Teachers’ situational physiological stress and affect

This is a sel -a chi ed e sion o an o iginal a icle. This e sion may di e om he o iginal in pagina ion and ypog aphic de ails. Au ho (s): Ti le: Yea : Ve sion: Copy igh : Righ s: Righ s u l: Please ci e he o iginal e sion: CC BY 4.0 h ps://c ea i ecommons.o g/licenses/by/4.0/ Teache s’ si ua ional physiological s ess and affec © 2023 The Au ho s. Published by Else ie L d. Published e sion Jõgi, Anna-Liisa; Malmbe g, La s-E ik; Paka inen, Eija; Le kkanen, Ma ja-K is iina Jõgi, A.-L., Malmbe g, L.-E., Paka inen, E., & Le kkanen, M.-K. (2023). Teache s’ si ua ional physiological s ess and affec . Psychoneu oendoc inology, 149, A icle 106028. h ps://doi.o g/10.1016/j.psyneuen.2023.106028 2023 Psychoneu oendoc inology 149 (2023) 106028 A ailable online 13 Janua y 2023 0306-4530/© 2023 The Au ho s. Published by Else ie L d. This is an open access a icle unde he CC BY license (h p://c ea i ecommons.o g/licenses/by/4.0/). Teache s’ si ua ional physiological s ess and a ec Anna-Liisa J˜ ogi a , c , * , La s-E ik Malmbe g b , Eija Paka inen c , d , Ma ja-K is iina Le kkanen c , d a Tallinn Uni e si y, Es onia b Uni e si y o Ox o d, UK c Uni e si y o Jy ¨ askyl¨ a, Finland d Uni e si y o S a ange , No way ARTICLE INFO Keywo ds: Physiological s ess Co isol A ec S uc u al equa ion modeling In aindi idual ela ions Teache s ABSTRACT Since eaching is a demanding and s ess ul p o ession, he s udy o eache s’ physiological s ess in he class- oom se ing is an eme ging ield. In c oss-sec ional s udies sel - epo ed s ess and a ec a e ela ed, bu less is known abou he in aindi idual ela ions be ween si ua ional physiological s ess and co esponding posi i e and nega i e a ec . The aim o ou s udy was o in es iga e he associa ions be ween si ua ional physiological s ess (six sali a y co isol samples pe day) and sel - epo ed si ua ional a ec (Posi i e and Nega i e A ec Schedule ou imes a day) among 61 Finnish p ima y school eache s o e wo wo kdays. We p esen a no el mul ile el s uc u al equa ion model (MSEM) ha includes co isol, wi h ime since awakening as a lexibly coded ime- a ying co a ia e and a ec wi h ime since co isol measu emen as a ime- a ying co a ia e. Highe le els o eache s’ si ua ional physiological s ess we e ela ed o lowe si ua ional posi i e a ec (e.g., en husiasm) and highe nega i e a ec (e.g., ne ousness), demons a ing he acu e/si ua ional e ec s o s ess on a ec . In ou discussion, we emphasize he impo ance o he sequence o sampling and obse a ions o u he heo e ical modeling o ela ions be ween s ess and a ec . We also p opose p ac ical implica ions o imp o ing eache s’ awa eness o hei well-being. 1. In oduc ion The eaching p o ession is conside ed mo e s ess ul han o he oc- cupa ions, impac ing bo h eache s’ physical and psychological well- being (Johnson e al., 2005), which has consequences o bo h hem- sel es and hei s uden s. Lowe le els o eache s’ well-being ha e been associa ed wi h poo e s uden ou comes and less posi i e s u- den – eache ela ionships (Ald up e al., 2018; A ens and Mo in, 2016). Teache s’ posi i e emo ions (e.g., enjoymen ) ha e been associa ed wi h s uden engagemen while nega i e ones (e.g., anxie y) wi h s uden disengagemen (F enzel e al., 2020; Li e al., 2022). Thus a , mos s udies in es iga ing eache s’ well-being ha e u ilized sel - epo ed s ess (e.g., Kelle e al., 2014) and ha e ocused on indi idual di e ences be ween eache s (e.g., Hamama e al., 2013). We go beyond p e ious s udies o eache s’ s ess and a ec by in es iga ing he ime-dependen wi hin-le el ela ionship be ween eache s’ physiolog- ical s ess and a ec . We apply mul ile el s uc u al equa ion models (MSEMs) o in es- iga e he wi hin-pe son associa ions be ween eache s’ si ua ional co isol le els and si ua ional posi i e and nega i e a ec o e wo wo kdays. Bo h physiological and a ec i e eac ions a e conside ed as esponses o s ess (Schlo z, 2019). Thus a , esea ch on si ua ional co isol and a ec has mos ly u ilized mul ile el models (MLMs) ha allow o only one ( ime) dependen a iable (Doane and Adam, 2010; Hoppmann and Klumb, 2006). S uc u al equa ion modeling (SEM) has been used in combining di e en co isol samples om di e en ime- poin s and/o days wi h la en ai s (Doane e al., 2015; Mille e al., 2016). We speci ied MSEMs ha pa i ioned a iances in o wi hin- ( imepoin s) and be ween-le el a iances ( eache s), bu allowed o co a ia e e ec s on bo h he dependen (a ec ) and independen (co isol) a iables, p o iding mo e lexibili y in he model-speci ica ion han MLMs o e all (Aspa ouho and Mu h´ en, 2020). 1.1. Physiological s ess and a ec Physiological s ess has majo nega i e consequences on one’s physical and men al heal h, and well-being in gene al (Schneide man e al., 2005). Howe e , s udies hus a ha e indica ed ha physiological s ess as measu ed by exc e ion o he co isol ho mone migh be weakly o no a all ela ed o sel - epo ed s ess o exhaus ion (e.g., Belling a h * Co espondence o: School o Educa ional Sciences, Tallinn Uni e si y, Na a mn 25, Tallinn 10120, Es onia. E-mail add ess: [email p o ec ed] (A.-L. J˜ ogi). Con en s lis s a ailable a ScienceDi ec Psychoneu oendoc inology jou nal homepage: www.else ie .com/loca e/psyneuen h ps://doi.o g/10.1016/j.psyneuen.2023.106028 Recei ed 28 Augus 2022; Recei ed in e ised o m 1 Decembe 2022; Accep ed 12 Janua y 2023 Psychoneu oendoc inology 149 (2023) 106028 2 e al., 2008), which makes i especially impo an o add ess physio- logical s ess sepa a ely. A ec is a eeling o s a e o mind o which one is conscious (F e- d ickson, 2001). Posi i e a ec as a ai is all-inclusi ely ela ed o be e heal h and well-being in gene al, and as a s a e, i mi iga es si ua ional physical o psychological dis ess (Lyubomi sky e al., 2005). In con as , nega i e heal h ou comes a e ela ed o nega i e a ec (Diene e al., 2017). In labo a o y s udies, he physiological and a - ec i e (i.e., sel - epo ed) s ess esponses measu ed in si ua ion end o be weakly and a he inconsis en ly ela ed (Campbell and Ehle , 2012). Fu he mo e, he supp ession o physiological s ess esponse does no in ol e he dec ease in emo ional esponse (Ali e al., 2017). Inconsis en associa ions a e pa ly explained by me hodological issues (see Campbell and Ehle , 2012 o a e iew) and migh be pa ly due o he he e ogenei y o he sample o pa icipan s as shown in Simon e al. (2022). Ambula o y s udies o da e ha e mos ly ocused on associa ions be- ween physiological s ess and a ec a he be ween-pe son le el (e.g., Mille e al., 2016). In mos cases, i has been concluded ha highe posi i e a ec is ela ed o lowe daily co isol le els (see S ep oe, 2019 o a e iew; Polk e al., 2005 o he excep ion), while be ween-pe son ela ions be ween nega i e a ec and co isol ha e shown ei he posi- i e (e.g., Polk e al., 2005) o no ela ions (e.g., Mille e al., 2016). A ecen me a-analysis by Joseph e al. (2021) emphasized he need o clea ly dis inguish si ua ional wi hin-pe son ela ions be ween physio- logical s ess and a ec om be ween-pe son ones. They also showed ha wi hin-pe son ela ions be ween co isol and a ec a e qui e ho- mogeneous and in he same di ec ion as be ween-pe son ela ions (Jo- seph e al., 2021). 