The Christi-Muslim Gp i the Sub-Shr Fertility Trsitio Drft: Plese do ot quote or cite Victor Agdji Deprtmet of Sociology Arizo Stte Uiversity Agdji@su.edu To be preseted t the 2004 Aul Meetigs of the Popultio Associtio of Americ 1-3 April 2004 Bosto, MA Mrch 2004
Itroductio The study sets out to exmie Christi-Muslim differetils i cotrceptive use d fertility i sub-shr coutries with dequte DHS dt d discusses the implictios of these differetils for the sub-shr fertility trsitio. Although religio is ofte metioed i referece to fertility d fertility regultio, o systemtic d cosistet evidece hs yet emerged. The diversity of religious trditios i differet societies gretly complictes the study of this reltioship. The quest for the roots of this reltioship typiclly strts with scrutiy of the religious doctrie. This quest, however, is usully brief d usuccessful: most religious techigs, formulted well before fertility regultio becme coceivble idividul optio d pressig public cocer, re mbiguous o reproductive mtters llowig for wide rge of iterprettios. The frustrtio with the theologicl perspective quickly moves reserchers i the opposite directio tht of the so-clled chrcteristics hypothesis. This hypothesis posits tht the religio-fertility reltioship is etirely due to other fctors ssocited with religio but re ot rooted i religious doctrie per se. This view is prticulrly commo i demogrphy whose methodologicl pprtus offers seemigly strightforwrd d covicig wys to idetify d cotrol for such fctors. The chrcteristics ssumptio is the most ifluetil i demogrphic studies d is implicitly preset eve i lyses tht do ot mge to effce the sttisticl effect of religio through multivrite exercise: becuse y residul effect remis substtively uexplied, it is implied tht it owes to some other fctors tht could ot be idetified from the dt. The miority-sttus perspective, which is ofte proposed s other heuristic ltertive, c lso be see s offshoot of the chrcteristics hypothesis. This perspective looks for swers i the socio-politicl sttus of religious (or ethic) groups d the ifluece of this sttus o the groups fertility strtegies d choices. Testig this hypothesis, especilly with demogrphic dt, presets mjor chllege s religious group s socio-politicl positio is ofte difficult to defie d disetgle from the socioecoomic d ethoculturl fctors tht re usully cosidered uder the chrcteristics perspective. 2
Reserch o the religio-fertility reltioship i sub-shr Afric hs illustrted this coceptul ucertity. Most of these studies hve looked t idividul coutries d sigle cross-sectiol dtsets, without likig fertility levels to cotrceptive use. Much of this reserch hs focused o Muslims. Severl studies hve oted tht Muslims fertility i sub- Shr Afric teds to be lower th tht of o-muslims cotrry to the ptter typiclly observed i other regios; Other studies, however, furish evidece to the cotrry (Biley 186; Gisie 172; Krim 17; Kollehlo 14; Sembjwe 180). Yet, o covicig rgumets hve bee proposed to expli d recocile these coflictig fidigs. Whtever Muslim-Christi fertility comprisos hve bee ttempted i sub-shr Afric hve bee usully cst withi the chrcteristics prdigm mily becuse the other perspectives ppered ill-suited to expli the observed ptters. Thus, Islmic techigs do ot offer y uequivocl positio o fertility d fertility regultio; Christi deomitios, despite the Ctholic Church s geerl dispprovl of cotrceptio, re i prctice tolert d eve supportive of fmily plig. At the sme time, pprochig the evidece from the miority-sttus perspective hs fced isurmoutble methodologicl difficulties. Theoreticl premises d hypotheses If the spred of fmily plig is to be viewed s prt of Afric s itegrtio ito the globl socio-culturl system defied o the bsis of wester vlues d techologies, Christis re to be expected to disply greter ffiity towrd fertility cotrol d cotrceptive use th re Muslims. Yet, Muslims lesser ccess or greter resistce to these wester iovtios is ot give. It is predicted o the politicl d historicl cotext i which both Muslims d Christis exist: where historiclly Muslims hve bee mrgilized prcticlly or symboliclly o the politicl re, their ccess to fertility limittio my be more costried d/or resistce to it my be stiffer; o the cotrry, i settigs where o historicl brriers to Muslims itegrtio ito the coloil d lter idepedet body politic hve existed, their disdvtge should be less visible. These fctors my ffect the Muslim-Christi differetils directly, i.e., through ttitudes d ccess to cotrceptives, d idirectly, by ifluecig ge t mrrige, post-prtum bstiece, wome s ccess to eductio, d host of other coditios pertiig to wome s sttus. This coceptul pproch hs importt implictios for theorizig the role of religio i fertility. Religio is o 3
