1
1
2
3
4
This is he pos p in e sion o he a icle, which has been published in
Science o he To al En i onmen 2018, 624, 1504–1512. The inal
publica ion is a ailable a h ps://doi.o g/10.1016/j.sci o en .2017.12.126
5
6
7 Running head: IAQ and social clima e in schools
8
9
10
Indoo ai p oblems and he pe cei ed social clima e in schools: a mul ile el s uc u al equa ion
11
analysis
12
13
Ee ika Finell*
14
Asko Tol anen
15
Ulla Ha e inen-Shaughnessy
16
Seppo Laaksonen
17
Saka i Ka onen
18
Reijo Sund
19
Pauliina Luopa
20
Juha Pekkanen
21
Timo S åhl
22
23
*Reques s o ep in s should be add essed o Ee ika Finell, Facul y o Social Sciences, Linna, 33014
24
Uni e si y o Tampe e, Tampe e, Finland. E-mail: [email p o ec ed]
25
26
27
2
O he a ilia ions:
1
2
Asko Tol anen: Me hodology Cen e o Human Sciences, P.O. Box 35, 40014 Uni e si y o
3
Jy äskylä, Finland
4
Ulla Ha e inen-Shaughnessy: Depa men o Heal h P o ec ion, Na ional Ins i u e o Heal h and
5
Wel a e, P.O. Box 95, 70701 Kuopio, Finland
6
Seppo Laaksonen: Depa men o Social Resea ch, P.O. Box 68, 00014 Uni e si y o Helsinki,
7
Finland
8
Saka i Ka onen: Depa men o Heal h and Social Ca e Sys ems, Na ional Ins i u e o Heal h and
9
Wel a e, P.O. Box 30, 00271 Helsinki, Finland
10
Reijo Sund: 1) Cen e o Resea ch Me hods, Depa men o Social Resea ch, P.O. Box 18, 00014
11
Uni e si y o Helsinki, Finland, 2) Ins i u e o Clinical Medicine, Uni e si y o Eas e n Finland,
12
Kuopio, Finland
13
Pauliina Luopa: Depa men o Wel a e, Na ional Ins i u e o Heal h and Wel a e, P.O. Box 30,
14
00271 Helsinki, Finland
15
Juha Pekkanen: 1) Depa men o Public Heal h, P.O. Box 20, 00014 Uni e si y o Helsinki, Finland,
16
2) Depa men o Heal h P o ec ion, Na ional Ins i u e o Heal h and Wel a e, Finland
17
Timo S åhl: Depa men o Wel a e, Na ional Ins i u e o Heal h and Wel a e, Bioka u 10,
18
33520 Tampe e, Finland
19
20
21
3
Abs ac
1
Indoo ai p oblems in schools appea o in luence lea ning ou comes and absence a es.
2
Howe e , p e ious esea ch has no in es iga ed whe he indoo ai p oblems in luence he social
3
clima e o schools. The e o e, we s udied whe he indoo ai p oblems obse ed in schools
4
associa e wi h s uden s’ pe cep ions o he eache -s uden ela ionship and class spi i . The
5
na ionwide sample o Finnish schools (N = 194 schools/27 153 s uden s) was analyzed using
6
mul ile el s uc u al equa ion modeling. Da a on he schools’ social clima e collec ed om
7
s uden s we e me ged wi h independen ly collec ed da a on obse ed indoo ai p oblems om
8
school p incipals. We ound ha he eache -s uden ela ionship was epo ed o be wo se in
9
schools wi h obse ed indoo ai p oblems compa ed o hose wi hou obse ed indoo ai
10
p oblems. Fu he mo e, he epo ed class spi i was wo se in schools wi h obse ed indoo ai
11
p oblems, bu only among s uden s wi h a high g ade poin a e age. Ou indings indica e ha
12
indoo ai p oblems may a ec he s uden -pe cei ed social clima e.