1.2. Teache s’ s udies Teaching is deemed as a e y s ess ul job (B ough on, 2010) and eache s’ sel - epo ed s ess is he e o e widely s udied. In gene al, he sel - epo ed s ess is nega i ely ela ed o posi i e a ec and posi i ely ela ed o nega i e a ec (Hamama e al., 2013; Mon gome y and Rupp, 2005). Howe e , s udying eache s’ physiological s ess in he class oom en i onmen is s ill in i s ea ly s ages. To da e, i has been ound ha eache s’ physiological s ess is highe a he end o he school yea , compa ed o he beginning o he school yea (Ka z e al., 2018). Teache s’ co isol le els a e also highe on wo kdays han on weekends (We s ein e al., 2020). Teache s’ physiological s ess does no appea o co ela e wi h hei sel - epo ed s ess o bu nou (Ka z e al., 2016; Nislin e al., 2016). Thus a , e y ew s udies on eache s ess ha e di e en ia ed be- ween in e - and in ape sonal le els o analyses, al hough i is s ongly ecommended in he analy ical li e a u e (Hamake , 2012). Using hea a e as a s ess indica o has shown ha lowe s uden engagemen coupled wi h eache -cen e ed ins uc ional p ac ice is ela ed o eache s’ highe si ua ion-speci ic physiological s ess (Junke e al., 2021). Highe physiological a ousal, also measu ed by hea a e, in associa ion wi h class oom agency is ela ed o highe enjoymen in he class oom (Donke e al., 2020). To he bes o ou knowledge, no p e- ious s udies ha e explo ed he in aindi idual ela ions be ween eache s’ si ua ional physiological s ess measu ed by co isol and a ec du ing he wo kday. 1.3. The cu en s udy Taken oge he , he associa ion be ween physiological s ess and a ec has been qui e widely s udied, bu hus a , mos ly as associa ions in c oss-sec ional s udies. When MLMs ha e been speci ied, models ha e only included one dependen a iable a a ime, limi ing he ways in which mo e complex ela ionships be ween a iables can be speci ied. In he cu en s udy, we ollow an in aindi idual app oach o modeling, in which ime-poin s a e nes ed in pe sons. Speci ying mul ile el s uc u al equa ion models (MSEM) allows us o sepa a ely model as- socia ions and di ec ional ela ionships (he e, co isol p edic ing expe- ienced a ec ) a he wi hin and be ween le els. This is consis en wi h he e godici y-assump ion (Molenaa , 2004), ha i is no possible o d aw conclusions abou in aindi idual p ocesses, changes, and a i- abili y based on indi idual di e ences (i.e., mean-le el di e ences be- ween pe sons). Be ween-pe son ela ions do no allow making claims abou he co a ia ions be ween wo indica o s wi hin one pe son (Molenaa , 2004). Indica o s ha a e posi i ely ela ed in be ween-pe son analysis, migh be nega i ely ela ed in wi hin-pe son le el and ice e sa (Cu an and Baue , 2011; Hamake , 2012). Also, he MLMs impede using wo o mo e dependen a iables and, consequen ly, manage he disc epancies in co isol measu emen s and a ec obse a ions. S udies using a ec o p edic co isol le els ha e mos ly measu ed a ec and co isol a exac ly he same ime, wi hou conside ing he ime di e ence be ween co isol sampling and a ec obse a ions (Doane and Adam, 2010; Hoppmann and Klumb, 2006). We assumed ha on wo kdays and in ambula o y se ings, i would be di icul o ensu e he exac same ime s amps o bo h co isol and a ec , a isk also acknowledged by Joseph e al. (2021). This wa an s aking he ime lag in o accoun when modeling co isol and a ec . Speci ically, we aimed o in es iga e wi hin-pe son ela ions be- ween eache s’ physiological s ess and a ec du ing he wo kday. Consis en wi h p e ious s udies (Hoppmann and Klumb, 2006; Joseph e al., 2021), we expec ed ha eache s’ highe physiological s ess is ela ed o lowe posi i e and highe nega i e a ec a he wi hin-pe son le el (Hypo hesis 1). Ou heo e ical wo-le el model wi h wo ime- a ying indica o s is p esen ed in Fig. 1 (see also Sec ions 2.2.3 and 2.3). 2. Me hod 2.1. Sample and p ocedu e Ou s udy was pa o a la ge longi udinal esea ch p ojec (Le k- kanen and Paka inen, 2016-2022) in es iga ing p ima y school each- e s’ and s uden s’ s ess in e ac ions in he class oom. Ou s udy sample consis ed o 61 p ima y school eache s (5 male) om Cen al Finland. Fo y-nine o hem augh 8–9-yea -old s uden s in G ade 2, and 12 o hem augh 9–10-yea -olds in G ade 3. The mean age o he pa ici- pa ing eache s was 45.3 yea s (SD =9.4), and hei mean wo k expe- ience was 17.8 yea s (SD =10.1). Da a we e collec ed o e wo wo kdays du ing he sp ing semes e o he yea 2019 (49 eache s) o 2020 (12 eache s). O e all, da a Fig. 1. Theo e ical model o wo-le el associa ions be ween co isol and posi- i e o nega i e a ec . No e. a Time a e awakening o co isol sampling in hou s. b Time lag be ween he a ec sel - epo and he co isol sampling in hou s. A.