loger competig expltio for fertility behvior d chge; isted it is prt of the historico-socil cotext tht costrits idividul behvior. Accordigly, the three competig theoreticl perspectives o religio d fertility become mutully complemetry d eve erichig: religious techig ffect other chrcteristics, while these chrcteristics ifluece the wy religious techigs re perceived d implemeted. Likewise religious miority s positio i society is closely relted to its other chrcteristics d the type d stregth of religious doctrie d idividuls dherece to this doctrie. This holistic d cotext-grouded pproch guides this study, lthough ievitbly i such brod overview costried by dt limittios my importt coutry-specific detils of fertilityreligio ssocitios cot be ddressed. Dt d methods Dt This study uses publicly vilble wome s files from Sub- Shr Demogrphic d Helth Surveys. Demogrphic dt impose mjor restrictios o studies of the religio-fertility reltioship, d this study is o exceptio. As most such studies, this compriso relies o religious ffilitios s the oly mrker of religio, levig outside the complex gmut of religious commitmet d prticiptio. Yet, lthough the religio-relted iformtio i DHS is limited to oe questio o religious self-idetifictio, the richess of reproductive d cotrceptive detils d the comprbility of the DHS dt cross coutries d time offer uique opportuity to explore religious differeces i cotrceptive choices d fertility outcomes i differet sub-shr settigs. This study is limited to DHS dtsets i which the two religious groups of iterest Christi d Muslims re clerly idetified. The lysis is restricted to ever-mrried wome d looks t stdrd cotrceptive d fertility idictors use of y cotrceptive method t y time; use of moder method t y time; curret use of y method; curret use of moder method; umber of childre ever bor, d religio-specific totl fertility rtes. This study is lso limited to coutries where both groups of iterest Christis d Muslims re clerly idetified i the dt. This excludes most coutries where Muslims re tiy miority (Souther Afric) or, o the cotrry, Muslims lmost completely domite (the Shel). I most remiig coutries 4
where t lest oe DHS ws coducted the Muslim segmet of the popultio is smller th the Christi oe. Filly, by defiitio the study excludes people who do ot declre ffilitio with either world religio, which i some coutries pper to be lrge portio of the popultio. Most comprisos i this study re limited to ever mrried wome. Costructed o the bsis of the DHS stdrd questios, ever-mrried wome re mple ctegory: it icludes ll wome who report hvig lived permetly with m regrdless of the type of mritl uio. We exclude ever mrried wome becuse their childberig d especilly cotrceptive behvior re likely to be misreported d re ofte drive by differet set of fctors th those of wome who hve bee i mritl uio. The exclusio of ever-mrried wome should lso help to obti more coservtive estimtes of religious differetils becuse for resos rooted i religio-sctioed geder ideology d other socil orms Christi wome would hve much higher level of premritl fertility th would Muslim wome. However, becuse childberig mog ever-mrried wome i sub-shr Afric is still tiy frctio of ll childberig, this restrictio would ot bis the overll picture. It must be ckowledged from the outset tht Christis re diverse group, d deomitiol differeces my ply big role i fertility d cotrceptive ptters s literture o the US d other settigs suggests (e.g., Addi 1; Agdji 2001; 2003; Avog 2001; Berhu 14; Goldscheider d Mosher 11; Gregso et l. 1; Mosher et l. 12). Yet, the DHS do ot lwys llow oe to distiguish mog differet types of Christi deomitios, especilly differeces withi Protestts. Islm i sub-shr Afric is, of course, ot uiform either, d tretig Muslims s oe group is dmittedly problemtic. However, the iterl heterogeeity tht cot be ccouted for i this study should ot deril the study s mi objective: to pit d explore the big picture of Christi- Muslim differetils. Methods We strt by presetig d discussig the reltive size d sptil distributio of Muslims d Christis d their geerl socio-demogrphic chrcteristics i the selected coutries tht my be relevt to fertility such the ge t first mritl uio d eductiol level. We the cosider fertility d 5