13
Keywo ds (6): indoo ai quali y, social clima e, eache -s uden ela ionship, psychosocial e ec ,
14
lowe seconda y school, class spi i
15
16
17
18
19
20
21
22
23
4
1. In oduc ion
1
P oblems in schools’ indoo ai quali y (IAQ) a e widesp ead. Fo example, some 24% o schools in
2
Finland, 20% in he Ne he lands, and 40% Spain we e es ima ed o ha e mois u e p oblems, such
3
as dampness o mold. 1 Fu he mo e, many s udies ha e ound ha a la ge p opo ion o schools
4
a e inadequa ely en ila ed. 2–4 Such p oblems can inc ease he isk o s uden s’ heal h, 5,6
5
absence a es, 7,8 and dec eased academic pe o mance (e.g., s anda dized es sco es). 3,7,9
6
Despi e hese indings, li le is known abou whe he and how poo IAQ associa es wi h he
7
s uden s’ pe cep ion o social clima e in schools. The social clima e e lec s, o ins ance, he goals,
8
no ms, in e pe sonal ela ionships and eaching p ac ices, o ganiza ional pa e ns, and school
9
acili ies p esen in a speci ic school. 10,11 Poo s uden -pe cei ed social clima e associa es wi h
10
s uden s’ dec eased well-being and academic pe o mance, 10–13 lowe sel -es eem 14 and
11
p oblems in hei school adjus men 15,16, o example. Gi en ha he pe cei ed social clima e
12
plays an impo an ole in s uden s’ psychosocial well-being and lea ning, i is essen ial o s udy
13
how schools’ IAQ p oblems ela e o i . The need o s udy he e ec s o IAQ on he social clima e
14
in schools is acknowledged in he “indoo ai esea ch” li e a u e as well. 17
15
16
1.1. S uden -pe cei ed social clima e and indoo ai p oblems in schools
17
Al hough he e is no p e ious esea ch assessing whe he and how schools’ IAQ p oblems ela e o
18
he s uden -pe cei ed social clima e, some indi ec e idence sugges s ha he e may be an
19
associa ion be ween hem. P e ious esea ch has associa ed schools’ IAQ p oblems wi h inc eased
20
a es o eache s’ sick lea e and dec eased well-being, 18, 19 and i may induce in e pe sonal
21
con lic s as well as anxie y and ea .20 Fu he mo e, i has been shown ha eache s, who pe cei e
22
hei school’s physical en i onmen unsa is ac o y, epo mo e nega i e a i udes and lowe
23
5
mo al owa d hei wo k and no ice mo e p oblems in he school’s social clima e han eache s
1
who pe cei e he en i onmen as sa is ac o y. 21–23 Finally, also p incipals’ pe cep ions o hei
2
school’s physical en i onmen has been associa ed wi h hei unde s anding o he school’s abili y
3
o deli e ins uc ion.24
4
Howe e , i is unlikely ha IAQ p oblems in luence only school pe sonnel. Fi s , gi en ha
5
school’s o ganiza ional cul u e and eache s’ well-being and a i udes ela e o he s uden -
6
pe cei ed social clima e, 25–28 i is plausible ha also s uden s in schools wi h IAQ p oblems epo
7
lowe social clima e han s uden s in schools wi hou such p oblems. Fu he mo e, p e ious
8
esea ch indica es ha also s uden s’ pe cep ions o hei school’s physical en i onmen associa e
9
wi h how hey pe cei e he school’s social clima e. 11,29 Finally, s uden s’ absence a es ha e been
10
ound o be highe in schools wi h poo IAQ han in schools wi hou such p oblems, 8 which may
11
also e lec issues in he social clima e, since he social clima e associa es wi h school
12
absen eeism. 30 Based on he e idence, i appea s ha schools’ IAQ p oblems may ela e o ac o s
13
which cons i u e he s uden -pe cei ed social clima e, such as pe cei ed social suppo , discipline,
14
and pee ela ionships, in many ways.
15
16
1.2. The p esen s udy
17
In his s udy, we ocus on wo highly ele an ac o s o he s uden -pe cei ed social clima e:
18
pe cei ed quali y o eache -s uden ela ionship and pe cei ed class spi i (i.e., he class oom’s
19
emo ional clima e). 10,31 These ac o s a e e y in luen ial o s uden s’ academic pe o mance,
20
school engagemen , heal h, and heal h- ela ed beha io s, o example. 32–34 The e o e, i is
21
impo an o know whe he and how IAQ p oblems in schools ela e o hese ac o s. Ou i s
22
esea ch ques ion asks whe he he pe cei ed eache -s uden ela ionship and class spi i among
23

6
lowe seconda y school s uden s di e be ween schools wi h obse ed IAQ p oblems and schools
1
wi hou such p oblems. Based on he li e a u e ci ed abo e, 18,20,22,26 we hypo hesize ha s uden s
2
in schools wi h IAQ p oblems epo wo se pe cei ed quali y o he eache -s uden ela ionship
3
and class spi i han s uden s in schools wi hou such p oblems (H1).
4
Ou second esea ch ques ion asks whe he he associa ions be ween s uden -le el
5
cha ac e is ics (e.g., gende ) and he s uden -pe cei ed social clima e a e di e en in schools wi h
6
and wi hou IAQ p oblems. The need o s udy his issue is based on p e ious esea ch showing
7
ha s uden s’ pe cep ions o hei school’s social clima e is associa ed wi h many s uden -le el
8
cha ac e is ics, such as gende , socioeconomic s a us, e hnici y, social suppo , and academic
9
pe o mance; 13,35,36 di e en s uden s pe cei e he social clima e o hei school di e en ly. Gi en
10
ha no only s uden cha ac e is ics bu also con ex ual ac o s can ha e e ec s on he s uden -
11
pe cei ed social clima e, 37 we also es ed whe he he associa ion be ween school-le el
12
cha ac e is ics (e.g., school size) and he s uden -pe cei ed social clima e a e di e en in hese
13
wo school con ex s. Analyzing hese in e ac ions was conside ed impo an since is p o ides
14
in o ma ion on whe he he e a e ce ain ypes o s uden s o schools ha a e especially
15
suscep ible o epo poo eache -s uden ela ionship o poo class spi i when acing IAQ
16
p oblems. Gi en ha he e a e no p e ious s udies conside ing his issue, no hypo hesis is posed.