-L. J˜ ogi e al. Psychoneu oendoc inology 149 (2023) 106028 3 collec ion las ed om la e Feb ua y o he beginning o May in 2019 and om la e Feb ua y o he middle o Ma ch in 2020. Fo 47 eache s, he wo wo king days we e consecu i e; in 9 cases, hey we e h ee days apa , including o e a weekend; in 5 cases, he e we e 3–6 days be- ween wo measu emen s, and 2 eache s did no p o ide da a on Day 2. P io o da a collec ion, he pa icipan s ecei ed w i en ins uc ions o collec ing co isol samples and esponding o a ec ques ionnai es, and hey we e pe sonally ins uc ed by ained esea ch assis an s. The esea ch assis an s we e p esen in class du ing he i s day o da a collec ion o o e suppo . The eache s could also ge in ouch wi h a esea ch assis an ia phone a any ime. The da a collec ion and analysis p ocedu es ollowed he p inciples o The Decla a ion o Helsinki. The uni e si y e hics commi ee app o ed he s udy be o e he da a collec ion s a ed. All eache s p o- ided hei w i en consen o pa icipa e in he s udy. 2.2. Measu es 2.2.1. Physiological s ess Sali a y co isol was used as an indica o o he eache s’ physio- logical s ess. I is ecommended as an ecologically alid measu e o ambula o y assessmen s (Kudielka e al., 2012). Co isol is eleased in he body du ing he s ess esponse and i a ec s a wide ange o issues. The main aim o co isol is o suppo o ganisms o cope wi h s ess and main ain homeos asis (de Kloe e al., 1998). Co isol elease in he body has a ce ain diu nal hy hm—co isol le els inc ease apidly in he mo ning, a e awakening (i.e. co isol awakening esponse—CAR), hen ollowed by a sha p decline and hen dec ease smoo hly du ing he day (Kudielka e al., 2012). The eache s we e asked o p o ide six sali a samples (a he ime o awakening, 30 and 45 min a e awakening, a 10 am., a he end o he school day app oxima ely a 12–13 pm, and be o e bed ime) pe day o e wo wo kdays o a o al o 12 samples pe eache . Teache s e- po ed hei ime o awakening and he ime o each co isol sampling. Syn he ic Sali e e® Co isol swabs (by Sa s ed ) we e used o sali a sampling. The eache s we e ins uc ed no o ea o d ink any hing o he han wa e , b ush hei ee h, o smoke 30 min be o e sampling. The pa icipan s we e asked o gen ly chew he swabs o one minu e immedia ely upon awakening, 30 and 45 min a e awakening, a 10 am, a he end o he wo kday (a ound 12–1 pm), and a bed ime. Sali a samples we e s o ed in he deep eeze a e he sampling. Resea ch assis an s collec ed samples om schools a e he Day 2 sampling, labeled and s o ed hem in a deep eeze a he uni e si y. Samples we e anspo ed o he lab by cou ie se ice in 1–2 days. Sali a samples we e assayed a he D esden LabSe ice GmpH acili ies, using Co isol Luminesence Immunoassay (CLIA RE62011 by IBL In- e na ional) o de e mine co isol concen a ions. O he samples, 20% we e andomly selec ed o be assayed wice, gi ing an in e -assay co- e icien o a ia ions below 7%. Upon sc eening he quali y o he da a, he ollowing co isol samples we e excluded om analyses (see Fig. 2): 3 co isol samples we e collec ed on an e oneous sampling day; 15 samples iola ed ea ing es ic ions, as de e mined by sel - epo ed meal o snack imes; and 12 samples had co isol concen a ions la ge han 73 nmol/l (equal o 60 nmol/l assayed by andem mass-spec ome y [LC-MS/MS] as a e e - ence me hod (Mille e al., 2013)) and we e excluded as physiologically implausible. The i s h ee mo ning samples om wo eache s om one day we e excluded because he gap be ween awakening and he i s Fig. 2. Da a exclusion c i e ia. A.-L. J˜ ogi e al. Psychoneu oendoc inology 149 (2023) 106028 4 sample was longe han 60 min, as indica ed by he sel - epo s. One eache ’s i s sampling ime was 30 min a e awakening; he e o e, he i s and second samples we e ede ined as he second and hi d samples, espec i ely, and he ac ual hi d sample was emo ed om he da a. In o al, 691 sali a y co isol samples we e included in he analyses. The dis ibu ions o he eache s’ co isol le els o e wo days a e p esen ed in Fig. 3. Raw co isol alues (nmol/l) we e na u al loga i hm ans- o med o he analyses because o he posi i e skewness (Adam and Kuma i, 2009). 2.2.2. Posi i e and nega i e a ec A Finnish e sion o he Posi i e and Nega i e A ec Schedule (PANAS) was adminis e ed (C aw o d and Hen y, 2004; Hie alah i e al., 2016), which included i e i ems cap u ing he eache s’ si ua ional posi i e a ec (e.g., “a en i e,” “en husias ic”) and i e i ems ep e- sen ing nega i e a ec (e.g., “ne ous,” “sca ed”). The eache s esponded using sma phones o in one case pe sonal compu e s ou imes a day (upon awakening, a 10 am, a he end o he wo kday, and be o e bed ime), ega ding o wha ex en each emo ion bes desc ibed hem a ha pa icula momen (1 =“does no desc ibe me a all,” 5= “desc ibes me e y well”). Teache s we e asked o comple e he PANAS ques ionnai e app oxima ely a he same ime as hey ga e a co isol sample. The esponse ime was eco ded in he de ice. The mean o i e i ems was used as a posi i e a ec indica o (C onbach’s α anged om.74 o.94 o 8 imepoin s). The i ems measu ing nega i e a ec we e highly skewed, wi h ou ou o i e i ems ha ing he answe “does no desc ibe me a all” 83–94% o he ime. Two easons migh be plausible o he skewness. Fi s , he PANAS scales ask abou e y in ense emo ions (Diene e al., 2017), which end o occu less equen ly (see Hoy e al., 2016 o a compa ison). Second, linguis ic and cul u al di e ences migh be p esen , as emo ions a e less easily exp essed in Finnish han in English (Chen e al., 2012). In conclusion, we used nega i e a ec as a bina y indica o , coding i as 0 i he eache had no el any nega i e emo ions a all a pa icula ime poin , and coding i as 1 i he eache indica ed ha a leas one nega i e a ec i em desc ibed hem a a pa icula ime poin . In e nal consis ency o he nega i e a ec scale (o dinal α ) anged om.50 o.96 o 8 ime poin s. In o al, we ecei ed 412 a ec epo s, up o ou epo s o each o wo days om each pa icipan , in o al up o eigh epo s pe pa ici- pan . The ollowing obse a ions we e excluded be o e he analyses (see Fig. 2 o he igh ): obse a ions close han 15 min o he p e ious obse a ion, which we e conside ed epea measu emen s (n =35 e- po s); one obse a ion on he w ong day (n =1), and iming was no a ibu able o any imepoin (n =15). Fo example, i a eache had illed in he ques ionnai e upon awakening and also an hou la e , and again a 10 am, one o he i s wo epo s was omi ed. In o al, 361 a ec epo s we e included in he analyses. The dis ibu ion o he means o posi i e a ec o e he day a e p esen ed in Fig. 3 and aw da a in Table 1. 