cotrceptive differetils strtig with comprisos of totl fertility rtes, umber of childre ever bor, d cotrceptive prevlece rtes. We the move o to more refied, multivrite comprisos of prity d cotrceptive use. Filly, we look t religious differetils i fertility prefereces focusig o the itetio to hve o more childre. Logistic regressio for biomil outcomes is employed for the lyses of cotrceptive use d fertility prefereces d Poisso regressio is employed for the lysis of the umber of childre ever bor. The lyses of cotrceptive use cotrol for such sociodemogrphic chrcteristics s ge, prity (umber of livig childre, i curret-use models d fertility prefereces), eductio, re of residece, work, mteril coditios, d curret mritl sttus (ot mrried, mrried moogmously, or mrried polygyously). The lysis of prity lso cotrols for ge t first mritl uio. Results Christi-Muslim demogrphics Tble 1 presets selected sociodemogrphic chrcteristics of Muslims d Christis i the coutries uder cosidertio. I ll the coutries, sve Burki Fso, Christis outumber Muslims. The rtio of Muslims to Christis rges from s low s.04 i 18 Key DHS to s high s 2 i 18 Burki Fso DHS. For coutries with more th oe DHS we get ide bout treds i the reltive size of the two religious groups. I West Afric, the umber of Muslims reltive to tht of the Christi popultio icresed i Cote d Ivoire d Burki Fso, decresed i Togo, Bei, Nigeri, d Cmeroo, d remied firly stble i Gh. Notbly, most vritio i the reltive sizes of the two groups ws due to cosiderble chges i the shre of Christis; i compriso, chges i the shre of Muslims were, i most cses, rther smll. Est Afric coutries with multiple DHS disply greter stbility i the reltive size of both groups; i ll of these coutries, however, the Muslim-Christi rtio registered slight icrese. Tble 1 bout here It is, of course, impossible to seprte the demogrphic growth compoet of these chges from coversios sice o dt o coversio re vilble. The use of DHS dt for popultio estimtes is i itself problemtic, d dt with cosiderble chges i the religious compositio (mily i the size of Christis) re prticulrly suspect. Yet oe hs to keep i mid tht for most, if ot ll of these coutries, DHS 6
sttistics o religious ffilitio re the best or t lest, the most imprtil dt vilble. Muslims d Christis do ot just differ i size; they re lso sptilly seprted, reflectig both the historicl ptters of both religios peetrtio ito Afric d correltio betwee ethicity d religio. Oe simple wy to ssess these sptil ptters is by costructig the idex of dissimilrity (ID) mesure of segregtio defied o the bsis of regiol distributio of Christis d Muslims recorded i the DHS dt files. Becuse the regiol clssifictio used i differet coutries is fr from uiform, the IDs tht we re ble to clculte from the DHS re ot fully ccurte d comprble but they oetheless provide good sese of the degree of coutrywide religious segregtio. The ID vlues re preseted i Tble 1: the higher the vlue o the scle betwee 0 d 1, the greter the degree of coutrywide segregtio. The vlues of ID vry from (suspiciously) very low.08 i Liberi to roud.70 i the lte 180s-erly 10s i Nigeri, Cmeroo, d Key. I most coutries with multiple DHS the IDs ted to decrese over time or to remi uchged; o cler regiol ptters emerge. The fertility literture typiclly poits to wome s eductio s mjor determit of fertility chge. The eductiol differeces betwee Muslims d Christis i sub-shr Afric re impressive (Tble 1). I ll coutries, with the exceptio of Ugd, Muslims re t disdvtge. The gp i the me umber of school yers completed is prticulrly lrge i Nigeri d Cmeroo, lthough the religious gp teded to rrow somewht betwee the erlier d lter DHS, s it did i ll other coutries. Oce gi, however, the biggest cotrst is betwee West d Cetrl Afric, o the oe hd, d Est d Southest Afric, o the other. I the ltter the religious gp i eductio is geerlly much smller d i Ugd the eductiol levels of Muslims d Christis re sttisticlly idistiguishble. Amog proximte determits of fertility religious differeces re likely to mifest themselves i ge t the oset of exposure to the risk of coceptio. Tble 1 presets religiospecific me ges t first mrrige, proxy for the strt of childberig creer (reports o ge t first itercourse seem much less relible). As oe could expect bsed o the eductiol differeces, Muslims cosistetly mrry erlier th Christis, d the gp geerlly persists over time, eve though the ge t first mrrige mog both Muslims d 7
