Zhang, Lixia; Li, Shao ing; Ren, Yanjun
A icle — Published Ve sion
Does in e ne use bene i he men al heal h o olde
adul s? Empi ical e idence om he China heal h and
e i emen longi udinal s udy
Heliyon
P o ided in Coope a ion wi h:
Leibniz Ins i u e o Ag icul u al De elopmen in T ansi ion Economies (IAMO), Halle (Saale)
Sugges ed Ci a ion: Zhang, Lixia; Li, Shao ing; Ren, Yanjun (2024) : Does in e ne use bene i he
men al heal h o olde adul s? Empi ical e idence om he China heal h and e i emen longi udinal
s udy, Heliyon, ISSN 2405-8440, Else ie , Ams e dam, Vol. 10, Iss. 3, pp. 1-15,
h ps://doi.o g/10.1016/j.heliyon.2024.e25397 ,
h ps://www.sciencedi ec .com/science/a icle/pii/S2405844024014282
This Ve sion is a ailable a :
h ps://hdl.handle.ne /10419/282018
S anda d-Nu zungsbedingungen:
Die Dokumen e au EconS o dü en zu eigenen wissenscha lichen
Zwecken und zum P i a geb auch gespeiche und kopie we den.
Sie dü en die Dokumen e nich ü ö en liche ode komme zielle
Zwecke e iel äl igen, ö en lich auss ellen, ö en lich zugänglich
machen, e eiben ode ande wei ig nu zen.
So e n die Ve asse die Dokumen e un e Open-Con en -Lizenzen
(insbesonde e CC-Lizenzen) zu Ve ügung ges ell haben soll en,
gel en abweichend on diesen Nu zungsbedingungen die in de do
genann en Lizenz gewäh en Nu zungs ech e.
Te ms o use:
Documen s in EconS o may be sa ed and copied o you pe sonal
and schola ly pu poses.
You a e no o copy documen s o public o comme cial pu poses, o
exhibi he documen s publicly, o make hem publicly a ailable on he
in e ne , o o dis ibu e o o he wise use he documen s in public.
I he documen s ha e been made a ailable unde an Open Con en
Licence (especially C ea i e Commons Licences), you may exe cise
u he usage igh s as speci ied in he indica ed licence.
h p://c ea i ecommons.o g/licenses/by-nc-nd/4.0/
Heliyon 10 (2024) e25397
A ailable online 1 Feb ua y 2024
2405-8440/© 2024 Published by Else ie L d. This is an open access a icle unde he CC BY-NC-ND license
(h p://c ea i ecommons.o g/licenses/by-nc-nd/4.0/).
Resea ch a icle
Does in e ne use bene i he men al heal h o olde adul s?
Empi ical e idence om he China heal h and e i emen
longi udinal s udy
Lixia Zhang
a
,
b
, Shao ing Li
a
, Yanjun Ren
a
,
c
,
*
a
College o Economics and Managemen , No hwes A&F Uni e si y, 3 Taicheng Rd, Yangling 712100, Shaanxi, China
b
Six Indus ial Resea ch Ins i u e, No hwes A&F Uni e si y, 3 Taicheng Rd, Yangling 712100, Shaanxi, China
c
Leibniz Ins i u e o Ag icul u al De elopmen in T ansi ion Economies (IAMO), Theodo -Liese -S . 2, 06120, Halle (Saale), Ge many
ARTICLE INFO
Keywo ds:
In e ne use
Men al heal h
Olde adul s
China
ABSTRACT
The men al heal h (MH) o olde adul s is a p ominen public heal h conce n. Howe e , esea ch
ega ding he impac o eme ging In e ne use on MH among olde adul s emains limi ed,
pa icula ly in ansi ional economies expe iencing a apidly aging popula ion such as China.
Thus, o add ess his esea ch gap, his s udy uses da a om he 2013–2018 wa es o he China
Heal h and Re i emen Longi udinal S udy. To in es iga e he causal ela ionship be ween
In e ne use and MH among olde adul s and explo e he unde lying channels h ough which his
ela ionship ope a es. The esul s e eal a no able posi i e associa ion be ween In e ne use and
MH among olde adul s. Fu he mo e, he s udy highligh s social in e ac ion, social us , a -
eling expenses, and heal hy habi s as c ucial channels h ough which In e ne use can impac MH
among olde adul s. The analysis also e eals how In e ne use demons a es a s onge posi i e
e ec on olde indi iduals who ha e ewe ch onic diseases and li e wi h hei o sp ing
compa ed wi h hei coun e pa s. These indings ha e signi ican policy implica ions, which hus
emphasizes he need o enhance In e ne use among olde adul s as a means o imp o ing hei
MH.
1. In oduc ion
An inc easingly aging popula ion has eme ged as a subs an ial challenge ac oss he wo ld; consequen ly, p omo ing heal hy aging
o main ain a high quali y o li e and independence has become a global conce n. The p opo ion o indi iduals aged 65 yea s and
abo e in de eloped egions is p ojec ed o ise om 14.3 % in 2000 o 25.9 % in 2050 [1]. Wi h he ad ancemen o In e ne ech-
nology, his p opo ion will g ow apidly. The olde popula ion becomes inc easingly ulne able o a ious heal h isks and diseases as
hei li e expec ancy inc eases [2]. The exis ing s a is ics e eal ha he p e alence a es o ch onic diseases, disabili y a es, and he
pe capi a consump ion o heal h esou ces among olde adul s a e 3.2, 3.6, and 1.9 imes highe han hose o he o al popula ion,
espec i ely [3]. This si ua ion is pa icula ly challenging o China because o i s apid popula ion g ow h compa ed wi h o he
de eloping coun ies. By he end o 2020, he popula ion aged 60 o abo e in China will each 264 million, wi h an aging a e o 18.70
% [4]. The cu en p ojec ions indica e ha he p opo ion o he popula ion o e 60 yea s old in China will each 34.9 % by 2050,
* Co esponding au ho . College o Economics and Managemen , No hwes A&F Uni e si y, 3 Taicheng Rd, Yangling 712100, Shaanxi, China.