2.2.3. Linking co isol and a ec obse a ions We linked co isol obse a ions wi h sel - epo s so ha he i s co isol sample coincided wi h he i s PANAS obse a ion, and he second and hi d co isol samples we e un ela ed wi h he a ec epo s. The h ee emaining co isol samples coincided wi h h ee PANAS ob- se a ions. The esul ing da ase included up o 12 samples o each eache ’s co isol (six samples pe day), and up o eigh esponses o posi i e and nega i e a ec ( ou epo s pe day). This linkage ga e a da a s uc u e wi h wo a ec epo s missing pe day by design. The lexibly coded ime indica o o co isol was he ime o sam- pling om he ime o awakening in hou s. Fo sel - epo ed a ec , he ime o obse a ion om awakening was e y highly co ela ed o he co isol ime indica o . In o de o a oid mul icollinea i y in ou models, we used he lexibly coded ime indica o , gi ing he ime lag om he co isol measu emen . Fo example, i he ou h co isol sample was aken a 12 pm, 5.5 h since awakening, and he sel - epo ed a ec a 12:10, he wo ime- a ying co a ia es would be 5.5 o co isol and 0.167 h since he co isol sample. Fi e eache s we e missing wake-up ime da a on day 1 and 10 eache s on day 2; he e o e, hei wake-up Fig. 3.. Va iabili y o sali a y co isol (A) and posi i e a ec (B) du ing wo measu emen days. No e. The hick ed- illed line ep esen s he mean co isol le els (A) and posi i e a ec (B) o e he day. A.-L. J˜ ogi e al. Psychoneu oendoc inology 149 (2023) 106028 5 ime was eplaced wi h ime o he i s co isol sampling. Ou da a-s uc u e had up o 12 imepoin s o co isol (6 pe day) and up o eigh imepoin s o a ec (up o ou pe day). In p e ious esea ch, co isol da a om di e en measu emen days ha e been agg ega ed o la en co isol indica o s o es ima e be ween-pe son di e ences (Doane e al., 2015; Mille e al., 2016). Ou app oach en- ables us o answe esea ch ques ions conce ning in aindi idual di e ences. P elimina y inspec ion o he da a (we eg essed co isol on day) sugges ed no wi hin-pe son di e ence in co isol esponse be ween he wo days, suppo ing his decision. A ec esponses di e ed on he wo days, necessi a ing he need o include day as a co a ia e in he models. In conclusion, o modeling, we hus used h ee indica o s o he ime s uc u e: con inuous ime o co isol sampling elapsed since awakening in hou s, he ime lag be ween he co isol esponse and sel - epo ed a ec , and day (0 =Day 1, 1 =Day 2). 2.3. S a is ical p ocedu es Fi s , we inspec ed he wi hin- and be ween-le el a ia ions in he eache s’ physiological s ess and a ec . We speci ied h ee sepa a e MLMs o change o a) co isol (Eq. 1), b) posi i e a ec , and c) nega i e a ec (Eq. 2) in he MSEM amewo k, using MPlus 8.6 (Mu hen and Mu hen, 1998–2017). Wi h imepoin s ( ) nes ed in indi iduals (i) and a alue o co isol le el o indi idual (i) a ime ( ) as a dependen a i- able, Eq. (1) was as ollows: Co isol i =β0 +β1Time i +β2Timepoin 2 i +β3Timepoin 3 i +β4Day i + υ 0i + ε 1i (1) β0 is he o e all co isol in e cep , β1 is he slope (change in co isol o e ime), υ 0i is he andom pa o he in e cep , and ε 1i is he esidual. We also speci ied he CAR by including wo dummy-coded a iables, β2 and β3 (co isol samples aken 30 and 45 min a e awakening) o e lec CAR (Doane and Adam, 2010; Ka z e al., 2018). Fo a ec , we speci ied a simila eg ession model wi hou dummies o e lec CAR. In o de o inspec whe he he ime lag be ween he sali a sample and sel - epo had an e ec on a ec obse a ions, we es ed he model wi h wo ime indica o s, including β2ΔTime i (see Eq. 2) o explain posi i e o nega i e a ec , as ollows: A ec i =β0 +β1Time i +β2ΔTime i +β3Day i + υ 0i + ε 1i (2) In o de o answe ou main esea ch ques ion—how physiological s ess is ela ed o a ec in wi hin-le el—we speci ied a join MSEM model as shown in Fig. 1. A he wi hin-le el, we eg essed co isol on ime a e awakening and day, and we eg essed a ec on co isol, ime o sel - epo ed a ec since he co isol sample, and day. Co isol sam- pling ime a e awakening and ela i e ime o PANAS obse a ion we e allowed o co ela e. We used he maximum likelihood (ML) es ima o o he co isol only model, and co isol and posi i e a ec models; and we used ML wi h Mon e Ca lo in eg a ion o acili a e con e gence o modeling co isol and he bina y nega i e a ec ou come. Raw es ima ed pa ame e s and s anda d e o s a e epo ed in he Resul s sec ion, as we used nega i e a ec as a bina y dependen a iable. MPlus does no p o ide s an- da dized pa ame e es ima es o wo-le el models wi h wi hin le el eg essions on bina y ou comes. 3. Resul s Desc ip i e s a is ics o un ans o med co isol concen a ions and PANAS obse a ions a each imepoin as well as ime lags om awak- ening o co isol and he ime di e ences be ween PANAS and co isol measu emen s a e p esen ed in Table 1. Raw co isol concen a ions and he le els o eache s’ posi i e a ec o e he wo measu emen days a e shown in Fig. 3. We i s p esen indings o physiological s ess and posi i e and nega i e a ec sepa a ely, and hen p esen indings o posi i e and nega i e a ec eg essed on physiological s ess. 3.1. Teache s’ physiological s ess and a ec du ing he wo kday The co isol-only model (Eq. 1) indica ed a decline o co isol le els (B = − 0.11, p <.001 o ime a e awakening) and highe co isol Table 1 Desc ip i es o indica o s o each imepoin . Indica o Day 1 Day 2 N M o % SD ange N M o % SD ange Co isol (nmol/l) TP1 59 27.07 12.67 6–67 56 26.91 10.79 9–58 Co isol (nmol/l) TP2 58 41.85 11.78 13–67 58 40.60 14.05 12–69 Co isol (nmol/l) TP3 59 40.50 12.91 10–71 59 39.89 14.34 11–72 Co isol (nmol/l) TP4 60 13.75 9.15 5–62 56 10.77 5.32 4–31 Co isol (nmol/l) TP5 61 9.28 4.88 4–34 57 9.87 7.83 2–53 Co isol (nmol/l) TP6 56 3.79 1.91 1–11 52 5.78 9.72 1–70 Posi i e a ec TP1 48 3.62 0.75 1.8–5.0 53 3.31 0.84 1.6–5.0 Posi i e a ec TP4 36 4.03 0.55 2.5–4.8 41 4.07 0.78 1.2–5.0 Posi i e a ec TP5 54 3.85 0.71 2.2–5.0 49 3.65 0.88 1–5.0 Posi i e a ec TP6 51 2.70 0.97 1.0–5.0 49 2.66 0.94 1.2–4.6 Nega i e a ec a TP1 48 83 53 59 Nega i e a ec a TP4 36 72 41 51 Nega i e a ec a TP5 54 43 49 59 Nega i e a ec a TP6 51 43 49 31 Co isol ime b TP1 59 0.02 0.04 0.0–0.3 56 0.01 0.02 0–0.1 Co isol ime b TP2 58 0.50 0.06 0.3–0.8 58 0.51 0.07 0.3–0.8 Co isol ime b TP3 59 0.79 0.08 0.8–1.3 59 0.78 0.06 0.7–1.0 Co isol ime b TP4 60 3.90 0.48 3.1–5.7 56 4.03 0.84 3.0–7.8 Co isol ime b TP5 61 7.05 0.67 5.9–9.0 57 7.07 0.97 4.1–10.3 Co isol ime b TP6 56 16.01 0.82 13.9–17.6 52 15.98 1.14 13.4–19.5 PANAS ime c TP1 46 1.15 0.58 0.3–2.8 49 1.04 0.42 0.4–2.5 PANAS ime c TP4 36 0.33 0.52 -0.6–2.1 40 0.40 0.74 -0.8–2.6 PANAS ime c TP5 54 0.86 1.26 -0.9–5.6 46 0.39 0.82 -0.9–4.4 PANAS ime c TP6 48 -0.17 0.65 -2.4–1.0 42 -0.21 0.68 -1.9–1.0 No e. a Due o skewness o he scale-sco e (i.e., he mean o he i e i ems eco ded on 1–5 scales) nega i e a ec was ecoded in o 0 =all esponses a “1”, 1 =one o mo e esponses a “2” o abo e). b Time a e awakening o co isol sampling in hou s. c Time lag be ween he a ec sel - epo and he co isol sampling in hou s. TP = imepoin . A.