Christis hs bee risig stedily. Agi, the gp is prticulrly wide i Nigeri d Cmeroo, lrgely becuse of very erly ge t first mrrige mog Muslims i these two coutries. Totl fertility rtes Religio-specific totl fertility rtes re preseted umericlly i Tble 1 d grphiclly i Figure 1. The regiol ptter of religio-specific TFRs merits ttetio. Thus i most Wester d Cetrl Afric, Muslims hve higher TFRs th Christis. The most cospicuous d suspicious exceptio is Liberi, the coutry of the first ever DHS, where the Muslim TFR is lmost three times lower th the Christi TFR. I Est Afric d Southest Afric the excess of Christi fertility is predomit ptter with Mlwi beig the oly cler exceptio. While TFRs of both Christis d Muslims teded to declie i lmost ll coutries i which more th oe DHS ws coducted, the differeces betwee Muslims d Christis persisted i most of them. I Est Afric the excess of Christi totl fertility teded to dimiish i Key, reversed itself i Ugd, but ppered to icrese i Tzi. I West d Cetrl Afric the chges were ot uiform either: the excess of Muslim totl fertility icresed i Cote d Ivoire, Nigeri d Cmeroo, decresed to reverse itself i Burki Fso, somewht decresed i Togo d Bei, d remied prcticlly uchged i Gh. Figure 1 bout here Figure 2 reltes the differeces betwee Muslim d Christi TFRs (expressed s TFR rtios) to the treds i coutry-level TFRs. The TFR rtio teds s the TFR declies, suggestig tht overll Muslim fertility declies less rpidly th tht of Christis, eve though the tedecy is ot very strog. Figure 2 bout here As idirect d forml test of the miority-sttus thesis we c exmie how religious differeces i TFR re relted to the reltive shre of the group i coutry s popultio. Figure 3 plots the Muslim/Christi TFR rtio by the Muslim/Christi popultio rtio. (Although the popultio s religious compositio remis reltively stble over time, we iclude ll vilble coutry-yers to visulize possible coutry-yer vritios.) The grph suggests tht s the proportio of the Muslim miority rises, the TFR rtio rises too. However, whe Muslims d Christi become mjority, s i Burki Fso, the 8
differeces seem to dimiish gi dimiish, lthough this reversl should ot be extrpolted to coutries with eve lrger Muslim mjorities. Figure 3 bout here Multivrite lysis of childre ever bor Tble 2 presets the estimtes for Muslims (reltive to Christis) of Poisso regressio models for the umber of childre ever bor. I Togo d Bei, two smll, demogrphiclly d historiclly similr djcet coutries, the multivrite results coform to the differeces i TFRs: Muslims, ceteris pribus, hve more childre th Christis. I Gh d Côte d Ivoire, two other eighbors, Muslim-Christi differeces dispper fter cotrollig for other fctors. Filly, i the remiig coutries s diverse s Nigeri, Liberi, d the CAR, Muslims ted to hve fewer childre fter cotrols re dded, d oly for Liberi this treds coforms to the TFR differetils. These ptters re very similr i rurl d urb res (ot show). Wht fctors primrily ccout for the dispperce of the Muslim excess of births or for its reversl i some cses i West d Cetrl Afric? Age t mrrige is oe: s we sw erlier, Muslims ted to mrry much erlier th do Christis (Muslims re lso much less likely to hve premritl births, lthough this compoet of fertility is lrgely beyod the scope of this pper focused o fertility of ever-mrried wome The erlier observed lower levels of Muslims eductio is other strog iterveig fctor. I most Est/Southest Afric Muslim wome ted to hve sigifictly lower fertility, cotrollig for other fctors. Oly Ugd i 15 displys the opposite tred. Muslims deficit of births seems to dimiish over time i Key d Tzi. Oe strikig differece from West/Cetrl Afric is tht the tedecies detected i multivrite tests closely prllel the TFR differetils, which lrgely owes to much smller eductiol differeces th i West d Cetrl Afric. I most Est/Southest Afric coutries, Muslims lso ted to hve sigifictly loger birth itervls (Tble 2), lthough this is lso the cse for Ugd 15, where the effect of religio o umber of childre ever bor bucks the regiol tred. Agi, s i West d Cetrl Afric rurl-urb vritio i religious prity differetils re miiml (ot show). Cotrceptive use The DHS do ot collect complete cotrceptive history d therefore likig prity d cotrceptive use directly is ot possible. Yet this coectio is cetrl to fertility trsitio. Cotrceptive prevlece i sub-shr coutries remis lower