E-mail add ess: [email p o ec ed] (Y. Ren).
Con en s lis s a ailable a ScienceDi ec
Heliyon
jou nal homepage: www.cell.com/heliyon
h ps://doi.o g/10.1016/j.heliyon.2024.e25397
Recei ed 19 June 2023; Recei ed in e ised o m 22 Janua y 2024; Accep ed 25 Janua y 2024
Heliyon 10 (2024) e25397
2
ushe ing he coun y in o a s age o p o ound aging. Olde adul s ha e placed conside able p essu e on he heal hca e sys em [5,6].
Howe e , men al heal h (MH) issues a e mo e se ious and less no iceable han physical heal h p oblems in olde adul s. S a is ically,
hal o he adul s aged 65–84 yea s old ha e expe ienced MH issues h oughou hei li es, including dep ession, loneliness, anxie y,
and s ess [7]. Poo dep ession signi ican ly a ec s he quali y o li e o olde adul s and is associa ed wi h an ele a ed isk o mo ali y
and mo bidi y [8]. The e o e, explo ing e ec i e s a egies o enhance he MH o olde adul s is essen ial.
A p esen , he In e ne has become inc easingly popula among olde adul s ollowing he de elopmen o echnology and he
popula i y o sma phones. Fo ins ance, 7.3 million senio adul s (1.3 % o he senio popula ion) used he In e ne in 2009, and he
numbe s wen up o 940 million, o which olde use s (i.e., hose o e 60 yea s old) accoun o 10.3 % [9,10]. The In e ne pene-
a ion a e o he olde adul popula ion aged 60 yea s and abo e is 43.2 %. In e ne use has subs an ial e ec s on eemploymen ,
income enhancemen , die a y imp o emen , and CO
2
emissions among olde adul s [11–14]. In e ne use-based in e en ions a e
p omising in e ms o suppo ing he heal h o olde adul s and ha e p o en o be e ec i e in acili a ing li es yle changes and disease
managemen in coun ies such as Aus alia, he Ne he lands, and Swi ze land [15]. Thus, such in e en ions migh cons i u e a
p omising app oach o b idging he exis ing gaps in MH ca e o olde adul s in China [16].
Thus a , he exis ing esea ch has p o ided some co ela ional e idence bu has ye o a i e a consis en conclusions. Mos s udies
ha e eliably demons a ed ha In e ne use o e s a con enien and e ec i e means o communica ion among olde adul s, he eby
enhancing a ious aspec s o hei well-being [17–21]. Fo example, using OLS eg essions, Zhang e al. [22] ound ha In e ne use is
signi ican ly associa ed wi h be e MH o olde Chinese esiden s. Clinical da a also indica e ha In e ne use has been e ec i e in
educing anxie y le els among pa ien s wi h ca diomyopa hy (Min o e al. [18]. The eason o such indings is ha In e ne use could
inc ease li e sa is ac ion, happiness, and o e all well-being while concu en ly educing eelings o loneliness, anxie y, s ess, and
dep essi e symp oms [23]. Fu he mo e, he In e ne p o ides an e icien and con enien channel h ough which olde adul s can
communica e wi h hei iends, ul ill hei spi i ual needs, enhance hei cogni i e abili ies, and educe hei likelihood o expe i-
encing dep ession [24]. Howe e , some s udies ha e shown ha In e ne use has nega i e implica ions o olde adul s in ha i
educes social pa icipa ion, leads o a na owing o social ci cles, and weakens iendship ne wo ks and he sense o communi y
belonging [4,20,25].
Al hough exis ing s udies ha e analyzed he impac o In e ne use on MH, ew empi ical es s ha e been conduc ed on he speci ic
channels by which In e ne use imp o es MH and educes he isk o dep ession. Social pa icipa ion se es as an in e media y ac o in
he ela ionship be ween In e ne use and he MH o olde adul s [26–28]. Empi ically, In e ne use no only imp o es he sel - epo ed
heal h, MH, and social adap a ion o olde adul s bu also enhances hei social pa icipa ion. In pa icula , In e ne use p omo es
communi y engagemen among olde adul s, he eby enhancing hei li e sa is ac ion and o e all heal h. Fo ins ance, Heo e al. [29]
epo ha In e ne use educes loneliness and enhances he li e sa is ac ion and well-being o olde adul s wi h social suppo ac ing as
a media ing a iable. O he media ing e ec s ha e also been p oposed. Fo example, Co en e al. [30] demons a e ha household size
media es he posi i e impac o In e ne use among e i ed adul s in he Uni ed S a es, especially o hose li ing alone. Simila ly, Yuan
[31] e eals ha olde adul s who engage in equen In e ne use a e less likely o expe ience MH issues and iden i y he mode a ing
e ec s o ch onic diseases and household income on his ela ionship.
The li e a u e e iew e eals ha he e is no cu en consensus on he in luence o In e ne use on he MH o olde adul s, and some
limi a ions a e in ol ed. Fi s , mos s udies ha e ocused on de eloped coun ies, such as he Uni ed S a es and New Zealand, and ha e
paid limi ed a en ion o de eloping coun ies. Such s udies ha e also been based on small samples, he eby limi ing he s a is ical
sophis ica ion and obus ness o he indings. Second, while ea lie s udies ha e sepa a ely ocused on in luence channels by
conside ing he impac o In e ne use on social pa icipa ion, social suppo , o income, insu icien a en ion has been paid o social
us and heal h habi s. P e ious s udies also did no conside hese aspec s wi hin he same amewo k o analyze he po en ial
channels o in luence. Thi d, conce ning esea ch me hodology, he exis ing s udies ha e mainly ocused on simple desc ip i e s a-
is ics and mul iple linea eg ession analysis, which ha e been c i icized o hei es ima ion e ec i eness, especially in e ms o
endogenei y and he e ogenei y. Finally, mos s udies ea olde adul s as a homogeneous g oup and epo a ied, e en con lic ing,
ou comes. Few s udies ha e explo ed he une en impac o In e ne use on subg oups wi h di e en MH condi ions.
