he second bigges p o ince, bo h inancial and popula ion wise (S a is ics
Canada, 2016). I is also he p incipal ancophone e i o y. B i ish Columbia and Mani-
oba a e no as big as he la e , bu hey a e no small ei he , being he hi d and he i h
p o inces by populace (see Table 1).
23
The se ing o he cha ac e is ics o he benchma k amily should no be an a bi a y
decision, as i can ease o make mo e di icul he subsequen in e p e a ion. Ha ing an-
alyzed he magni ude o he explana o y a iables, we se he benchma k cha ac e is ics
as ollows:
(a) Family li ing in Mani oba, as Quebec, Saska chewan and B i ish Columbia had a big-
ge di e en ial e ec s when in oduced as dummies ( he e o e hese h ee dummies
a e se o ze o).
(b) The home language is English, as i is he mos common ou come.
(c) The con inuous a iables a e se o he mean, hese includes he amily weal h, he
occupa ional s a us, and he educa ional le el o bo h mo he s and a he s.
(d) The emaining dummy a iables ha e been se o he mos common ou come, wi h
siblings and wi h no g andpa en s li ing in he amily uni .
Figu e 1 shows he changes in p obabili y o a ze o-one, o s anda d de ia ion change,
o each a iable depending on i s na u e. The Rlanguage did no ha e an implemen ed
lib a y o his pu pose (unlike S a a), so e e y hing in he p ocess is sel coded.
S a ing wi h he con inuous a iables, we can app ecia e ha ha hei e ec is gen-
e ally no big. The cul u al possessions,occupa ional s a us and he educa ional le el o
bo h pa en s show li le o none in luence o e he p obabili y changes. In spi e o his,
he amily weal h pa ame e gains weigh when we ake in accoun he di e en ial e ec s
o he p o inces. In bo h Saska chewan and B i ish Columbia he gain is no iceable, bu
no as huge as in Quebec. We can app ecia e ha he p obabili y o choosing a public
school (1and 3) dec eases, implying ha o iche amilies, p i a e schools a e p e e ed.
Fo Saska chewan, public F ench schools ha e a highe p obabili y o being chosen, his
could come because o he in e es o iche households o hei child en speaking mo e
24
O e all, he es ima ion o he pa ame e s om bo h models a e obus . The Wald es
con i ms ha all a iables a e ele an ac oss bo h models. The p obabili y o choosing
schools whe e F ench is he ins uc ional language is a ec ed he mos by hose a iables
ela ed o he home language. This p o es ha bilingualism is he main d i e o school
choice. Among he p o inces, Quebec and New B unswick unco e he g ea e in luence
o F ench when compa ed o o he s. Besides, he second mos impo an g oup o pa-
ame e s a e hose ela ed o amily weal h. Once again, Quebec and On a io showcase
bigge di e en ial e ec s, which is logical, gi en ha hey a e among he iches e i o ies
in Canada (S a is ics Canada, 2016). Ha ing analyzed all o he a iables, a ew s and
ou o hei lack o ele ance, such as siblings and he educa ional le el o bo h pa en s,
which a e ele an in o he pape s, such as in he Ma iel e al s udy o Japan (2019).
In conclusion, simila o some s udies in Basque Coun y (Ma iel e al Spanish s udy,
2015), we ha e ga he ed o Canada compelling e idence o bilingualism dilu ing he
socio-economic seg ega ion p e iously ound in he school choice. I would be in e es -
ing o analyze o he bilingual o mul ilingual coun ies. None heless, we need o ake in o
accoun ha no many coun ies boas om he un ema kable si ua ion o Canada o he
Basque Coun y. In o de o p ope ly unde s and he bilingualism e ec on he school
choice, he ins uc ional language mus no be es ic ed by e i o y (e.g. Belgium, whe e
he language depends on he a ea in which you s udy).