-L. J˜ ogi e al. Psychoneu oendoc inology 149 (2023) 106028 6 le els o CAR imepoin s (B =0.71, p <.001 o bo h imepoin 2 and imepoin 3). Teache s’ co isol le els did no di e by measu emen day (B = − 0.02, p <.50). The co isol-only model was sa u a ed wi h he pe ec model i . In he posi i e a ec model (Eq. 2), eache s epo ed a lowe posi- i e a ec la e in he day (B = − 0.07, p <.001), and his did no depend on he ime lag o PANAS obse a ions om he co isol sampling (B = −0.05, p =.36). Posi i e a ec was lowe on he second measu emen day (B = − 0.51, p =.03). The posi i e a ec -only model was also sa u a ed wi h he pe ec model i like he co isol-only model. In he nega i e a ec model, he nega i e a ec was also lowe la e in he day (B = − 0.13, p <.001), and i did no depend on he ime lag om co isol sampling (B = − 0.30, p =.08). Simila o posi i e a ec , he eache s ended o epo ewe nega i e eelings on he second measu emen day (B = − 0.56, p =.03). 3.2. In ape sonal ela ions be ween physiological s ess and a ec Nex , we combined ini ial co isol-only and a ec -only models and eg essed si ua ional a ec on si ua ional co isol. Co isol was explained by con inuous ime o sampling a e awakening, wo dummy CAR imepoin s, and day (see Model 1 in Fig. 2). Posi i e o nega i e a ec was p edic ed by ime o co isol sample p io o he sel - epo a e awakening, he di e ence be ween he ime o PANAS obse a- ion and co isol sampling, and day. The same model was applied wice: once o posi i e a ec as an ou come and ano he o nega i e a ec . In bo h models, we ound he wi hin-le el e ec o physiological s ess on a ec in expec ed di ec ions. As o models using Mon e Ca lo in eg a- ion, he absolu e model i indices a e no a ailable; we could no examine he i o he nega i e a ec model. Ne e heless, he i indices o he posi i e a ec model did no demons a e good model i ( χ 2 =213.47, d =9, p <.001, CFI =0.86, RMSEA =0.18, SRMR wi hin =0.14, SRMR be ween =0.17) (Hu and Ben le , 1999). The e o e, we decided o emo e co isol imepoin s indica ing CAR ( imepoin s 2 and 3) om ou models and es ou hypo heses using he da a om ou imepoin s in a day. We also speci ied he ime sequence o ou si ua ion-speci ic da a collec ion by excluding bo h co isol and PANAS da a i co isol samples we e collec ed a e he PANAS obse - a ion o he co isol sampling ime was missing (N =74) om ou inal model (see Model 2 in Fig. 2, n =403 co isol samples and n =307 PANAS obse a ions). The model i indices o co isol and he posi i e a ec model we e e y good ( χ 2 =4.09, d =3, p =.25, CFI =0.99, RMSEA =0.03, SRMR wi hin =0.03, SRMR be ween =0.18). Simila o he co isol only model, eache s’ co isol le els did no di e be ween he i s and second sampling day (see Table 2). Time a e awakening nega i ely p edic ed he co isol le els, indica ing no mal physiological co isol decline o e he day. Si ua ional posi i e a ec was explained by ime and day o co isol sampling a e awakening, simila o he posi- i e a ec only model. In acco dance wi h Hypo hesis 1, he si ua ional co isol le el explained posi i e a ec in wi hin-le el (B = − 0.31, p=.001). The highe he eache s’ physiological s ess he less hey el posi i e a ec . Combining physiological s ess and nega i e a ec , we ound ha he p obabili y o eeling nega i e a ec di e ed on wo obse a ion days, as in he nega i e a ec -only model (see Table 2). Du ing he day, ime did no explain he p obabili y o eeling nega i e a ec any mo e in he combined model. The ime disc epancy om co isol sampling was ma ginally ela ed o nega i e a ec (B = − 0.38, p=.051) showing ha close he a ec obse a ion was o he co isol sampling, highe was he p obabili y o eeling nega i e a ec . Ou main in e es was he in aindi idual associa ion be ween physiological s ess and nega i e a ec . We ound a posi i e ela ion be ween hese wo indica o s (B =0.94, p =.004, odds a io =2.55, 95% CI-s[1.35, 4.82]). The highe he eache s’ physiological s ess he highe was he p oba- bili y o eeling nega i e a ec a he same ime. 4. Discussion We in es iga ed eache s’ daily physiological s ess and a ec and he e ec o physiological s ess on a ec in a si ua ion. Ou s udy o e s wo no el con ibu ions o he ield. Fi s , we applied a mul ile el SEM model o in es iga e wo ime- a ying indica o s concu en ly. Second, we showed ha a he in ape sonal le el, measu emen ime accoun ed o bo h s ess and a ec indica o s; eache s’ physiological s ess has an e ec on bo h posi i e and nega i e a ec du ing he wo kday. 4.1. Modeling ime- a ying physiological s ess and a ec Ha ing in aindi idual ela ions in physiological s ess in in e es , MLMs ha e been he mos p e alen analysis me hod ( o example see Doane and Adam, 2010). Un o una ely, MLMs do no allow o mo e han one dependen a iable in he model, no do hey p o ide model i indices o es ing how consis en he model is wi h he da a. The e o e, we p oposed a mul ile el SEM model o in es iga e he wi hin-le el ela ions be ween wo ime- a ying indica o s—physiological s ess measu ed by sali a y co isol, and si ua ional a ec . I was qui e easonable o expec ha he physiological s ess and a ec measu e- men s would no be exac ly ime complian , we needed o use wo ime indica o s in ou s a is ical models. Fo a oiding mul icollinea i y o ime indica o s ha had inc eased he Type II e o a es (G ewal e al., 2004) we p oposed using wo ime indica o s o PANAS obse a- ions— he ime o co isol sampling a e awakening and he ime o he sel - epo ed a ec a e he co isol sampling. Nex , as sample size and esul ing low powe is o en p oblema ic issue in s udies collec ing physiological da a (Adam and Kuma i, 2009), we modeled all a ailable da a wi hou agg ega ion, i.e., up o 12 co isol and 8 a ec epo s pe pe son. In line wi h he cu en esea ch, eache s’ co isol le els do no di e by wo kday (Belling a h e al., 2008). Inco po a ing co isol samples om di e en ime poin s om one day and/o om days o he same analysis le el has so a been used in la en ai models (Doane e al., 2015; Mille e al., 2016). Ou esul s Table 2 Pa ame e es ima es o wo-le el SEM models o ela ionships be ween co isol and a ec . Pa ame e Posi i e a ec Nega i e a ec Es ima e SE p- alue Es ima e SE p- alue Wi hin-pe son le el Co isol on: Co isol ime a -0.12 0.01 <0.001 -0.12 0.01 <0.001 Day 0.00 0.05 0.99 0.00 0.05 0.99 A ec on: Co isol ime a -0.10 0.01 <0.001 -0.01 0.05 0.98 Δ PANAS - Co isol ime b -0.02 0.06 0.76 -0.38 0.19 0.05 Day -0.15 0.08 0.04 -0.66 0.27 0.02 Co isol -0.31 0.08 <0.001 0.94 0.32 0.01 Residual a iances Co isol 0.22 0.02 <0.001 0.22 0.02 <0.001 A ec 0.40 0.04 <0.001 NA NA NA Be ween-pe son le el Means/ h eshold Co isol 3.03 0.05 <0.001 3.03 0.05 <0.001 A ec 3.90 0.12 <0.001 1.26 1.00 0.21 Va iances Co isol 0.07 0.02 <0.001 0.07 0.02 <0.001 A ec 0.39 0.90 <0.001 0.90 0.48 0.06 Co a iance Co isol wi h a ec 0.02 0.03 0.42 -0.07 0.07 0.30 No e. a Time a e awakening o co isol sampling in hou s. b Time lag be ween he a ec sel - epo and he co isol sampling in hou s. NA =no a ailable o bina y ou come in MPlus. A.