The e o e, he p esen esea ch aims o in es iga e how In e ne use a ec s he MH o olde adul s based on he 2013–2018 wa es
o he China Heal h and Re i emen Longi udinal S udy (CHARLS). Mo eo e , we examine whe he In e ne use has a media ing e ec
on MH by in luencing ou key aspec s: social in e ac ion, social us , a eling expenses, and heal h habi s. This s udy also conside s
he he e ogenei y o olde adul s in China and analyzes he impac o In e ne use on MH among di e en g oups.
Unlike p e ious s udies, his esea ch con ibu es o he exis ing li e a u e in he ollowing h ee aspec s. Fi s , using he na ionally
ep esen a i e da a o he 2013–2018 wa es o CHARLS, his s udy enhances he alidi y o he es ima ed causal ela ionship be ween
In e ne use and MH among olde adul s by add essing endogenei y conce ns using he ins umen al a iable (IV) me hod. Second, o
he bes o ou knowledge, his is he i s s udy o examine he media ing e ec o a eling expenses and conside ou key aspec s in
he same amewo k o u he he exis ing unde s anding o he unde lying channels be ween In e ne use and MH. Thi d, we analyze
he he e ogenei y o he esul s ac oss he numbe o ch onic diseases ha olde adul s ha e and hei li ing a angemen s (i.e., li ing
wi h o wi hou hei child en) o cap u e he e ec o In e ne use on hese speci ic g oups o olde adul s. This s udy o e s aluable
insigh s o u u e esea ch endea o s ha aim o explo e he ela ionship be ween In e ne use and MH among olde adul s in China.
The s udy is o ganized as ollows. Sec ion 2 p esen s he da a and a iables, while Sec ion 3 desc ibes he empi ical me hods.
Sec ion 4 p o ides he es ima ion esul s, and Sec ion 5 discusses he indings based on he empi ical esul s. Finally, Sec ion 6
concludes by summa izing he main indings and p oposing policy implica ions.
L. Zhang e al.
Heliyon 10 (2024) e25397
3
2. Da a
2.1. Sample
The da a used he ein a e aken om he 2013–2018 wa es o CHARLS, a na ionally ep esen a i e longi udinal su ey conduc ed
by he Ins i u e o Social Science Su ey a Peking Uni e si y. CHARLS ocuses on households and indi iduals o e 45 yea s o age in
China. The su ey employs a mul is age sampling me hod wi h p obabili y p opo ional o size. The na ional baseline su ey was i s
conduc ed in 2011, ollowed by subsequen wa es e e y 2–3 yea s. The da a om he 2013–2018 wa es a e he mos cu en da a
a ailable. Encompassing 150 coun ies and 450 illages o esiden commi ees in 28 p o inces, hese wa es o he su ey included
app oxima ely 17,000 indi iduals om app oxima ely 10,000 households. The ques ionnai e collec s a a ie y o in o ma ion,
including de ails on he esponden s’ socioeconomic s a us and heal h ci cums ances. The CHARLS da abase is widely ecognized as a
eliable and aluable sou ce o da a wi hin he academic communi y. Acco ding o he ‘Law o P o ec ion o Righ s and In e es s o
Olde Adul s’ in he People’s Republic o China, indi iduals o e he age o 60 yea s a e conside ed olde adul s. The e o e, o he
sample, indi iduals aged 60 yea s and abo e in he 2013–2018 wa es o CHARLS da a a e selec ed. A e sc eening and elimina ing
samples ha lacked ele an a iables, a o al o 27,561 alid samples we e ob ained. All da a analyses a e pe o med using STATA
16.1 so wa e (STATA Co p. LLC, College S a ion, TX, USA).
2.2. Va iables
2.2.1. Dependen a iable
In his s udy, he dependen a iable is MH, speci ically e e ed o as “dep ession” in he CHARLS da ase . To measu e dep ession,
CHARLS uses a simpli ied e sion o he Cen e o Epidemiologic S udies Dep ession Scale (CES-D10) [32]. This scale comp ises eigh
ques ions ega ding nega i e emo ions and wo ques ions add essing posi i e emo ions. The scale encompasses di e se aspec s such as
daily mood, loneliness, sleep su iciency, and he li e si ua ion o olde adul s. The esponse op ions, namely, “ a ely o none o he
ime,” “some o a li le o he ime,” “occasionally o a mode a e amoun o he ime,” and “mos o all o he ime” a e assigned alues
o 0, 1, 2, and 3, espec i ely. In con as , he sco es o he posi i e emo ion ques ions a e in e sely coded. The sum o he sco es o
he 10 i ems anged om 0 o 30, wi h highe alues indica ing highe dep ession le els. A sco e o 10 o highe is conside ed a high
dep essi e le el and assigned a alue o 1, whe eas a sco e below 10 indica es no dep ession o a low dep essi e le el and is assigned a
alue o 0 [33,34]. I should be no ed ha CES-D10 se es as a p elimina y dep ession sc eening ool and canno eplace clinical
diagnos ic conclusions. The eliabili y coe icien s o he CES-D10 i ems as in es iga ed by C onbach in 1999 and 2003 a e 0.84 and
0.86, espec i ely [35].
2.2.2. Independen a iable
The independen a iables examined in his s udy a e In e ne use and equency o In e ne use. Owing o da a limi a ions, we
examined In e ne use and equency o In e ne use o he 2013–2018 wa es o he CHARLS ques ionnai e using he ollowing
ques ion, “Ha e you done any o hese ac i i ies in he las mon h?” The ques ion p esen ed 12 esponse op ions. The esponden s who
selec ed he op ion “Used he In e ne ” a e classi ied as In e ne use s in his s udy and assigned a alue o 1. O he wise, he e-
sponden s a e assigned a alue o 0 [28,36,37]. The equency o In e ne use a iable, as obus ness, was examined by he ollowing
ques ion, “How o en did you use he In e ne in he las mon h?” The esponse op ions o his ques ion a e ca ego ized as ollows:
“almos daily” and “almos e e y week” a e assigned a alue o 1, whe eas “no egula ly” is assigned a alue o 0.