The e a e 55 o icially bilingual coun ies in he wo ld (Uni e si y o O awa, 2020), so
he e is plen y o po en ial esea ch ahead. Rega ding Canada, i would also be allu ing
o use o he da abases o es ou indings. S a is ics Canada, is he o icial ins i u e o
s a is ics o he s a e, and i con ains enough s udies o ca y a simila analysis o he
one we ha e conduc ed, such as he Classi ica ion o Ins uc ional P og ams (S a is ics
Canada, 2016). S udying school choice om ano he poin o iew could be in e es ing:
as a comp omise be ween ul illing he p e e ences o he households (people will end
o seg ega e on hei own), o making an e o o e adica e he exis ing seg ega ion on a
social, acial, o cul u al le el.
31
Re e ences
[1] Bi ulco, R., & Ladd, H. F. (2006). School choice, acial seg ega ion, and es -sco e
gaps: E idence om No h Ca olina’s cha e school p og am*. Jou nal o Policy Anal-
ysis and Managemen , 26(1), 31-56.
[2] Bo ge s, A., Oppewal, H., Ponj´
e, M., & Timme mans, H. (1999). Assessing he Im-
pac o School Ma ke ing: Conjoin Choice Expe imen s Inco po a ing A ailabili y and
Subs i u ion E ec s. En i onmen and Planning A: Economy and Space, 31(11), 1949-
1964.
[3] B e on, A. (1978). Na ionalism and Language Policies. The Canadian Jou nal o Eco-
nomics, 11(4), 656.
[4] B e on, A. (1998). Economic app oaches o language and o bilingualism. O awa:
Canadian He i age.
[5] B ochu, P., Deussing, M., Houme, K. & Chuy, M. (2013). Measu ing up: Canadian e-
sul s o he OECD PISA s udy: The pe o mance o Canada’s you h in science, eading
and ma hema ics: 2012 i s esul s o Canadians aged 15. Council o Minis e s o Ed-
uca ion, Canada.
[6] Calsamiglia, C., Hae inge , G., & Klijn, F. (2010). Cons ained School Choice: An
Expe imen al S udy. Ame ican Economic Re iew, 100(4), 1860-1874.
[7] Ch is o ides, L. N., & Swidinsky, R. (2010). The Economic Re u ns o he Knowledge
and Use o a Second O icial Language: English in Quebec and F ench in he Res -o -
Canada. Canadian Public Policy, 36(2), 137-158.
[8] CICIC, 2020. Minis ies/depa men s esponsible o educa ion in Canada. Re ie ed
om he CICIC webpage.
32
[9] Cons i u ion Ac , 1982. Re ie ed om he Go e nmen o Canada webpage.
[10] Economic Ad an ages o Bilingualism. Li e a u e Re iew (2016). Canadian Associa-
ion o Second Language Teache s.
[11] Fa e, L., O ega, F., & Tanaka, R. (2018). Immig a ion and he public–p i a e school
choice. Labou Economics, 51, 184-201.
[12] Fid muc, J. (2011). The Economics o Mul ilingualism in he EU. Resea ch Handbook
on he Economics o Eu opean Union Law.
[13] F eyne , N., & Cl´
emen , R. (2015). Bilingualism in mino i y se ings in Canada: In e-
g a ion o assimila ion? In e na ional Jou nal o In e cul u al Rela ions, 46, 55-72.
[14] Ganzeboom, H. B., G aa , P. M., & 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.
[15] H´
e u, C. (1991). Educa ional e o m and adi ion: The case o public and p i a e high
schools in Quebec. In e na ional Jou nal o Educa ional Resea ch, 15(5), 483-497.
[16] Lank o d, R., Lee, E., & Wycko , J. (1995). An Analysis o Elemen a y and Seconda y
School Choice. Jou nal o U ban Economics, 38(2), 236-251.
[17] Long, J. S. (1997). Reg ession Models o Ca ego ical and Limi ed. Dependen Va i-
ables, SAGE Publica ions, Thousand Oaks.