17
18
19
7
2. Ma e ial and me hods
1
2.1. Da a and pa icipan s
2
The da a we e ob ained om wo sou ces: a) he School Heal h P omo ion S udy (SHP) 2013,
3
ocusing on comp ehensi e school s uden s in g ades eigh and nine (14–16 yea s old) and b) he
4
Benchma king Sys em o Heal h P omo ion Capaci y Building (BSHPCB) da a collec ions om
5
comp ehensi e schools in 2013 ( o mo e de ails, see Finell e al., 2017). 38 The ques ionnai e o
6
BSHPCB is usually illed in by he school’s p incipals. Two a iables om BSHPCB we e used in he
7
analysis. The i s a iable measu ed i and when he iennial inspec ion equi ed by Heal h Ca e
8
Ac 1326/2010 had been ca ied ou in he school. This egula o y inspec ion is done in co-
9
ope a ion wi h he ep esen a i es o he school (e.g., a p incipal), he heal h au ho i ies, and
10
au ho i ies esponsible o cons uc ion and main enance o he school building. I includes a
11
numbe o ac o s (no jus IAQ ela ed) which can in luence he well-being o he school
12
communi y. The inspec ion o he indoo en i onmen includes e iew o exis ing documen s, such
13
as esul s om possible ques ionnai es on IAQ and heal h, IAQ measu emen s, and o he building
14
ela ed epo s as well as a building walk h ough ocusing on a eas whe e p oblems ha e been
15
ound o suspec ed (see Finell e al., 2017; Hie anen-Pel ola and Ko pilah i, 2015). 38, 39
16
The second a iable measu ed whe he o no he e we e biological exposu es obse ed in
17
he school du ing his egula o y inspec ion (see 2.2.2.). BSHPCB ques ionnai e also asks abou
18
physical and chemical exposu es. Howe e , we ocus only on he i em measu ing biological
19
exposu es he e because hese issues, especially dampness and mould (“mouldy schools”) ha e been unde
20
in ensi e public conce n since 1990’s in Finland. In he u u e, i would be in e es ing o pe o m compa a i e
21
analyses on he o he wo i ems o es whe he he associa ions wi h schools’ social clima e would be
22
di e en . Fo he analysis, we included only hose schools whe e a) he inspec ion was ca ied ou in
23
8
2012 o 2013 and b) whe e 1) biological exposu es we e iden i ied du ing he inspec ion bu he
1
p oblems had no been emedia ed o 2) no indoo ai p oblems we e iden i ied. This means ha
2
73 schools (N=10181 s uden s, 27%) we e excluded om he g oup o schools whe e he egula o y
3
iennial inspec ion was ca ied ou in 2012 and 2013. In hese schools a) he biological exposu es
4
we e no included in he inspec ion, b) p oblems we e ound and hey we e al eady co ec ed o c)
5
no in o ma ion we e p o ided (see Measu es). Six een schools wi h ewe han 10 s uden s we e
6
also excluded om he analysis ( o al o 66 s uden s, 0.2%). In addi ion, esponden s who epo ed
7
hei age as a leas wo yea s younge o h ee yea s olde han he a e age age o hei classma es
8
we e excluded om he analysis (n= 64, 0.2%). Two schools wi h 16 and 12 s uden s we e excluded,
9
because hey we e conside ed o be school-le el ou lie s (in he eache -s uden ela ionship). The
10
inal da a consis o 27153 s uden s om 194 schools. Abou 63 pe cen o he s uden s we e om
11
schools wi hou obse ed indoo ai p oblems (17127 s uden s om 127 schools).
12
13
2.2. Measu es
14
2.2.1. Ou come a iables
15
The pe cei ed quali y o he eache -s uden ela ionship was measu ed by h ee i ems: “Teache s
16
a e in e es ed in how I am doing”, “Teache s ea us, he s uden s, ai ly”, and “The opinions o
17
s uden s a e aken in o conside a ion in he de elopmen o school wo k”. The esponse scale was
18
1 = ully ag ee, 2 = ag ee, 3 = disag ee, and 4 = ully disag ee. These i ems ha e been used in many
19
p e ious s udies as indica o s o he eache -s uden ela ionship. 27,40,41 We modeled eache -
20
s uden ela ionship as a la en ac o wi h hese h ee i ems. The s anda dized loadings o he
21
con i ma o y ac o analysis a e p esen ed in Table 2. The i indica o s a e epo ed in Resul
22
sec ion (see also Suppo ing in o ma ion). The C onbach alpha was 0.63.