2.2.3. Channel a iables
To comp ehend he unde lying channels h ough which In e ne use ( equency o In e ne use) may impac dep ession among
olde adul s, a iables associa ed wi h social in e ac ion, social us , a eling expenses, and heal h habi s a e examined. In he
2013–2018 wa es o he CHARLS ques ionnai e, he a iable o social in e ac ion is assessed using he ques ion, “Ha e you done any o
hese ac i i ies in he las mon h?” The ques ion included a ious esponse op ions such as “In e ac ed wi h iends,” “Played Ma-jong,
played chess, played ca ds, o wen o a communi y club,” “P o ided help o amily, iends, o neighbo s who do no li e wi h you,”
“Wen o a spo , social, o o he kinds o he club,” “Took pa in a communi y- ela ed o ganiza ion,” and “A ended an educa ional o
aining cou se.” I esponden s selec any o hese op ions, hey a e assigned a alue o 1; o he wise, hey a e assigned a alue o 0.
The a iable o social us is assessed by asking he ollowing ques ion: “Ha e you done any o hese ac i i ies in he las mon h?” In
CHARLS, esponden s a e asked, “In he las yea , how much did you household spend on he ollowing i ems?” I he answe o his
ques ion is g ea e han 0, a alue o 1 is assigned; o he wise, a alue o 0 is assigned.
Fo he a iable o a el expenses, we ocused on enjoyable consump ion such as a eling expenses. The ques ion is “Ha e you
done any o hese ac i i ies in he las mon h?” The esponses included op ions such as “Done olun a y o cha i y wo k” and “Ca ed
o a sick o disabled adul who does no li e wi h you.” Responden s who selec ed ei he o hese op ions a e assigned a alue o 1,
whe eas o he s a e assigned a alue o 0.
The a iables o heal h habi s included smoking and physical ac i i y. The esponses o smoking a e de i ed om ques ions such as
“Ha e you e e chewed obacco, smoked a pipe, smoked sel - olled ciga e es, o smoked ciga e es/ciga s?” and “Do you s ill ha e he
habi , o ha e you qui ?” The s a us o physical ac i i y is de i ed by asking, “Do you usually engage in his ype o ac i i y o a leas
10 min e e y week?” To measu e he a iable o heal h habi s, “yes” esponses a e assigned a alue o 1, whe eas “no” esponses a e
assigned a alue o 0.
L. Zhang e al.
Heliyon 10 (2024) e25397
4
2.2.4. Co a ia es
Acco ding o G ossman’s heal h needs heo y [38], a ious ac o s—such as biological ac o s, income, social s a us, educa ion,
cul u e, ea ly child de elopmen , heal h sys ems, social suppo , popula ion p o iles, and geog aphy—play signi ican oles in in lu-
encing dep ession. In ela ed s udies on In e ne use ( equency o In e ne use) and dep ession among olde adul s, con ol a iables
a e selec ed om h ee dimensions: indi idual, amily, and social cha ac e is ics. In his s udy, he con ol a iables included gende ,
age, educa ion, ma i al s a us, he loga i hm o income, he numbe o ch onic diseases, u ban esidence, insu ance, and p o ince and
ime con ols [25,32,36,37]. In addi ion, gi en he inc easing numbe o “emp y nes e s,” we ook he in luence o child en on olde
adul s in o conside a ion, and a iables such as cohabi a ion wi h child en and he numbe o child en a e also included. These
a iables a e selec ed because o hei close associa ions wi h he occu ence o dep ession symp oms. The da a used he ein can be
ob ained om he CHARLS 2013–2018 da ase , al hough u he de ails a e no p o ided.
3. Empi ical me hod
3.1. E ec o in e ne use ( equency o in e ne use) and MH in olde adul s
To in es iga e how In e ne use ( equency o In e ne use) could a ec olde adul s’ dep ession, he panel andom e ec s (RE)
P obi model is employed o examine he causali y be ween MH and In e ne use ( equency o In e ne use) among olde adul s. This
model is es ablished on he basis o he s anda d no mal dis ibu ion equa ion.
P(Hi =1) = Φ(β0+β1Ii +β2Zi )(1)
In eq (1), H
i
is he MH o he i h olde adul s in yea , I
i
ep esen s In e ne use ( equency o In e ne use), and Z
i
indica es se e al
con ol a iables o accoun o po en ial con ounding ac o s. Eq (1) can be gi en by:
Hi ∗=
α
0+
α
1Ii +
α
2Zi +δ +ξi ,Hi=1(Hi
∗>0)(2)
In eq (2), Hi∗is a la en a iable, δ is a dummy a iable ha con ols he ime end and p o ince, ξi symbolizes andom dis u bance
e ms, and we can ob ain he unbiased es ima o Co (Ii ,ξi ) = 0. To ensu e unbiased and consis en es ima ion esul s while add essing
po en ial endogenei y issues, his s udy used he IV-RE P obi model. The IV-RE P obi model, which co esponds o bina y ac o
a ia ion, is selec ed as he benchma k o he bi a ia e panel P obi model. Unde he assump ion o con e gence, employing he IV-
RE P obi model o es ima ion p o ed o be mo e e icien . The undamen al concep is o iden i y ins umen al a iables ha a e
associa ed wi h In e ne use ( equency o In e ne use) bu no wi h MH and cons uc an ins umen al a iable equa ion o eplace he
endogenous a iables in he equa ion. The speci ic implemen a ion p ocess is ou lined as ollows:
Hi ∗=β0+β1Ii +β2Zi +δ +ξi Hi =1(Hi
∗>0)(3)
Ii =γ0+γ1IVi +γ2Zi +ξ(4)
In eq (3) – eq (4), IVi indica es he ins umen al a iables and Co (IVi ,ξi ) = 0, Co (ξ,ξi ) = 0, and Co (IVi ,Ii ) ∕= 0.