th i most of the less developed world but it hs bee risig lmost i ll of them, s the DHS evidece suggest. Although most of the chges re drive by the use of moder methods, some West Afric DHS detect very high prevlece of turl d idigeous cotrceptio (lthough the reportig of trditiol cotrceptio my be less ccurte th tht of moder methods). Religious differeces i cotrceptive prevlece rtes Figures 4 d 5 depict curret cotrceptive prevlece rtes mog ever-mrried Christi d Muslim wome for ll methods d moder methods, respectively. I ll West d Cetrl Afric coutries Muslims hve lower levels of overll cotrceptive use, d i some cses (e.g., Nigeri, Cmeroo) the religious differetils re very proouced. The sme ptter emerges whe we look t the use of moder methods oly: gi i ll West- Cetrl Afric Christis hve higher rtes, with the exceptio of the Cetrl Afric Republic, coutry o the geogrphic mrgis of the regio, where Muslims seem slightly hed of Christis. The religious differetils ted to persist or oly slightly declie over time i coutries where more th oe DHS ws coducted. Figures 4 d 5 bout here The picture i Est d Southest Afric is less cosistet. While Key, Mozmbique, d Mlwi re similr to West-Cetrl Afric i tht Christis hve higher moder cotrceptive rtes, i Tzi d Ugd Muslims re t lest s likely to use cotrceptio s Christis. I fct, Tzi Muslims dvtge i the use of moder methods seems to wide over time d i Ugd the sme is true of y method use. Yet, i both regios, the Christi-Muslim differetils ted to declie s cotrceptive prevlece rises. This tred is illustrted i Figures 6 d 7 tht plot the Muslim/Christi cotrceptive odds rtios (udjusted) gist the prevlece of ll d moder cotrceptive methods, respectively. Figures 6 d 7 bout here The regiol differeces re lso illustrtive. I West/Cetrl Afric, s cotrceptive prevlece rises, the Muslim disdvtge decreses i ll methods. However, i moder methods loe the tred is less cler. I Est/Southest Afric, the llmethod tred is i the opposite directio, which lso suggests covergece s the Muslim dvtge, where preset, is eroded with the rise of overll cotrceptive prevlece. The crossover is more potet with respect to moder methods, s the Muslim dvtges turs ito disdvtge. As with TFRs, we look t possible ssocitio betwee the reltive sizes of ech religious commuity d their cotrceptive prevlece. Oe wy to ssess the relevce of miority-sttus hypothesis is by exmiig the ssocitio 10
betwee religious differetils i cotrceptive prevlece d the religious structure of the popultio. Figures 8 d plot the Muslim/Christi odds rtios i ll-methods d modermethods use gist the proportio of Muslims i the popultio ( miority i ll the exmied coutries but Burki Fso). A tred, lbeit wek oe, ideed emerges: the higher the shre of Muslims the greter their cotrceptive disdvtge. (The tred would hve probbly bee more proouced hd we bee ble to iclude coutries where Christis re smll miority.) Iterestigly, the tred seems prticulrly strog i West/Cetrl Afric. I dditio to the reltive size of the two religious groups, oe c lso look t their sptil distributio. Although the roots d shpes of sptio-culturl cofigurtios of Muslim d Christi popultios re coutry-specific, oe c rgue, i the spirit of diffusioist rgumets, tht the more religiously segregted cotext is, the greter cotrceptive differetils betwee Muslims d Christi would be. Figure 10 plots the Muslim/Christi cotrceptive odds rtios gist the correspodig vlues of the idex of dissimilrity. The grph suggests tht Muslim disdvtge icreses with greter segregtio. This tred is most proouced i Est-Southest Afric, lthough the smll umber of dt poits clls for cutio i iterprettio. Figure 10 bout here Multivrite lysis of cotrceptio To isolte the et effect of religio differetils o cotrceptive use, we gi cotrolled for vilble sociodemogrphic chrcteristics. Tble 2 reports the Muslim/Christi odds rtios (djusted for cotrols) for y-method curret use. Figure 11 depicts them grphiclly. The disdvtge of Muslims i West Afric is overwhelmig: oly i Togo d Gh, two eighborig coutries, the odds rtios re ot sigifict. I the other coutries the differeces re strog d ted to icrese over time; the wideig of the Muslim-Christi gp is prticulrly oticeble, gi, i Nigeri d Cmeroo. Figure 11 bout here I Est d Southest Afric the Muslim disdvtge is geerlly less proouced d the temporl ptter is somewht more errtic. Thus the Muslim-Christi differetil is sigifict i the first Key s DHS but disppers i the subsequet two. I Tzi, o the cotrry, oly the ltest DHS ttests to y Muslim disdvtge. Filly, Ugd is the oly coutry where Muslims re mrgilly hed of Christis i y-method cotrceptive prevlece, cotrollig for other fctors. Whe we look t moder methods oly (Tble 2 d Figure 12), the et religious differetils dimiish cosiderbly. Yet, i Nigeri, Bei, d Cmeroo they ted to icrese over time. The 11