23
9
The class spi i was measu ed by h ee i ems: “The s uden s in my class ge along well”, “The
1
class oom discipline in my class is good” and “The mood in ou class is such ha I da e o exp ess
2
my opinion eely”. The esponse scale was he same as abo e. These i ems ha e been used
3
p e iously by Ka onen and colleagues 40 as an indica o o class spi i . As abo e, we modeled class
4
spi i as a la en ac o . The s anda dized loadings a e p esen ed in Table 2. The i indica o s a e
5
epo ed in Resul sec ion (see also Suppo ing in o ma ion). The C onbach alpha was 0.67. The
6
da a sou ce o he ou come a iables was he SHP.
7
8
2.2.2. P edic o
9
The BSHPCB su ey i em “We e he ollowing issues e alua ed in he mos ecen inspec ion:
10
Exposu e o biological agen s (indoo ai , mold, e c.)” had he ollowing esponse op ions: no da a
11
a ailable; no included in he inspec ion; inspec ed, no de iciencies de ec ed; inspec ed,
12
de iciencies de ec ed bu no ye co ec ed; inspec ed, de iciencies de ec ed and co ec ed. In his
13
s udy, we ocus only on he ollowing op ions: 0 = inspec ed, no de iciencies de ec ed and 1 =
14
inspec ed, de iciencies de ec ed bu no ye co ec ed. The esul ing a iable is e e ed o as
15
“obse ed IAQ p oblems”.
16
17
2.2.3. Backg ound a iables
18
Gende (0 = boy, 1 = gi l), age, pe cei ed heal h, pa en al suppo , and educa ion o he mo he
19
we e included only a he s uden -le el (see S a is ical analyses and Figu es 1 and 2 below).
20
Pe cei ed heal h was measu ed by he i em: “How is you heal h in gene al?” The esponse scale
21
was 1 = e y good, 2 = ai ly good, 3 = mode a e, and 4 = ai ly o e y poo The scale was
22
ca ego ized in o wo ca ego ies (0 = e y good, 1 = no e y good), because only a small
23
p opo ion epo ed mode a e o bad heal h. The simila kind o ca ego iza ion has been used in
24
16
3.1. The associa ion be ween obse ed IAQ p oblems and he s uden -pe cei ed schools’ social
1
clima e
2
Fi s , we es ed he i o each measu emen model (see Figu es 1 and 2) by ca ying ou a wo-
3
le el con i ma o y ac o analysis o each a iable sepa a ely. In o de o do ha we ixed he
4
loadings equally a he s uden and school-le els. 47 The model i s we e good bo h o he
5
eache -s uden ela ionship (RMSEA = 0.01, CFI = 1.00, SRMR: s uden -le el = 0.00, SRMR: school-
6
le el = 0.08) and class spi i (RMSEA = 0.01, CFI = 1.00, SRMR: s uden -le el = 0.00, SRMR: school-
7
le el = 0.05). The s anda dized ac o loadings a e p esen ed in Table 2 (see u he in o ma ion on
8
measu emen models and he i indica o s in Suppo ing in o ma ion).
9
Table 2. S anda dized loadings o he eache -s uden ela ionship and class spi i la en
10
a iables (s uden -le el: N = 27018–27025/ School-le el: N = 194).
11
Teache -s uden ela ionship
Class spi i
S uden -le el
S uden -le el
λ
λ
The opinions o s uden s a e aken
in o conside a ion in he de elopmen
o school wo k.
0.65*
The s uden s in my class ge along well.
0.73*
Teache s a e in e es ed in how I am
doing.
0.56*
The class oom discipline in my class is good.
0.54*
Teache s ea us, he s uden s, ai ly.
0.60*
The mood in ou class is such ha I da e o
exp ess my opinion eely.
0.63*
School-le el
School-le el
The opinions o s uden s a e aken
in o conside a ion in he de elopmen
o school wo k.
0.91*
The s uden s in my class ge along well.
0.87*
Teache s a e in e es ed in how I am
doing.
0.67*
The class oom discipline in my class is good.
0.57*
Teache s ea us, he s uden s, ai ly.
0.71*
The mood in ou class is such ha I da e o
exp ess my opinion eely.a
1.00*
* p < 0.001
12
a The esidual a iance is se o ze o a he school-le el.
13
14

17
The e was a s a is ically signi ican a iabili y bo h in he la en eache -s uden ela ionship
1
- a iable wi hin (σ2W = 0.204, p < 0.001) and be ween schools (σ2B = 0.010, p < 0.001) and in he
2
la en school spi i - a iable wi hin (σ2W = 0.276, p < 0.001) and be ween schools (σ2B = 0.012, p <
3
0.001). The ICC o he bo h la en a iables we e 0.05 and 0.04 espec i ely, meaning ha abou
4
4–5% o he a iabili y occu ed be ween schools. Al hough he ICCs we e ela i ely small, hey
5
had s ong design e ec s (DEFF;61 eache -s uden ela ionship: DEFF = 7.46; school spi i : DEFF =
6
6.76).