3.2. Possible channels explaining he heal h e ec s o in e ne use and equency o in e ne use
The o he aim o his s udy is o shed ligh on he po en ial channels h ough which In e ne use ( equency o In e ne use) a ec s
dep ession. Some schola s ha e p oposed ha he use o s epwise es s o media ing e ec s should be discon inued and ha e-
sea che s should e ain om es ima ing he size o indi ec e ec s o es ing hei s a is ical signi icance [39]. Ins ead, acco ding o
such schola s, he ocus o esea ch should shi owa d enhancing he c edibili y o he iden i ied causal ela ionship be ween
explana o y a iable I and explained a iable H. As pe economic heo y, one o se e al in e media y a iables, M, a e in oduced o
e lec he causal pa hway om I o H. The impac o M on H should be di ec and e iden , wi h an emphasis on iden i ying he
co ela ion be ween I and M. In his s udy, ou p ima y channels, ha is, social in e ac ion, social us , a eling expenses, and heal h
habi s, a e examined. The model is gi en as ollows:
Mi =β0+β1Ii +β2IVi +β3Zi +δ +
τ
i (5)
In eq (5), Zi ep esen s he con ol a iable, and Mi deno es he media o a iable.
3.3. Iden i ica ion s a egy
Gi en he longi udinal s uc u e o ou da a, we can simply un a RE es ima ion and a ixed e ec s es ima ion and use he Hausman
es o check which model is p e e ed. Howe e , we ha e unbalanced panel da a, which sugges s ha pa icipan s a e no consis en ly
in e iewed o did no consis en ly espond o he CHARLS ques ions we a e in e es ed in. The ixed e ec s model assumes ha some
ime-in a ian indi idual cha ac e is ics co ela e wi h he p edic o o independen a iables and may be biased in he es ima es. The
RE model assumes ha a ia ion be ween indi iduals is andom and unco ela ed wi h he p edic o o independen a iables included
in he model, unlike he ixed e ec s model. In his s udy, he explana o y a iable “dep ession” is a dicho omous a iable, so i is
L. Zhang e al.
Heliyon 10 (2024) e25397
5
es ima ed using a panel p obi . Howe e , i will lead o an inconsis en es ima ion o he coe icien
α
_1 i he ixed e ec o he panel
p obi is es ima ed. A he same ime, he LR es o he RE model in Table 2 ejec s he hypo hesis o he mixed p obi eg ession model.
The e o e, he panel RE p obi model is inally selec ed o es ima ion. Howe e , he po en ial endogenei y p oblem s ill a ec s ou
p ima y esul s. Namely, he decision o use he In e ne , along wi h an olde indi idual’s dep essi e symp oms may be de e mined o
in luenced by some missing o unobse ed a iable. In addi ion, MH condi ions a ec an olde indi idual’s beha io , which in u n
may a ec hei online beha io , c ea ing a e e se causali y p oblem. To o e come hese p oblems and es he main esul s, we
employ he IV-RE p obi me hod o e-es ima e he model. This s udy uses he a e age In e ne use le els o olde adul s in each illage
as he ins umen al a iable o whe he he subjec s used he In e ne o no . The IV-RE p obi me hod is a wo-s age eg ession. In he
i s s age, he ins umen al a iables o he p incipal independen a iables a e de e mined and hei co ela ions a e e alua ed. In he
second s age, he ins umen al a iables a e subs i u ed in o he model o eg ession, and he ex e nali y o In e ne use ( equency o
In e ne use) is e i ied acco ding o he endogenei y es pa ame e s.
4. Resul s
4.1. S a is ical desc ip ion
Table 1 displays he desc ip i e s a is ics o he a iables, including he numbe o obse a ions, means, and s anda d de ia ions.
36 % o he esponden s epo ed expe iencing dep ession. The independen a iable, “In e ne ”, is u ilized by 26 % o he olde
popula ion in he las mon h, indica ing a ela i ely low p e alence o In e ne use among olde adul s in China. The a e age equency
o In e ne use o he sample is 1.1, wi h a s anda d de ia ion o 0.36. The pa icipan s’ educa ional le els ange om 0 o 22 yea s,
wi h an a e age o 6.6 yea s o educa ion. The majo i y o he pa icipan s (76 %) a e ma ied. Fu he de ails o he desc ip i e
s a is ics a e p esen ed in Table 1. Table A1 p o ides a isual compa ison o he di e ences be ween In e ne use s and non-use s. The
esul s indica e a s a is ically signi ican di e ence in he le el o dep ession be ween In e ne use s (mean =0.39) and non-use s
(mean =0.28). In addi ion, s a is ically signi ican di e ences a e no ed in 8 co a ia es be ween he a o emen ioned wo g oups. I
sugges s ha a causal link be ween In e ne use and dep ession le els in olde adul s canno be ex apola ed.
4.2. Baseline eg ession
Using he panel RE P obi model, Table 2 p esen s he impac o In e ne use and equency o In e ne use on MH. In Model (1),
In e ne use demons a es a signi ican nega i e coe icien o −0.401 (p <0.01), indica ing ha i educes dep ession le els. Model (2)
yields consis en esul s e en a e inco po a ing co a ia es, hus highligh ing he signi ican ly nega i e ela ionship be ween he
selec ed con ol a iables and dep ession wi h a coe icien o −0.300. To minimize he bias o analysis esul s, we also conside he
equency o In e ne use in he baseline eg ession analysis. Model (3) in Table 2 p esen s he es ima ed esul s o he equency o
In e ne use a iable. The esul s show ha he coe icien is es ima ed as −0.432 hus indica ing a s a is ically signi ican impac a he
1 % le el. Upon inco po a ing co a ia es, Model (4) displays consis en ly signi ican ly nega i e esul s wi h a coe icien o −0.261. I
Table 1
Desc ip ions and basic s a is ics o he selec ed a iables.