puzzlig (d suspicious) dvtge of CAR Muslims, lredy observed t the bivrite level, remis mrgilly sigifict; the sme dvtge of Tzi s Muslims is cofirmed oly for the 16 DHS. Figure 12 bout here Cotrceptio d fertility While the egtive ssocitio betwee cotrceptive levels d fertility rtes is geerlly see s sig of cotrceptio s ggregte effect o fertility, it hs bee lso oted tht this ssocitio is less proouced i sub-shr Afric th i the rest of the developig world (Westoff d Bkole 2001). Figures 13 d 14 depict the ssocitio betwee moder methods prevlece d TFRs for both regios d both religious groups. As we c see, the expected egtive ssocitio is wekest mog Muslims of West/Cetrl Afric. I compriso, Est/Southest Afric Muslims disply very similr ptter to tht of Christis (lthough the smll umber of dt poits gi cutios gist overgeerliztios). Figures 13 d 14 bout here Fertility prefereces Although fertility prefereces s sub-shr wome stte them i surveys re ot likely to be fully implemeted, there is strog evidece tht these prefereces re predictive of fertility outcomes d re criticl for fertility chge (Bkole 15; Bkole d Westoff 18; Pritchett 14). It my therefore be istructive to compre these prefereces mog Muslims d Christis. The itetio to hve o more childre seems the most strightforwrd of ll the stdrd bttery of DHS fertility preferece mesures; it is lso believed to predict subsequet fertility better th other mesures (Hermli et l. 17; Lightboure 185; Rodgers 176). Here it is formulted s dichotomy: wishig to stop childberig vs. wishig to cotiue or beig udecided. Tble 2 presets Muslim/Christi odds rtios, djusted for the umber of livig childre d other cotrols. (The odds rtios re lso depicted i Figure 15) Christis re overwhelmigly more likely th Muslims to wish to ed childberig (occsiol devitios from this ptter, s i the cse of Cmeroo 11 re difficult to expli). Eve i Est d Southest Afric, where Christi- Muslim differetils i cotrceptive use proved less strog d cosistet, Christis geerlly look more ti-tlist th Muslims. Iterestigly, however, the religious gp teds to declie i most of West Afric (with puzzlig exceptio of Cmeroo), but ppers to icrese i Est Afric coutries 12
with multiple DHS (eve if it remis geerlly smller th i West Afric coutries). Figure 15 bout here Discussio d coclusio The Muslim-Christi divide i sub-shr Afric hs differet fces: from reltively iocuous culturl distictios, to prejudice, segregtio, d discrimitio, to bloody cofrottios betwee ggs of religious ftics. Demogrphic profiles d treds i both religious groups re fudmetl to the shpe, cotet, d itesity of Muslim-Christi coexistece d iterctios. Yet these demogrphic chrcteristics re i tur shped by umber of fctors re both iterl d exterl to ech group. I explortory study tht offers but bird s view of Muslim- Christi fertility d cotrceptive differetils it is impossible to ddress ll the relevt coutry-specific demogrphic d culturl uces. Thus to keep our presettio focused d cocise, we did ot ddress pre-mritl childberig. A cursory explortio of this mtter o the side of this study poits to cosiderbly higher pre-mritl fertility mog Christis. As both ge t first mrrige d premritl childberig rises, the religious differetils i it my ply d icresig role i overll Christi-Muslim fertility differetils. Aother chllege tht this study did ot ddress is disetglig the effects of religio from tht of ethicity d liege systems. Historiclly, the spred of both Islm d vrious Christi deomitios i Afric ofte followed ethic boudries, lthough i more recet er religio hs icresigly cut through ethic lies. The study did ot ccout for the diversity withi the Christi popultio, which c be quite substtil d c ifluece the shpe d stregth of Muslim-Christi differetils. The most importt limittio of this study, however, is tht it lcked pproprite dt to exmie differet spects of religious existece d coexistece, d especilly to exmie religious groups s commuities (Agdji 2001; Goldscheider d Mosher 188). Withi such religious commuities, dherece to certi prctices tht is iitilly iflueced by the religious doctrie d the positios tke by religious leders is further reiforced by socil pressure d cotrol. 13