7
Then we es ima ed wo andom in e cep models (i.e., Model 1a and Model 1b) in o de o
8
es whe he he pe cei ed eache -s uden ela ionship and class spi i di e ed be ween schools
9
wi h and wi hou obse ed IAQ p oblems (see Table 3). The model i s we e good bo h o he
10
eache -s uden ela ionship (Model 1a: RMSEA = 0.02, CFI = 0.97, SRMR: s uden -le el = 0.01,
11
SRMR: school-le el = 0.07, see Table 2) and o school spi i (Model 1b: RMSEA = 0.02, CFI = 0.99,
12
SRMR: s uden -le el = 0.01, SRMR: school-le el = 0.06, see Table 2). A e all backg ound a iables
13
we e inse ed in o Model 1a, obse ed IAQ p oblems associa ed signi ican ly wi h he eache -
14
s uden ela ionship. The e ec size was 0.5, indica ing a medium-sized e ec 57. Howe e , a e
15
all backg ound a iables we e inse ed in o Model 1b, obse ed IAQ p oblems did no associa e
16
signi ican ly wi h school spi i (see u he in o ma ion on uni a ia e associa ions in Suppo ing
17
in o ma ion) .
18
In Model 1a, he mos impo an s uden -le el backg ound a iables we e pa en al suppo ,
19
pe cei ed heal h, and GPA. The mos impo an school-le el backg ound a iables we e e hnic
20
composi ion (measu ed by s uden s’ sel - epo ed e hnic backg ound) and he school’s a e age
21
GPA. In Model 1b, he mos impo an s uden -le el backg ound a iables we e pa en al suppo ,
22
gende , and pe cei ed heal h. The mos impo an school-le el backg ound a iables we e school
23
18
size and he school’s a e age GPA (see Table 2). Model 1a explained 11% o he o al a iance a
1
he s uden -le el (R2 = 0.114) and 22% a he school-le el (R2 = 0.222). Model 1b explained 8% o
2
he o al a iance a he s uden -le el (R2 = 0.077) and 16% a he school-le el (R2 = 0.157),
3
espec i ely (see Table 2).
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
19
Table 3. Pa ame e es ima es o he eache -s uden ela ionship and he class oom’s social
1
clima e1 as a unc ion o s uden - and school-le el a iables: A andom in e cep model
2
(N=27153 s uden s/194 schools).
3
Model 1a
Teache -s uden ela ionship
Model 1b
Class spi i
B (SE)a
β (SE)b
B (SE)a
β (SE)b
S uden -le el
Gende c
-0.06 (0.01)***
-0.07 (0.01)***
0.13 (0.01)***
0.13 (0.01)***
Aged
0.01 (0.01)
0.02 (0.01)
-0.02 (0.01)
-0.02 (0.01)
Age squa ede
-0.06 (0.01)***
-0.06 (0.01)***
Pe cei ed heal h
0.08 (0.01)***
0.09 (0.01)***
0.16 (0.01) ***
0.15 (0.01)***
Pa en al suppo g
0.12 (0.00)***
0.26 (0.01)***
0.07 (0.01)***
0.13 (0.01)***
Pa en al suppo squa ede
0.03 (0.00)***
0.06 (0.01)***
0.02 (0.01)***
0.03 (0.01)***
Educa ion o mo he h
0.05 (0.01)***
0.06 (0.01)***
-0.02 (0.01)
-0.01 (0.01)
G ade poin a e agei
-0.06 (0.01)***
-0.11 (0.01)***
0.01 (0.01)*
0.02 (0.01)*
G ade poin a e age squa ede
-0.02 (0.01)*
-0.02 (0.01)*
E hnic backg oundj
-0.02 (0.03)
-0.01 (0.01)
0.08 (0.03)**
0.03 (0.01)**
School-le el
Obse ed IAQ p oblemsk
0.05 (0.02)**
0.25 ( 0.08)**
0.02 (0.02)
0.08 (0.08)
G ade poin a e aged
-0.22 (0.09)*l
-0.28 (0.10)**
-0.24 ( 0.09) **
-0.27 (0.10)**
E hnic backg ound
-0.36 (0.12)**
-0.19 (0.09)*n
0.03 (0.19)
0.02 (0.09)
School size o
0.01 (0.00)m
0.16 (0.08)*
0.01 (0.01)*
0.24 (0.09)*
σ2W
0.17 (0.01)
0.89 (0.01)
0.24 (0.01)
0.92 (0.01)
σ2B
0.01 (0.00)
0.78(0.07)
0.01 (0.00)
0.84 (0.08)
*p < 0.05, **p < 0.01, *** p < 0.001
4
1 A highe alue in he eache -s uden ela ionship and he class oom’s social clima e means mo e p oblems.
5
a Uns anda dized be a
6
b S anda dized be a
7
c Scale 0–1 (0 = boy)
8
d G and mean cen e ed
9
e Quad a ic e m. Only i he quad a ic e m was signi ican (p < 0.05), i was included in he inal model and epo ed.