Va iables De ini ion Mean S.D.
Dependen a iable
MH I high dep essi e le el =1, o he wise =0 0.36 0.48
Independen a iable
In e ne I indi idual uses In e ne =1, o he wise =0 0.26 0.44
F e_in e ne The equency o In e ne use, I no egula ly =1, I almos e e y week =2, I almos daily =3 1.1 0.36
Channel a iables
Social Joining social ac i i ies 0.46 0.50
T us T us o he s including s ange s 0.03 0.16
T a eling expenses Long-dis ance a eling expenses 0.37 0.48
Smoke Smoking 0.34 0.47
Physical ac i i ies Doing physical ac i i ies e e y week 0.57 0.49
Con ol a iables
Gende I male =1, o he wise =0 0.50 0.50
Age Yea s o age 70 7.40
Age
2
Age squa ed 4885 1079
Educa ion Yea s o educa ion 6.60 5.00
Ma ied I ma ied =1, o he wise =0 0.76 0.42
Lincome The Loga i hm o esponden ’s income in he las yea (yuan) 5.50 3.10
Illness Numbe s o ch onic diseases 1.30 1.40
U ban I cu en ly li ing in u ban =1, o he wise =0 0.39 0.49
Li ing wi h child I li ing wi h child =1, o he wise =0 0.40 0.49
Child numbe Numbe s o child en 1.90 2.00
Insu ance I had basic endowmen insu ance =1, o he wise =0 0.32 0.46
Obse a ions: 27,561
Sou ce: Au ho s’ calcula ion based on he CHARLS da a om he 2013–2018 wa es.
L. Zhang e al.
Heliyon 10 (2024) e25397
6
sugges s ha in oducing con ol a iables can e ec i ely educe es ima ion bias. This inding suppo s he no ion ha In e ne use can
e ec i ely dec ease dep ession le els and enhance MH.
4.3. Accoun ing o he endogenei y o in e ne use and equency o in e ne use
Fo he ins umen al a iable, his s udy selec s he a e age In e ne use le els o olde adul s in each illage o communi y,
excluding he indi iduals being analyzed. Using high-le el a e age alues (e.g., p o ince- o communi y-le el ones) as ins umen s o
lowe le els (e.g., he household le el) is a common echnique in he li e a u e [40–42]. The eliabili y o he IV es ima ion depends on
he alidi y o he ins umen al a iable. The ins umen al a iable should be highly co ela ed wi h he endogenous a iable bu has
Table 2
Baseline eg ession esul s o he e ec o In e ne use and equency o In e ne use on MH.
Va iables (1) (2) (3) (4)
In e ne −0.401*** −0.300***
(0.027) (0.028)
F e_in e ne −0.432*** −0.261***
(0.036) (0.036)
Gende −0.413*** −0.408***
(0.026) (0.026)
Age 0.155*** 0.160***
(0.026) (0.026)
Age
2
−0.001*** −0.001***
(0.000) (0.000)
Educa ion −0.012*** −0.012***
(0.003) (0.003)
Ma ied −0.166*** −0.171***
(0.031) (0.031)
Lincome −0.018*** −0.018***
(0.005) (0.005)
Illness 0.131*** 0.132***
(0.008) (0.008)
U ban −0.334*** −0.363***
(0.028) (0.028)
Li ing wi h child 0.066*** 0.013
(0.023) (0.023)
Child numbe −0.026*** −0.022**
(0.010) (0.010)
Insu ance −0.254*** −0.252***
(0.034) (0.034)
P o ince and ime ixed e ec Yes Yes Yes Yes
lnsig2u 0.032 −0.221*** 0.056 −0.203***
(0.045) (0.049) (0.045) (0.049)
LR es 1741.13*** 1221.59*** 1800.37*** 1252.35***
N 27,561 27,561 27,561 27,561
No e: S anda d e o s a e in pa en heses. *p <0.1, **p <0.05, ***p <0.01.
Sou ce: Au ho s’ calcula ion based on he CHARLS da a om he 2013–2018 wa es.
Table 3
IV es ima ion o he impac o In e ne use and equency o In e ne use on MH.
Va iables (1) (2)
Second s age:
In e ne −0.581***
(0.077)
F e_in e ne −1.351***
(0.117)
Con ols
a
Yes Yes
P o ince and ime ixed e ec Yes Yes
Fi s s age:
IV −0.739*** −0.325***
(0.016) (0.014)
Wald es 19.52*** 43.95***
AR 57.65*** 60.08***
Wald 57.35*** 55.50***
N 27,561 27,561
No e: S anda d e o s a e in pa en heses. *p <0.1, **p <0.05, ***p <0.01.
a
Con ols include all con ol a iables as in Table 2.
Sou ce: Au ho s’ calcula ion based on he CHARLS da a om 2013 o 2018 wa es.
L. Zhang e al.
Heliyon 10 (2024) e25397
7
no di ec co ela ion wi h he dependen a iable. Fi s , i sa is ies he co ela ion equi emen . Following beha io al imi a ion heo y,
a highe In e ne use a e among olde adul s in a gi en illage inc eases he likelihood and he equency o indi idual In e ne use due
o spillo e e ec s. Second, i is deemed exogenous. In e ne use by o he s in he same illage is unlikely o ha e a di ec impac on
indi idual beha io and dep ession. Hence, he a e age In e ne use a e is conside ed an ideal ins umen al a iable. The CHARLS
samples a e ca ego ized acco ding o illage/ esidence code, and he a e age In e ne use a e o olde adul s in each illage o
neighbo hood commi ee is calcula ed as he egional-le el In e ne use a e. The s eng h o he ins umen al a iable is assessed using
he p- alue. I i is signi ican a he 1 %, 5 %, o 10 % le els, he absence o a weak ins umen al a iable p oblem in he eg ession is
indica ed. This app oach ensu es ha he ins umen al a iable selec ed is s ongly co ela ed wi h he dependen a iables and does
no in oduce bias in o he esul es ima o s.