Much of the effect of this milieu opertes through the complex web of socil orms, customs, d reltioships tht re ofte referred to s wome s sttus. DHS offers some crude proxies for wome s sttus such s wome s eductio or ge t first mrrige but the list of vilble mesures is too short d their vlidity is questioble. More comprehesive d cotextembedded mesures of wome s sttus re eeded if we re to explore the effects of religio further. Wome s eductio, ge t first mrrige, d other predictors used i multivrite tests helped expli some of the religious differetils. The excess of Muslims TFRs observed i some coutries disppered d i some cses reversed itself: fter ddig cotrols, especilly ge t first mrrige, Muslims hd lower fertility th Christis i most coutries. The presece of these differeces remis puzzlig, especilly give Muslims lower cotrceptive use. While Muslim disdvtge i modermethod use were ot s proouced s i y-method use, it is i fertility prefereces (the stoppig itetio) tht the Muslim-Christi differeces proved prticulrly strk. Filly, we lso oted ssocitio betwee religious differetils i fertility d cotrceptio, o the oe hd, d the two religious groups reltive size d sptil distributio, o the other. A covetiol coclusio from this study s fidigs would be to ckowledge tht it hs let some support to the chrcteristics hypothesis, some support to the mioritysttus perspective, but hs left much of the religious vritio uexplied. Yet, s we proposed t the outset, this study s fidigs should ot be see rrowly s other test of the estblished hypotheses. Isted these fidigs cll for more cotextulized view of reproductive choices d behvior. Our regiol focus o West d (djcet prt of) Cetrl Afric, o the oe hd, d Est d Southest Afric, o the other ws oe ttempt i tht directio. While either regio is homogeeous, meigful regiol-level differeces i fertility d especilly i cotrceptive use, s well s some of their determits could be clerly estblished. The differeces betwee West-Cetrl Afric d Est-Southest Afric should be iterpreted withi the coloil d postcoloil politico-culturl cotext i which Christiity d Islm hs coexisted for more th cetury. I West-Cetrl Afric, the Christi-Muslim divide hs trditiolly bee more 14
strogly rticulted. I Nigeri, for exmple, Islm ws the mobilizig ber of the fierce resistce to the Christi British expsio. Iroiclly, durig the coloil er the British idirect rule further stregtheed the Muslim elites d fueled Muslim-Christi tesios. Frech coloies were chrcterized by strog (eve if ot lwys explicit) politicl d culturl domitio of Christiity (especilly Ctholicism), which cotiuously mrgilized Islm. This ws lso the cse of Liberi, where the iterl coloilism ws prticulrly hrsh o Muslims. The Muslim culturl d politicl disdvtge hs edured ito the idepedece period, lthough i most coutries it hs ot resulted i commul cofrottios util very recetly. Muslim-Christi reltios i Est d Southest Afric hve very differet history. Muslims there, especilly i the core coutries Tzi, Key, d Ugd were ever i the vgurd of the resistce to the coloil coquest d domitio, s i Nigeri, or suffered excessive discrimitio s i Frech Afric. Iterestigly, whe the Muslim elites rose up gist the coloil rule, s it hppeed i Zzibr, they chose Mrxism-Leiism isted of Islm s the ideologicl tool of politicl mobiliztio. The sigificce of religious differetils i fertility should goes beyod the rrow demogrphic cofies. A curious yet ievitble reflectio of globl tedecies d evets, the Muslim-Christi politicl tesio i Afric is escltig where it ws simmerig for geertios d is emergig eve i plces tht re used to be thought of s religiously hrmoious. The differece i growth rtes betwee Muslims d Christis due to fertility differetils, coversio, or proselytism my upset the religious compositios of the popultio d cosequetly the precrious politicl blce i my sub-shr society. As globliztio brigs dow iformtiol brriers oe my lso expect virtuliztio of religious commuities, s differet religious groups Muslims, Ctholics, Protestts, Petecostls coect with their brethre cross the sub-cotiet d the world i growig perceptio of commo destiy. This my be prticulrly the cse of Islm whose followers ll roud the globe sub-shr ot beig exceptio icresigly perceive their religio d imgied de-territorilized commuity ( sort of virtul umm) s threteed by the Judeo-Christi politicl d culturl oslught. 15