[18] Maddaus, J. (1990). Pa en al Choice o School: Wha Pa en s Think and Do. Re iew
o Resea ch in Educa ion, 16, 267.
[19] Manski, C. F., & Wise, D. A. (1983). College choice in Ame ica. Camb idge, Mass.
u.a.: Ha a d Uni . P .
33
[20] M¨
ulle , S., Haase, K., & Seidel, F. (2012). Exposing Unobse ed Spa ial Simila i y:
E idence om Ge man School Choice Da a. Geog aphical Analysis, 44(1), 65-86.
[21] Nakhaie, M. (2000). Social o igins and educa ional a ainmen in Canada: 1985 and
1994. Re iew o Radical Poli ical Economics, 32(4), 577-609.
[22] OECD. (2012). P og amme o In e na ional S uden Assessmen (PISA).
[23] Pons-Ridle , S., & Ridle , N. B. (1989). The e i o ial concep o o icial bilingualism;
A cheape al e na i e o Canada? Language Sciences, 11(2), 147-158.
[24] P ie o, L. M., Ague o-Val e de, J., Za a e-Ca denas, G., & Maa se een, M. V. (2018).
Pa en al p e e ences in he choice o a special y school. Jou nal o School Choice,
13(2), 198-227.
[25] Ripley, B. & Venables, W. (2020). Feed-Fo wa d Neu al Ne wo ks and Mul inomial
Log-Linea Models. R package, e sion 7.3-14.
[26] S a is ics Canada (2020). Table 15-10-0003-01. Popula ion by mo he ongue and
geog aphy, 1951 o 2016. Re ie ed om he S a is ics Canada webpage.
[27] S a is ics Canada (2016). Abo iginal Popula ion P o ile. 2016 Census. Re ie ed om
he S a is ics Canada webpage.
[28] S a is ics Canada (2016). Language Highligh Table. 2016 Census. Re ie ed om
he S a is ics Canada webpage.
[29] S a is ics Canada (2016). Table: 36-10-0487-01. G oss domes ic p oduc (GDP) a
basic p ices, by sec o and indus y, p o incial and e i o ial . Re ie ed om he S a is-
ics Canada webpage.
34
[30] S a is ics Canada (2016). Classi ica ion o Ins uc ional P og ams (CIP) Canada
2016. Re ie ed om he S a is ics Canada webpage.
[31] Uni e si y o O awa (2020). Compendium o Language Managemen in Canada
(CLMC). Re ie ed om he Uni e si y o O awa webpage.
[32] Vaillancou , F. (2012). O icial language policies o he Canadian p o inces: Cos s
and bene i s in 2006. Vancou e , BC: F ase Ins i u e.
[33] Vega-Bayo, A., & Ma iel, P. (2015). School choice in he Basque Coun y: Public,
go e nmen -dependen and p i a e schools wi h di e en languages o ins uc ion. In-
e na ional Jou nal o Educa ional Resea ch, 74, 13-25.
[34] Vega-Bayo, A., & Ma iel, P. (2018). School Choice ac oss Di e en Regions o Spain.
Re is a Hacienda P´
ublica Espa˜
nola, 227(4), 11-36.
[35] Vega-Bayo, A., & Ma iel, P. (2018). A Disc e e Choice Expe imen Applica ion o
School Choice. Re is a Hacienda P´
ublica Espa˜
nola, 230(3), 41-62.
[36] Vega-Bayo, A., Sanko, N., & Ma iel, P. (2019). Social class seg ega ion in uppe -
seconda y school choice in Japan. Uni e sidad del Pa´
ıs Vasco, Depa amen o de
Econom´
ıa Aplicada III.
[37] Yee, T. & Mole , C. (2020). Vec o Gene alized Linea and Addi i e Models. R pack-
age, e sion 1.1-3.