O he a ious me hods used o es ablish causali y, IV analysis o e s dis inc ad an ages in e ms o add essing endogenei y. To
add ess endogenous issues, his s udy employs an IV-RE app oach o a comp ehensi e analysis. In he i s s age o he eg ession, as
shown in Table 3, he ins umen al a iable is s a is ically signi ican and nega i ely co ela ed wi h In e ne use and he equency o
In e ne use hus mee ing he dependency condi ion o he ins umen al a iables. The p- alue o he weak IV es is signi ican a he 1
% le el. Rejec ing he null hypo hesis (null hypo hesis H
0
: endogenous a iables a e no co ela ed wi h ins umen al a iables),
indica es ha he ins umen al a iables a e no weak. Thus, he ins umen al a iable is alid and can be used in u he es ima ions.
Following he indings in Model (1)–(2) o Table 3, In e ne use and he equency o In e ne use exhibi s a is ical signi icance (p
<0.01) o dep ession among olde adul s. These indings indica e ha a highe le el o In e ne use o a highe equency o In e ne
use has a nega i e impac on dep ession among olde adul s, which is consis en wi h he esul s ob ained om he baseline eg ession
analysis. I is simila o he esul s o p e ious s udies [36,43].
4.4. Robus ness es
To ensu e he obus ness o he indings, h ee addi ional me hods a e used o conduc obus ness es s. To explo e he obus ness o
he dep ession a iable, he measu emen o he independen a iable is changed o a con inuous a iable Table 4 p esen s he
es ima ed esul s a e subs i u ing he measu emen o he dependen a iable o obus ness es ing. Acco ding o he esul s, he
e ec is s a is ically signi ican a he 1 % le el. This inding u he suppo s he no ion ha In e ne use ( equency o In e ne use)
enhances he MH o olde adul s, which is consis en wi h he esul s ob ained om he baseline eg ession analysis.
Acco ding o he age c i e ia de ined by he Wo ld Heal h O ganiza ion o he olde popula ion, indi iduals aged 85 yea s and
abo e a e conside ed o be long-li ed olde adul s. Thus, o accoun o he po en ial impac o ex eme age alues, indi iduals
exceeding 85 yea s o age a e excluded om his analysis, and he obus ness o he esul s is e-es ima ed. Table 5 shows ha he
es ima ed ma ginal e ec s a e −0.570 and −0.811, demons a ing s a is ical signi icance a he 1 % le el. These esul s a e consis en
wi h p e ious esea ch indings; hus ein o cing he obus ness o he obse ed ela ionship be ween In e ne use ( equency o
In e ne use) and MH ou comes among he olde popula ion.
To add ess po en ial sel -selec ion bias and e e se causali y, he endogenous swi ching p obi (ESP) me hod is employed o in-
dep h analysis. This app oach accoun s o he possibili y ha In e ne use among olde adul s may esul om sel -selec ion and
ha he e could be a bidi ec ional ela ionship be ween In e ne use and MH. The ESP model has been designed o mi iga e bias due o
obse able and unobse able ac o s [44,45]. To con ol o sel -selec ion bias, he IV me hod is employed o co ec o any po en ial
bias [46].
The es ima ion esul s o he ESP model a e p esen ed in Table A2. The a e age In e ne use o o he olde adul s wi hin he same
illage had a signi ican posi i e e ec on an indi idual’s likelihood o using he In e ne . This sugges s ha a highe a e age le el o
In e ne use among olde adul s in he same illage inc eases he p obabili y o indi idual In e ne adop ion. The pa ame e
ρ
1
is
s a is ically signi ican a he 1 % le el, indica ing he p esence o selec i e bias in he eg ession model. Unobse able ac o s impac
bo h he le el o In e ne access and he MH o olde adul s. Fu he mo e,
ρ
1
de e mined o be posi i e, sugges ing ha he MH le el o
Table 4
IV es ima ion o he impac o In e ne use and equency o In e ne use on MH.
Va iables (1) (2)
Second s age:
In e ne −0.248**
(0.232)
F e_in e ne −0.522***
(0.109)
Con ols
a
Yes Yes
P o ince and ime ixed e ec Yes Yes
Fi s s age:
IV −0.353*** −0.353***
(0.247) (0.446)
F alue 20.79 62.87
N 27,561 27,561
No e: S anda d e o s a e in pa en heses. *p <0.1, **p <0.05, ***p <0.01.
a
Con ols include all con ol a iables as in Table 2.
Sou ce: Au ho s’ calcula ion based on he CHARLS da a om 2013 o 2018 wa es.
L. Zhang e al.
Heliyon 10 (2024) e25397
8
olde adul s using he In e ne is highe han ha o he o e all sample. The goodness-o - i es and model independence a e s a-
is ically signi ican a he 1 % le el, indica ing ha he use o he ESP model is app op ia e.
The ESP model is selec ed o empi ically analyze he impac wi hin a “coun e ac ual” amewo k, allowing o he calcula ion o
he a e age ea men e ec on he ea ed (ATT), a e age ea men e ec on he un ea ed (ATU), and a e age ea men e ec (ATE)
[47]. Howe e , because he ATU and ATE pa ame e s include he e ec s on samples una ec ed by joining coope a i es, his s udy
ocuses solely on he ATT, ep esen ing he mean change in he MH s a us o he sample using he In e ne compa ed wi h ha i hey
did no use he In e ne . Table 6 p esen s he es ima ed ea men e ec s o In e ne use on MH beha io s among olde adul s. The
ea men e ec is consis en wi h he p e iously es ima ed esul s, indica ing ha In e ne use can signi ican ly imp o e MH among
olde adul s. The esul s ob ained om he ESPmodel p o ide addi ional suppo o he obus ness o he indings.