-L. J˜ ogi e al. Psychoneu oendoc inology 149 (2023) 106028 7 also showed ha eache s’ co isol le els a di e en ime-poin s did no a y on he wo measu emen days. S ill, bo h posi i e and nega i e a ec we e lowe on he second measu emen day. Taking he modeling pa oge he , ou esul s i s showed he ex- pec ed physiological co isol cu e du ing he day, and also in ape - sonal changes in posi i e and nega i e a ec o e wo measu emen days. This allowed us o me ge wo ime- a ying models in o de o in es iga e he in ape sonal ela ions be ween eache s’ physiological s ess and a ec . 4.2. Rela ions be ween eache s’ si ua ional physiological s ess and a ec Nex , ou esul s shed ligh on he wi hin-le el ela ions be ween eache s’ physiological s ess and a ec du ing he wo kday in au hen ic class oom se ings. Labo a o y s udies o heal hy adul s ha e so a indica ed a weak ela ionship be ween physiological s ess and emo ional esponse (Campbell and Ehle , 2012). Recen ly, i has been p oposed ha associa ions be ween physiological and psychological indica o s migh be he e ogenous in he sample (Simon e al., 2022). In mos s udies conduc ed in ambula o y se ings hus a , adul samples ha e shown nega i e ela ions be ween si ua ional physiological s ess and posi i e a ec , and posi i e ela ions be ween si ua ional physio- logical s ess and nega i e a ec (Joseph e al., 2021). Howe e , eache s’ physiological s ess s udies a e s ill in hei i s s eps. I has been shown, o example, ha eache s’ physiological s ess is highe on wo k days compa ed o he weekend (We s ein e al., 2020) o ha eache s uden s expe ience physiological s ess esponse du ing he labo a o y s ess es bu no emo ional esponse (Becke e al., 2022). Consis en wi h s udies on he ela ionship be ween- eache s sel - epo ed s ess and a ec (e.g. Hamama e al., 2013) and wi hin-pe son ela ions be ween physiological s ess and a ec (Joseph e al., 2021), ou indings con i med ha a he wi hin-pe son le el, i eache s ha e highe physiological s ess, hey also ypically ha e lowe posi i e a ec . And, ice e sa o nega i e a ec — he highe he si u- a ional physiological s ess, he highe he nega i e a ec . The e o e, ou s udy adds aluable si ua ional in o ma ion o p e- ious esea ch abou eache s’ s ess and emo ions in he class oom (e.g F enzel e al., 2020; Kelle e al., 2014). Teaching is a e y demanding occupa ion in which s ess is conside ed o be a de aul ea u e (B ough on, 2010). Teache s’ s ess depends on se e al ex e nal (e.g s uden s’ beha io , wo kload, ela ions in o ganiza ion) and in e nal (e. g coping skills, emo ions) ac o s (Mon gome y and Rupp, 2005). Physiological s ess is ela ed o he class oom en i onmen and each- ing p ac ices a he in ape sonal le el (Junke e al., 2021). As eache s’ a ec is highly si ua ion-speci ic (Kelle e al., 2014) i is c ucial o g asp he momen a y ela ions o s ess and a ec . Unde s anding in ape - sonal ela ions be ween physiological s ess and a ec emphasizes he necessi y o acili a e educa o s’ awa eness o hei own s ess and a ec . I also helps wi h planning indi idual in e en ions o diminish he nega i e pe cep ions o s ess ul e en s, and suppo eache s’ s ess managemen skills and each eache ’s s ess egula ion du ing and a e he school day. Teache s’ well-being can be suppo ed, o example, by pee men o ing g oups ocused on de eloping hei pedagogical p ac- ices in he class oom (Vasalampi e al., 2021). The impo an ole o p ima y school p incipals on eache s’ well-being is also well ecognized (e.g., Ael e man e al., 2007). Fo example, p incipals hold he key in gi ing eache s’ oppo uni ies o in-se ice aining and supe ision (B edeson and Johansson, 2000). Ou esul s open se e al possible di ec ions o u he esea ch. Fi s , as ou ocus was on in ape sonal ela ionships be ween physio- logical s ess and a ec , we did no in es iga e possible be ween- eache o -class oom le el p edic o s o s ess and a ec and he indi idual di e ences in change in s ess and a ec du ing he day. In u he s udies, o example, he e ec o eache s’ wo k expe ience, numbe o s uden s in he class oom, and he p opo ion o s uden s wi h beha - io al o lea ning di icul ies on eache s’ physiological s ess and a ec should be in es iga ed. Nex , as eache s’ sel - epo ed s ess is ela ed o hei class oom p ac ices (Pen inen e al., 2020), i is easonable o in es iga e he e ec o physiological s ess on he quali y o class oom in e ac ions and ins uc ion and, u he , on s uden ou comes. P omp- ed by he esul s o he in e ac ions be ween physiological a ousal and class oom en i onmen on eache s’ enjoymen (Donke e al., 2020), he possible mode a o s on he associa ions be ween physiological s ess and a ec in he class oom should be s udied. 4.3. Limi a ions Ou s udy also has some clea limi a ions we need o aise. We ha e s udied si ua ional s ess and a ec ideally measu ed simul aneously. In ou s udy conduc ed in non-labo a o y se ings, he e a e o en dis- c epancies in he iming o co isol sampling and a ec obse a ion. Fi s , we ha e aken his in o accoun in ou model by adding he ime di e ence indica o o explain si ua ional a ec . This di ec ed us o exclude co isol da a ha we e sampled a e he sel - epo ed a ec (PANAS) and a ec epo s ha we e answe ed be o e he linked co isol sampling o enable speci ying he di ec ion o he ela ionship be ween physiological s ess and a ec in he MSEM model. Al hough we ha e speci ied he di ec ion o co isol le els o explain he a iance o a ec , we do ecognize ha he e is also heo e ical suppo o he opposi e eg ession model and call o u he in es iga ions abou he di ec ion o he ela ions be ween ime- a ying s ess and a ec . Fu he mo e, ou solu ion inc eases he amoun o missingness in ou da a se . Secondly, we combined da a om he same pa icipan s sampled and obse ed on di e en