35
Appendix I
Table 8: Mul inomial logi : i s model es ima ion
Dependen a iable:
P i a e English Public F ench P i a e F ench
(1) (2) (3)
Mo he no wo king 0.439∗∗∗ −0.292∗∗∗ 0.226∗∗
(0.100) (0.083) (0.109)
Two-pa en amily 0.268∗0.108 0.617∗∗∗
(0.155) (0.107) (0.179)
Siblings −0.471∗∗∗ 0.133 0.738∗∗∗
(0.135) (0.108) (0.190)
G andpa en s 0.187 −0.560∗∗∗ −0.589∗∗
(0.207) (0.170) (0.278)
Cul u al Possessions 0.483∗∗∗ −0.217∗∗∗ 0.229∗∗∗
(0.057) (0.043) (0.059)
Mo he Educ. le el 0.219∗∗∗ 0.035 0.127∗∗
(0.052) (0.033) (0.051)
Fa he Educ. le el 0.149∗∗∗ −0.029 0.055
(0.042) (0.027) (0.042)
Family Weal h 0.522∗∗∗ 0.330∗∗∗ −0.104
(0.113) (0.087) (0.148)
Occup. S a us 0.031∗∗∗ 0.010∗∗∗ 0.016∗∗∗
(0.003) (0.002) (0.003)
Home Language no English 0.231 2.413∗∗∗ 0.203
(0.338) (0.175) (0.383)
Quebec 1.453∗∗∗ 0.266 1.142∗∗∗
(0.204) (0.201) (0.208)
Saske chwan −0.223 −1.678∗∗∗ −0.099
(0.267) (0.331) (0.263)
B. Columbia −0.435 0.109 −0.439
(0.267) (0.202) (0.278)
Home Language no English: Quebec −0.668∗1.886∗∗∗ 1.093∗∗∗
(0.377) (0.239) (0.411)
Home Language no English: Saska chewan 0.323 0.145 −0.628
(0.575) (0.432) (0.830)
Home Language no English: B. Columbia 0.289 −1.356∗∗∗ 0.265
(0.444) (0.287) (0.509)
Family Weal h: Quebec −0.064 −0.752∗∗∗ −0.173
(0.132) (0.102) (0.167)
Family Weal h: Saska chewan −0.153 −0.129 0.070
(0.165) (0.198) (0.206)
Family Weal h: B.Columbia −0.160 −0.709∗∗∗ 0.417∗∗
(0.166) (0.147) (0.200)
Cons an −7.067∗∗∗ −3.636∗∗∗ −6.228∗∗∗
(0.406) (0.252) (0.403)
Akaike In . C i . 10,458.460 10,458.460 10,458.460
No e: ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01
Benchma k: Public English schools.
36
Appendix II
Table 9: Bina y logi : second model es ima ion
Dependen a iable:
Public F ench
Mo he no wo king −0.371∗∗∗
(0.069)
Two-pa en amily 0.025
(0.087)
Siblings −0.052
(0.084)
G andpa en s −0.752∗∗∗
(0.133)
Cul u al Possessions −0.356∗∗∗
(0.034)
Mo he Educ. le el 0.090∗∗∗
(0.028)
Fa he Educ. le el −0.022
(0.023)
Family Weal h 0.342∗∗∗
(0.113)
Occup. S a us 0.011∗∗∗
(0.002)
Home Language no English 2.744∗∗∗
(0.227)
Albe a −0.425∗∗
(0.187)
N. B unswick 0.333∗∗
(0.170)
On a io 1.411∗∗∗
(0.134)
Home Language no English: Albe a −0.778∗∗∗
(0.286)
Home Language no English: N. B unswick 2.225∗∗∗
(0.295)
Home Language no English: On a io −1.408∗∗∗
(0.241)
Family Weal h: Albe a −0.299∗∗
(0.142)
Family Weal h: N. B unswick −0.407∗∗∗
(0.152)
Family Weal h: On a io −0.288∗∗
(0.119)
Cons an −3.388∗∗∗
(0.216)
Akaike In . C i . 6,444.261
No e: ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01
Benchma k: Public English schools.
37