10
The e we e no signi ican quad a ic e ms a he school-le el.
11
Scale 0–1 (0 = e y good heal h)
12
20
g Scale 1–4 (1 = good suppo , g and mean cen e ed)
1
h Scale 0–1 (1 = uni e si y educa ion)
2
i Scale 0–2 (0 = GPA less han 7.5, g and mean cen e ed)
3
j Scale 0–1 (0 = one o bo h pa en s a e bo n in Finland)
4
k Scale 0–1 (0 = no obse ed IAQ p oblems in he school)
5
l p = 0.012
6
m p = 0.064
7
n p = 0.027
8
o In hese models he scale was changed as ollows: school size/100.
9
10
3.2. The e ec o he school’s IAQ p oblems on he associa ions be ween he backg ound and
11
ou come a iables
12
We es ed di e ences on he associa ions be ween a) s uden -le el cha ac e is ics and he
13
s uden -pe cei ed social clima e and b) school-le el cha ac e is ics and he s uden -pe cei ed
14
social clima e in schools wi h and wi hou IAQ p oblems. In o de o es hese di e ences we
15
used mul ig oup wo-le el s uc u al equa ion modeling and Wald- es . The associa ion be ween
16
s uden -le el GPA and class spi i was mode a ed by obse ed IAQ p oblems (Wald- es = 6.04, p =
17
0.014). This associa ion was s onge in schools wi h IAQ p oblems (s anda dized be a = 0.05, p <
18
0.001) han in schools whe e he e we e no such p oblems (s anda dized be a = 0.00, p = 0.821;
19
see u he in o ma ion on uni a ia e associa ions in Suppo ing in o ma ion). In o de o be e
20
unde s and his esul , we an h ee sepa a e analyses. We es ed s uden s whose GPA was a) less
21
han 7.5, b) be ween 7.5 and 8.4 and c) o e 8.4. Excep o GPA, he o he backg ound a iables
22
in hese h ee analyses we e he same as in Model 1b (see Table 2). S uden s wi h he highes GPA
23
(i.e., o e 8.4) pe cei ed mo e p oblems in class spi i in schools wi h obse ed IAQ p oblems han
24
in schools wi hou such p oblems (s anda dized be a = 0.19, p = 0.032). The e ec size is 0.4,
25
indica ing a small-sized e ec . 57 The e we e no di e ences be ween schools among s uden s in
26
he lowe GPA clus e s (GPA 7.5–8.4: s anda dized be a = 0.12, p = 0.193; GPA < 7.5: s anda dized
27
be a = 0.01, p = 0.948).
28
21
1
Figu e 3. The mean di e ence o class spi i be ween schools wi h and wi hou obse ed IAQ
2
p oblems di ided by he s anda d de ia ion o class spi i .
3
0 = GPA < 7.5, p = 0.948; 1 = GPA 7.5–8.4, p = 0.193; 2= GPA > 8.4, p = 0.032 (N = 193–194
4
schools).
5
6
Fu he mo e, he linea e ms o he pa en al suppo and eache -s uden associa ions we e
7
signi ican ly di e en (Wald- es = 4.75, p = 0.029) be ween he wo school con ex s (no obse ed
8
IAQ p oblems: s anda dized be a = 0.27, p < 0.001; obse ed IAQ p oblems: s anda dized be a =
9
0.23, p < 0.001; see u he in o ma ion on uni a ia e associa ions in Suppo ing in o ma ion). The
10
Wald- es be ween quad a ic e ms was no signi ican (Wald- es = 0.50, p = 0.479). This small
11
e ec indica es ha he di e ence in he pe cei ed quali y o he eache -s uden ela ionship
12
be ween he wo school con ex s was la ge among s uden s wi h high pa en al suppo han
13
s uden s wi h low pa en al suppo . In o al, s uden s wi h low pa en al suppo , howe e ,
14
e alua ed hei eache -s uden ela ionship mo e nega i ely han s uden s wi h high pa en al
15
suppo .
16
17
18
G ade poin a e age
Class spi i

22
4. Discussion
1
Ou main inding was ha on a e age, s uden s in schools wi h obse ed IAQ p oblems epo ed
2
wo se eache -s uden ela ionship han s uden s in schools wi hou such p oblems. Fu he mo e,
3
we ound ha s uden s wi h a high GPA epo ed wo se class spi i in schools wi h IAQ p oblems
4
han in schools wi hou such p oblems.