days. This was he solu ion we used o add ess he lack o powe p oblem, which o en occu s in ambula o y s udies col- lec ing physiological da a. S ill, we a e e y open o discussion abou he solu ion we p oposed. As a hi d limi a ion, da a collec ed wi h he nega i e a ec scale was much skewed and we we e no able o use nega i e a ec as a con inuous a iable as in he o iginal PANAS scale. We also ha e o no e he limi a ion o inding ha eache s’ a ec , bo h posi i e and nega i e, was lowe on Day 2. On Day 1 eache s we e obse ed by he esea ch assis an s, he da a collec ion Day 2 was a school day as usual. The e o e, we canno ell he eason o lowe a ec on Day 2, we can only say ha he day was con olled in ou models. We also used single co isol samples o wi hin-pe son analysis and no any combined co isol indica o . This makes i impossible o d aw any be ween-pe son conclusions abou he unc ionali y o ou sample’s HPA-axis. 4.4. Conclusions Taken oge he , de elopmen s in in aindi idual esea ch in educa- ion and he inc eased a ailabili y o so wa e pose an in e es ing u u e o his ield o esea ch. We ha e shown one op ion o modeling wo o mo e ime- a ying indica o s in a mul ile el SEM amewo k, and we hope o con inue academic discussions abou combining di e en in- dica o s in in aindi idual esea ch. We also showed ha despi e eache s’ a e age le els o physiological s ess and a ec , hei co isol le els and posi i e and nega i e a ec a e ela ed a a si ua ional le el. Ou indings emphasize he need o pay a en ion o each eache ’s s ess and a ec egula ion skills. Decla a ion o in e es None. Acknowledgemen s This s udy was unded by g an s om he Finnish Wo k En i onmen Fund (2017–2020 #117142), he Academy o Finland (2018–2022, #317610), and he Facul y o Educa ion and Psychology, Uni e si y o Jy ¨ askyl¨ a, Finland. The au ho s hank he EARLI Eme ging Field G oup, A.-L. J˜ ogi e al. Psychoneu oendoc inology 149 (2023) 106028 8 “The po en ial o biophysiology o unde s anding lea ning and eaching expe iences,” o b inging us oge he and acili a ing inspi ing discussions. Re e ences Adam, E.K., Kuma i, M., 2009. Assessing sali a y co isol in la ge-scale, epidemiological esea ch. Psychoneu oendoc inology 34, 1423–1436. h ps://doi.o g/10.1016/j. psyneuen.2009.06.011. Ael e man, A., Engels, N., Van Pe egem, K., Pie e Ve haeghe, J., 2007. The well-being o eache s in Flande s: he impo ance o a suppo i e school cul u e. Educ. S ud. 33, 285–297. h ps://doi.o g/10.1080/03055690701423085. Ald up, K., Klusmann, U., Lüd ke, O., G¨ ollne , R., T au wein, U., 2018. S uden misbeha io and eache well-being: es ing he media ing ole o he eache - s uden ela ionship. Lea n. Ins . 58, 126–136. h ps://doi.o g/10.1016/j. lea nins uc.2018.05.006. Ali, N., Ni schke, J.P., Coope man, C., P uessne , J.C., 2017. Supp essing he endoc ine and au onomic s ess sys ems does no impac he emo ional s ess expe ience a e psychosocial s ess. Psychoneu oendoc inology 78, 125–130. h ps://doi.o g/ 10.1016/j.psyneuen.2017.01.015. A ens, A.K., Mo in, A.J.S., 2016. Rela ions be ween eache s’ emo ional exhaus ion and s uden s’ educa ional ou comes. J. Educ. Psychol. 108, 800–813. h ps://doi.o g/ 10.1037/edu0000105. Aspa ouho , T., Mu h´ en, B., 2020. Compa ison o models o he analysis o in ensi e longi udinal da a. S uc . Equ. Model. 27, 275–297. h ps://doi.o g/10.1080/ 10705511.2019.1626733. Becke , S., Spina h, B., Di zen, B., D¨ o le , T., 2022. Psychological s ess =physiological s ess? J. Psychophysiol. h ps://doi.o g/10.1027/0269-8803/a000301. Belling a h, S., Weigl, T., Kudielka, B.M., 2008. Co isol dys egula ion in school eache s in ela ion o bu nou , i al exhaus ion, and e o – ewa d-imbalance. Biol. Psychol. 78, 104–113. h ps://doi.o g/10.1016/j.biopsycho.2008.01.006. B edeson, P.V., Johansson, O., 2000. The school p incipal’s ole in eache p o essional de elopmen . J. Se . Educ. 26, 385–401. h ps://doi.o g/10.1080/ 13674580000200114. B ough on, A., 2010. Wo k- ela ed s ess. Eu opean Founda ion o he Imp o emen o Li ing and Wo king Condi ions. 〈h ps://www.eu o ound.eu opa.eu/publica ion s/ epo /2010/wo k- ela ed-s ess〉. Campbell, J., Ehle , U., 2012. Acu e psychosocial s ess: does he emo ional s ess esponse co espond wi h physiological esponses. Psychoneu oendoc inology 37, 1111–1134. h ps://doi.o g/10.1016/j.psyneuen.2011.12.010. Chen, S.H., Kennedy, M., Zhou, Q., 2012. Pa en s’ exp ession and discussion o emo ion in he mul ilingual amily. Pe spec . Psychol. Sci. 7, 365–383. h ps://doi.o g/ 10.1177/1745691612447307. C aw o d, J.R., Hen y, J.D., 2004. The posi i e and nega i e a ec schedule (PANAS): cons uc alidi y, measu emen p ope ies and no ma i e da a in a la ge non- clinical sample. B . J. Clin. Psychol. 43, 245–265. h ps://doi.o g/10.1348/ 0144665031752934. Cu an, P.J., Baue , D.J., 2011. The disagg ega ion o wi hin-pe son and be ween-pe son e ec s in longi udinal models o change. Annu. Re . Psychol. 62, 583–619. h ps:// doi.o g/10.1146/annu e .psych.093008.100356. Diene , E., P essman, S.D., Hun e , J., Delgadillo-Chase, D., 2017. I , why, and when subjec i e well-being in luences heal h, and u u e needed esea ch. Appl. Psychol. Heal h Well-Being 9, 133–167. h ps://doi.o g/10.1111/aphw.12090. Doane, L.D., Adam, E.K., 2010. Loneliness and co isol: momen a y, day- o-day, and ai associa ions. Psychoneu oendoc inology 35, 430–441. h ps://doi.o g/10.1016/j. psyneuen.2009.08.005. Doane, L.D., Chen, F.R., Sladek, M.R., Van Len en, S.A., G ange , D.A., 2015. La en ai co isol (LTC) le els: eliabili y, alidi y, and s abili y. Psychoneu oendoc inology 55, 21–35. h ps://doi.o g/10.1016/j.psyneuen.2015.01.017. Donke , M.H., an Gog, T., Goe z, T., Roos, A.-L., Mainha d, T., 2020. Associa ions be ween eache s’ in e pe sonal beha io , physiological a ousal, and lesson- ocused emo ions. Con emp. Educ. Psychol. 63, 101906 h ps://doi.o g/10.1016/j. cedpsych.2020.101906. F ed ickson, B.L., 2001. The ole o posi i e emo ions in posi i e psychology: he b oaden-and-build heo y o posi i e emo ions. Am. Psychol. 56, 218–226. h ps:// doi.o g/10.1037//0003-066x.56.3.218. F enzel, A.C., Fiedle , D., Ma x, A.K.G., Reck, C., Pek un, R., 2020. Who enjoys eaching, and when? Be ween- and wi hin-pe son e idence on eache s’ app aisal-emo ion links. F on . Psychol. 