4.5. He e ogenei y analysis
In he p e ious s udy, esea che s ea ed all olde adul s as a homogeneous g oup when examining he a e age e ec o In e ne
use ( equency o In e ne use) on MH. Howe e , he In e ne use ( equency o In e ne use) o di e en olde g oups wi h a ying
indi idual cha ac e is ics is he e ogeneous is no able. P e ious s udies ha e demons a ed he impac o In e ne use ( equency o
In e ne use) on dep ession ac oss a ious demog aphic g oups [44,48,49], which di e in e ms o income le el, gende , educa ion
le el, age, and numbe o accompanying pe sons [50–52]. Howe e , limi ed e idence exis s ega ding he di e ences obse ed among
g oups based on hei physical heal h s a us o li ing a angemen s. To in es iga e he e ec s o In e ne use ( equency o In e ne use)
on MH among olde indi iduals wi h di e en cha ac e is ics, we examine he eg ession esul s o a iables such as he numbe o
ch onic diseases and li ing a angemen s (speci ically, li ing wi h child en). Mo eo e , he analysis employs he IV-RE me hod o
add ess any po en ial endogenei y issues.
Table 7 p esen s he indings o he he e ogenei y analysis be ween In e ne use and dep ession. In Model (1), he e ec o he
in e ac ion be ween In e ne use and li ing a angemen s (whe he he esponden s a e li ing wi h hei child en) on dep ession is
examined. The indings demons a e a signi ican nega i e in e ac ion e ec , sugges ing ha he ela ionship be ween In e ne use and
dep ession is a enua ed among olde adul s who cohabi wi h hei child en. In o he wo ds, In e ne use is mo e likely o be asso-
cia ed wi h be e MH among indi iduals li ing wi h hei child en. In Model (2), he e ec o he in e ac ion be ween In e ne use and
he numbe o ch onic diseases on dep ession is in es iga ed. The esul s e eal a signi ican posi i e in e ac ion, sugges ing ha he
associa ion be ween In e ne use and dep ession is s onge among olde adul s wi h a lowe numbe o ch onic diseases, which in-
dica es ha heal hie olde indi iduals end o expe ience g ea e bene i s om In e ne use han hose wi h poo e heal h condi ions.
Model (3) inco po a es bo h o hese in e ac ion e ms, and he esul s align wi h hose o Models (1) and (2). Thus, i can be concluded
ha he In e ne has he po en ial o educe MH dispa i ies among he olde adul popula ion. No ably, his e ec ends o be
signi ican ly posi i e o olde adul s wi h ewe ch onic diseases who li e wi h hei child en. Table A3 p esen s he indings o he
Table 5
Robus ness es esul s o MH consequences o In e ne use and equency o In e ne use.
Va iables (1) (2)
Second s age:
In e ne −0.570***
(0.083)
F e_in e ne −1.263***
(0.186)
Con ols
a
Yes Yes
P o ince and ime ixed e ec Yes Yes
Fi s s age:
IV −0.773*** −0.555***
(0.018) (0.025)
Wald es 18.86*** 37.97***
AR 47.80*** 49.99***
Wald 47.43*** 46.09***
N 21,833 21,833
No e: S anda d e o s a e in pa en heses. *p <0.1, **p <0.05, ***p <0.01.
a
Con ols include all con ol a iables as in Table 2.
Sou ce: Au ho s’ calcula ion based on he CHARLS da a om 2013 o 2018 wa es.
Table 6
The ea men e ec o In e ne use on olde adul s’ MH.
Va iables Obse a ions Mean S.D. Min Max
ATT 7053 −0.192 0.039 −0.334 −0.050
ATU 20,508 −0.167 0.042 −0.296 0.001
ATE 27,561 −0.173 0.039 −0.298 0.003
Sou ce: Au ho s’ calcula ion based on he CHARLS da a om he 2013–2018 wa es.
L. Zhang e al.
Heliyon 10 (2024) e25397
15
[54] T.C. Lang, The e ec o he In e ne on a el consume pu chasing beha iou and implica ions o a el agencies, J. Vaca . Ma k. 6 (4) (2000) 368–385,
h ps://doi.o g/10.1177/135676670000600407.
[55] R. Law, A. Law, E. Wai, The impac o he in e ne on a el agencies in Hong Kong, J. T a . Tou ism Ma ke . 11 (2–3) (2002) 105–126, h ps://doi.o g/
10.1108/09596110410519982.
[56] M. K oesen, J. De Vos, Does ac i e a el make people heal hie , o a e heal hy people mo e inclined o a el ac i ely? J. T anspo Heal h 1610 (2020) h ps://
doi.o g/10.1016/j.j h.2020.100844.
[57] A. Zamanzadeh, T. Ca oli, R. Bane jee, T a elling, anxie y and he impac o COVID-19: e idence om I aly, Cu . Issues Tou ism 8 (2022), h ps://doi.o g/
10.1080/13683500.2022.2153651.
[58] C.E. Moge, D.M. Romano, Con ex ualising ideo game engagemen and addic ion in men al heal h: he media ing oles o coping and social suppo , Heliyon 6
(11) (2020) 10, h ps://doi.o g/10.1016/j.heliyon.2020.e05340.
[59] R.F. McCloud, C.A. Okechukwu, G. So ensen, K. Viswana h, En e ainmen o heal h? Explo ing he in e ne usage pa e ns o he u ban poo : a seconda y
analysis o a andomized con olled ial, J. Med. In e ne Res. 18 (3) (2016), h ps://doi.o g/10.2196/jmi .4375.
[60] K. Mikkelsen, L. S ojano ska, M. Polenako ic, M. Bose ski, V. Apos olopoulos, Exe cise and men al heal h, Ma u i as (2017) 10648–10656, h ps://doi.o g/
10.1016/j.ma u i as.2017.09.003.
L. Zhang e al.