5
P e ious esea ch has epo ed associa ions be ween poo IAQ in schools and bo h heal h
6
and lea ning ou comes (e.g., espi a o y symp oms, illness absence, and academic pe o mance).
7
3,7,9,62 Ou indings con ibu e o his esea ch by showing ha IAQ p oblems in schools may ha e
8
an e ec on he s uden -pe cei ed social clima e. To ou knowledge, his is he i s ime when
9
hese e ec s ha e been assessed. Gi en ha social clima e is shown o in luence a ious ac o s,
10
such as academic pe o mance, absence a es, and s uden s’ psychosocial well-being, 10,30,31 ou
11
indings a e impo an . Fu he mo e, we ound ha a school’s obse ed IAQ p oblems signi ican ly
12
modi ied he associa ions be ween s uden -le el cha ac e is ics and he s uden -pe cei ed social
13
clima e. Al hough hese associa ions we e weak, he indings suppo ed he p e ious esea ch by
14
Finell and colleagues38 by concluding ha schools wi h and wi hou IAQ p oblems may ep esen
15
wo di e en social con ex s. They also sugges ha he social and psychological consequences o
16
IAQ p oblems in schools esul om complex p ocesses whe e bo h he s uden s’ and school
17
pe sonnel’s cha ac e is ics play a ole.
18
An impo an ask o u u e esea ch is o examine why IAQ p oblems in schools may
19
in luence he s uden -pe cei ed social clima e. Ou esul s and he li e a u e p o ide h ee
20
po en ial in e wined pa hs. The i s pa h ela es o he impac o poo IAQ on school pe sonnel,
21
o example by dec easing hei well-being, inc easing sick lea e, and causing possible p oblems in
22
managemen and decision-making p ocedu es. 19,20,26,28,63 The second pa h ela es o he impac o
23
23
poo IAQ on s uden s. The s uden s’ pe cei ed heal h is associa ed wi h he s uden -pe cei ed
1
social clima e, 10 as was also ound in ou s udy. The hi d pa h ela es o he poo condi ion o he
2
school acili ies as well as an icipa ed emedia ion o he school building. 22,23 In Finland, school
3
gene al managemen and acili y managemen a e independen om each o he . In o he wo ds,
4
school p incipals and s a do no ha e a di ec ole o o e sigh o he acili y managemen . They
5
can epo any issues o conce ns bu he alloca ion o unds o school acili y managemen and
6
emedia ion decisions a e made by he municipali y echnical boa d. Howe e , emedia ion
7
p ocesses o en esul in dis u bances (e.g., mo ing o empo al buildings and/o noise), which can
8
be s ess ul bo h o school pe sonnel and s uden s, and may he e o e in luence he s uden -
9
pe cei ed social clima e.
10
F om a p ac ical poin o iew, an impo an ques ion is how i is possible o p e en he
11
po en ial social consequences caused by IAQ p oblems in schools. The main ask is na u ally o
12
in es iga e and p ope ly emedia e indoo ai p oblems in he school building. Howe e , hese
13
p ocesses a e o en slow. In case he emedia ion is delayed, he p e en ion o heal h p oblems
14
among eache s and s uden s can in ol e sou ce con ol (e.g., mo e equen cleaning), enhanced
15
en ila ion, and exposu e p e en ion (e.g., no using p oblema ic a eas). Fu he mo e,
16
managemen and decision-making p ocedu es should be de eloped in o de o a oid con lic s
17
inside he school o ganiza ion and expe iences o injus ice.64 Open communica ion and sha ing
18
up- o-da e ( egula , ac ual) in o ma ion wi h school pe sonnel, s uden s, and pa en s should help
19
o main ain us be ween pa ies.
20
The s eng hs o he s udy include he la ge and ep esen a i e sample. The use o mul ile el
21
s uc u al equa ion models has many ad an ages, including be e s a is ical powe compa ed o
22
models using agg ega ed school-le el da a and a oiding spu ious indings om using s uden -le el
23
24
da a wi hou aking he hie a chical da a s uc u e in o accoun . 48 The limi a ions o he s udy
1
include he c oss-sec ional design and he use o ques ionnai e da a. Fu he mo e, we did no
2
ha e physical measu es. This means ha we ha e o ely on p inciples epo on he indings o
3
iennial inspec ion on biological exposu es and ha we a e canno be e dis inguish schools
4
based on he ac o s ha ha e caused hei IAQ p oblems. I is likely ha he e exis s a ia ion in
5
he se e i y o schools’ IAQ p oblems and hei social consequences. We also do no know he
6
deg ee eache s and s uden s we e awa e o hei school’s IAQ p oblems. In he u u e, hese
7
issues should be u he s udied using longi udinal da a and physical measu es. Fu u e esea ch
8
should also de ec ac o s ha can media e he associa ion be ween school’s indoo en i onmen
9
and he pe cei ed social clima e. Also possible building- ela ed con ounde s (e.g., poo sani a y
10
condi ions o unpleasan school en i onmen ) should be conside ed.