11. h ps://doi.o g/10.3389/ psyg.2020.01092. G ewal, R., Co e, J.A., Baumga ne , H., 2004. Mul icollinea i y and measu emen e o in s uc u al equa ion models: implica ions o heo y es ing. Ma k. Sci. 23, 519–529. h ps://doi.o g/10.1287/mksc.1040.0070. Hamake , E., 2012. Why esea che s should hink wi hin-pe son: a pa adigma ic iew. In: Mehl, M.R., Conne , T.S. (Eds.), Handbook o Resea ch Me hods o S udying Daily Li e. The Guil o d P ess, New Yo k; London, pp. 43–61. Hamama, L., Ronen, T., Shacha , K., Rosenbaum, M., 2013. Links be ween s ess, posi i e and nega i e a ec , and li e sa is ac ion among eache s in special educa ion schools. J. Happiness S ud. 14, 731–751. h ps://doi.o g/10.1007/s10902-012-9352-4. Hie alah i, M., Ran anen, J., Kokko, K., 2016. Do li e goals media e he link be ween pe sonali y ai s and men al well-being? J. Happiness Well-Being 4, 72–89. Hoppmann, C.A., Klumb, P.L., 2006. Daily goal pu sui s p edic co isol sec e ion and mood s a es in employed pa en s wi h p eschool child en. Psychosom. Med. 68, 887–894. h ps://doi.o g/10.1097/01.psy.0000238232.46870. 1. Hoy , L.T., Zeide s, K.H., Eh lich, K.B., Adam, E.K., 2016. Posi i e upsho s o co isol in e e yday li e. Emo ion 16, 431–435. h ps://doi.o g/10.1037/emo0000174. Hu, L., Ben le , P.M., 1999. Cu o c i e ia o i indexes in co a iance s uc u e analysis: con en ional c i e ia e sus new al e na i es. S uc . Equ. Model. 6, 1–55. h ps:// doi.o g/10.1080/10705519909540118. Johnson, S., Coope , C., Ca w igh , S., Donald, I., Taylo , P., Mille , C., 2005. The expe ience o wo k- ela ed s ess ac oss occupa ions. J. Manag. Psychol. 20, 178–187. h ps://doi.o g/10.1108/02683940510579803. Joseph, N.T., Jiang, Y., Zilioli, S., 2021. Momen a y emo ions and sali a y co isol: a sys ema ic e iew and me a-analysis o ecological momen a y assessmen s udies. Neu osci. Biobeha . Re . 125, 365–379. h ps://doi.o g/10.1016/j. neubio e .2021.02.042. Junke , R., Donke , M.H., Mainha d, T., 2021. Po en ial class oom s esso s o eache s: an audio isual and physiological app oach. Lea n. Ins . 75, 101495 h ps://doi.o g/ 10.1016/j.lea nins uc.2021.101495. Ka z, D.A., G eenbe g, M.T., Jennings, P.A., Klein, L.C., 2016. Associa ions be ween he awakening esponses o sali a y α -amylase and co isol wi h sel - epo indica o s o heal h and wellbeing among educa o s. Teach. Teach. Educ. 54, 98–106. h ps://doi. o g/10.1016/j. a e.2015.11.012. Ka z, D.A., Ha is, A., Abena oli, R., G eenbe g, M.T., Jennings, P.A., 2018. Educa o s’ emo ion egula ion s a egies and hei physiological indica o s o ch onic s ess o e 1 yea . S ess Heal h 34, 278–285. h ps://doi.o g/10.1002/smi.2782. Kelle , M.M., Chang, M.-L., Becke , E.S., Goe z, T., F enzel, A.C., 2014. Teache s’ emo ional expe iences and exhaus ion as p edic o s o emo ional labo in he class oom: an expe ience sampling s udy. F on . Psychol. 5. h ps://doi.o g/ 10.3389/ psyg.2014.01442. de Kloe , E.R., V eugdenhil, E., Oi zl, M.S., Jo¨ els, M., 1998. B ain co icos e oid ecep o balance in heal h and disease1. Endoc . Re . 19, 269–301. h ps://doi.o g/10.1210/ ed .19.3.0331. Kudielka, B.M., Gie ens, A., Hellhamme , D.H., Wüs , S., Schlo z, W., 2012. Sali a y co isol in ambula o y assessmen —some dos, some don’ s, and some open ques ions. Psychosom. Med. 74, 418–431. h ps://doi.o g/10.1097/ psy.0b013e31825434c7. Le kkanen, M.-K., Paka inen, E., 2016 2022. In: Teache and S uden S ess and In e ac ion in Class oom (TESSI). h ps://doi.o g/10.17011/jyx/ da ase /77741. Li, P.-H., Maye , D., Malmbe g, L.-E., 2022. Teache well-being in he class oom: a mic o- longi udinal s udy. Teach. Teach. Educ. 115, 103720 h ps://doi.o g/10.1016/j. a e.2022.103720. Lyubomi sky, S., King, L., Diene , E., 2005. The bene i s o equen posi i e a ec : does happiness lead o success. Psychol. Bull. 131, 803–855. h ps://doi.o g/10.1037/ 0033-2909.131.6.803. Mille , K.G., W igh , A.G.C., Pe e son, L.M., Kama ck, T.W., Ande son, B.A., Ki schbaum, C., Ma sland, A.L., Muldoon, M.F., Manuck, S.B., 2016. T ai posi i e and nega i e emo ionali y di e en ially associa e wi h diu nal co isol ac i i y. Psychoneu oendoc inology 68, 177–185. h ps://doi.o g/10.1016/j. psyneuen.2016.03.004. Mille , R., Plessow, F., Rauh, M., G ¨ oschl, M., Ki schbaum, C., 2013. Compa ison o sali a y co isol as measu ed by di e en immunoassays and andem mass spec ome y. Psychoneu oendoc inology 38, 50–57. h ps://doi.o g/10.1016/j. psyneuen.2012.04.019. Molenaa , P.C.M., 2004. A mani es o on psychology as idiog aphic science: b inging he pe son back in o scien i ic psychology, his ime o e e . Meas. In e discip. Res. Pe spec . 2, 201–218. h ps://doi.o g/10.1207/s15366359mea0204_1. Mon gome y, C., Rupp, A.A., 2005. A me a-analysis o explo ing he di e se causes and e ec s o s ess in eache s. Can. J. Educ. 28, 458. h ps://doi.o g/10.2307/ 4126479. Mu hen, L.K., Mu h´ en, B.O., 1998–2017. Mplus Use ’s Guide. Eigh h Edi ion. Mu hen & Mu hen, LA. Nislin, M., Sajaniemi, N., Sims, M., Suhonen, E., Maldonado, E.F., Hy inen, S., Hi onen, A., 2016. Occupa ional well-being and s ess among ea ly childhood p o essionals: he use o an inno a i e s a egy o measu e s ess eac i i y in he wo kplace. Open Re . Educ. Res. 3, 1–17. h ps://doi.o g/10.1080/ 23265507.2015.1128352. Pen inen, V., Paka inen, E., on Suchodole z, A., Le kkanen, M.-K., 2020. Rela ions be ween kinde ga en eache s’ occupa ional well-being and he quali y o eache - child in e ac ions. Ea ly Educ. De . 31, 994–1010. h ps://doi.o g/10.1080/ 10409289.2020.1785265. Polk, D.E., Cohen, S., Doyle, W.J., Skone , D.P., Ki schbaum, C., 2005. S a e and ai a ec as p edic o s o sali a y co isol in heal hy adul s. Psychoneu oendoc inology 30, 261–272. h ps://doi.o g/10.1016/j.psyneuen.2004.08.004. Schlo z, W., 2019. In es iga ing associa ions be ween momen a y s ess and co isol in daily li e: wha ha e we lea ned so a . Psychoneu oendoc inology 105, 105–116. h ps://doi.o g/10.1016/j.psyneuen.2018.11.038. Schneide man, N., I onson, G., Siegel, S.D., 2005. S ess and heal h: psychological, beha io al, and biological de e minan s. Annu. Re . Clin. Psychol. 1, 607–628. h ps://doi.o g/10.1146/annu e .clinpsy.1.102803.144141. Simon, L., Rab, S.L., Golds ein, P., Magal, N., Admon, R., 2022. Mul i- ajec o y analysis unco e s la en associa ions be ween psychological and physiological acu e s ess esponse pa e ns. Psychoneu oendoc inology, 105925. h ps://doi.o g/10.1016/j. psyneuen.2022.105925. A.-L. J˜ ogi e al.