11
The p esen s udy is pa o a la ge p ojec aiming o cla i y associa ions be ween
12
psychosocial ac o s and IAQ p oblems among comp ehensi e school s uden s in Finland. In a
13
p e ious s udy, we ound associa ions be ween bo h he physical and psychosocial en i onmen
14
and subjec i e IAQ in schools. 38 In he p esen s udy, psychosocial ac o s a e he ou come
15
a iables. Ou s udy unde lines he ac ha he associa ions be ween he pe cep ion o IAQ
16
p oblems and psychosocial ac o s a e highly complex and a e likely o in luence each ano he in
17
a ious ways.
18
19
20
21
22
25
5. Conclusions
1
Ou s udy ound ha p oblems in a school’s IAQ associa ed wi h he s uden -pe cei ed social
2
clima e. Fu he mo e, we ound ha he school’s obse ed IAQ p oblems signi ican ly modi ied
3
he associa ions be ween s uden -le el cha ac e is ics and he s uden -pe cei ed social clima e.
4
Ou s udy poin s o a need o s udy schools’ IAQ p oblems om a mul idisciplina y pe spec i e in
5
o de o be e unde s and how IAQ p oblems in luence s uden s’ psychosocial well-being.
6
Fu he mo e, om p ac ical poin o iew, when e alua ing he u gency o emedia ion, also IAQ
7
p oblems’ possible social consequences should be aken in o accoun . Fas and e ec i e
8
emedia ion p ocess no only p e en ad e se heal h e ec s bu can also p e en p oblems in
9
school’s social clima e.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
32
59. Rubin DB. In e ence and missing da a. Biome ika. 1976;63:581–592.
1
60. Ende s C, Bandalos D. The ela i e pe o mance o ull in o ma ion maximum likelihood
2
es ima ion o missing da a in s uc u al equa ion models. S uc . Equ. Model. Mul idiscip. J.
3
2001;8:430–457 .
4
61. Ukoumunne OC, Gulli o d, MC, Chinn S, e al. Me hods o e alua ing a ea-wide and
5
o ganisa ion-based in e en ions in heal h and heal h ca e: a sys ema ic e iew. Heal h
6
Technol. Assess. 1999;3.
7
62. Mendell MJ, Elisee a EA, Da ies MM, Lobscheid A. Do class oom en ila ion a es in Cali o nia
8
elemen a y schools in luence s anda dized es sco es? Resul s om a p ospec i e s udy.
9
Indoo Ai . 2016;26:546–557.
10
63. Mille RT, Mu nane RJ, Wille JB. Do eache absences impac s uden achie emen ?
11
Longi udinal e idence om one u ban school dis ic . Educ. E al. Policy Anal. 2008; 30: 181–
12
200.
13
64. Finell E, Seppälä T. Indoo ai p oblems and expe iences o injus ice in he wo kplace: A
14
quan i a i e and a quali a i e s udy. Indoo Ai . 2017; doi:10.1111/ina.12409.
15
16
17
18
19
20
21
22
23
24
25

33
1
2
Figu e 1
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
= measu emen model
33
34
Figu e 1. Hypo hesized s uc u al equa ion model: The eache -s uden ela ionship as a la en
35
ou come a iable.
36
Teache -s uden
ela ionship
Teache s a e in e es ed
in how I am doing
Gende
Teache s ea us, he
s uden s, ai ly
The opinions o
s uden s a e aken in o
conside a ion in he
de elopmen o school
wo k
Age
Pe cei ed
heal h
Pa en al
suppo
GPA
E hnic
backg ound
Educa ion
o mo he
Teache -s uden
ela ionship
S uden -le el
School-le el
Teache s a e in e es ed
in how I am doing
Teache s ea us, he
s uden s, ai ly
The opinions o
s uden s a e aken in o
conside a ion in he
de elopmen o school
wo k
GPA
E hnic
backg ound
School size
Obse ed
IAQ
p oblems
34
1
Figu e 2
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
= measu emen model
32
33
Figu e 2. Hypo hesized s uc u al equa ion model: Class spi i as a la en ou come a iable.
34
Class-spi i
S uden -le el
School-le el
The s uden s in my
class ge along well.
The class oom
discipline in my class is
good.
The mood in ou class is
such ha I da e o
exp ess my opinion
eely.
GPA
E hnic
backg ound
School size
Obse ed
IAQ
p oblems
Gende
Age
Pe cei ed
heal h
Class-spi i
Pa en al
suppo
The s uden s in my
class ge along well.
The class oom
discipline in my class is
good.
The mood in ou class is
such ha I da e o
exp ess my opinion
eely.
GPA
E hnic
backg ound
Educa ion
o mo he