As umerous studies of wester fertility hve show d s this study lso suggests religious fertility differetils i Afric should evetully dispper. However, the culturl d politicl schism log the religious lies, d especilly the cotiuig tgoiztio d mrgiliztio of Islm my slow dow the covergece by ffectig both Muslims cceptce of wester cotrceptio d perpetutig mritl prctices d geder orms tht re coducive to higher fertility. Refereces Addi, I. 1. Does religio mtter i cotrceptive use mog Ghi wome? Review of Religio Reserch 41: 328-343. Agdji, V. 2001. Religio, socil milieu, d the cotrceptive revolutio. Popultio Studies, 55 (2): 135-148. Agdji, V. 2003. Rom Ctholics s foreruers of the fertility revolutio i sub-shr Afric Pper preseted t the Aul Meetig of the Popultio Associtio of Americ, Miepolis, MN. Avog, H. N. 2001. Religio d fertility mog the Atyp i Nigeri. Jourl of Biosocil Sciece 33(1):1-12. Biley,M. 186. Differetil fertility by religious groups i rurl Sierr Leoe. Jourl of Biosocil Sciece 18: 75. Bkole, A. 15. "Desired Fertility d Fertility Behviour mog the Yorub of Nigeri: A Study of Couple Prefereces d Subsequet Fertility." Popultio Studies 4(2):317-328. Bkole, A., d C. F. Westoff. 18. "The cosistecy d vlidity of reproductive ttitudes: Evidece from Mrocco." Jourl of Biosocil Sciece 30(4):43-455. Berhu, B. 14. "Religio fertility differetils i Shew, Cetrl Ethiopi." Jourl of Fmily Welfre 40(1):22-. Gisie,S.K. 172. Fertility levels mog the Ghi tribes. I Popultio Growth d Ecoomic Developmet i Afric. Ed. By S.K. Omide d C.N. Ejiogu. Heiem, Lodo. Goldscheider, C. d W. D. Mosher. 188. Religious ffilitio d cotrceptive usge: Chgig Americ ptters, 155-82 Studies i Fmily Plig 1 (1): 48-57. Goldscheider, C. d W. D. Mosher. 11. Ptters of cotrceptive use i the Uited Sttes: The importce of religious fctors, Studies i Fmily Plig 22(2): 102-115. Gregso, S., T. Zhuwu, R.M. Aderso, S.K. Chdiw. 1. Apostles d Zioists; The ifluece of religio o demogrphic chge i rurl Zimbbwe. Popultio Studies 53(2): 17-3. 16
Hermli, A.I., R. Freedm, T.H. Su, d M.C. Chg. 17. Do itetios predict fertility? The experiece i Tiw, 167-74. Studies i Fmily Plig 10(3): 75-5. Krim, M. S. 17. Reproductive Behvior i Muslim Coutries. DHS Workig Ppers, Number 23. Mcro Itertiol d UNFPA: Clverto, Mryld d New York. Kollehlo, K. T. 14. "Religious ffilitio d fertility i Liberi." Jourl of Biosocil Sciece 26(4):43-507. Lightboure, R.E. 185. Idividul prefereces d fertility behviour i J. Cleld d J. Hobcrft (eds) Reproductive Chge i Developig Coutries New York: Oxford Uiversity Press, pp. 165-18. Mosher, W. D., L. B. Willims, D. P. Johso. 12. Religio d fertility i the Uited Sttes: New ptters, Demogrphy 2(2): 1-214. Pritchett, L. H. 14. Desired fertility d the impct of popultio policies Popultio d Developmet Review 20:1-55. Rodgers, G.B. 176. Fertility d desired fertility: logitudil evidece from Thild Popultio Studies 30 (3): 511-526. Sembjwe, I. 180. Religious fertility differetils mog the Yorub of Wester Nigeri Jourl of Biosocil Sciece 12:153. Westoff, C. F. d A. Bkole. 2001. The Cotrceptio Fertility Lik i Sub-Shr Afric d Other Developig Coutries. DHS Alyticl Studies No.4. 17
Tble 1. Bsic demogrphic chrcteristics of Christis d Muslims Coutry/DHS yer Popultio shre (%) Mus. Chris. Idex of dissim. Me eductio (yers) Mus. Chris. Me ge t first mrrige (yers) Mus. Chris. Totl fertility rte Mus. Cotrceptiv e prevlece y method (%) Cotrceptiv e prevlece rte, moder method (%) Chris. Mus. Chris. Mus. Chris. Togo 188 11 38 0.48 0. 3.8 17.8 18.5 7.41 5.00 31.8 32.8 2.3 6.4 Togo 18 11 51 0.53 1.2 3.5 18.2 18. 5.27 4.05 23.0 26.4 6. 10.1 Cote d'ivoire 14 33 43 0.23 0. 3.1 17.3 18.2 5.6 4.54 7.2 1.8 3.5 6.8 Cote d'ivoire 18 35 41 / 1.1 3. 17.3 18.6 6.20 4.11.3 25.5 5.6 11.4 Burki Fso 14 55 33 0.2 1.8 16.7 17.6 6.75 6.11 22.5 30.7 3.6 8.0 Burki Fso 18 56 27 0.2 0.5 1.2 17.0 17.8 6.36 6.42 11.0 18.1 4.7 7.7 Gh 188 10 70 0.30 1.8 5.6 17.2 17.8 6.0 6.02 11.3 14.6 5.0 6.2 Gh 13 12 72 0.45 2.0 6.2 18.3 18.6 5.67 4.72 11.3 23.3 6.0 11. Gh 1 11 78 0.33 2.5 6.6 18.2 18.8 4.7 4.20 16.2 22.3 10.4 13.5 Nigeri 10 48 48 0.75 0. 3. 15.3 18.0 6.25 5.83 2.4 10.7 1.6 6.1 Nigeri 1 43 55 0.54 1.8 6.5 15.6 18. 5.45 4.14 6.3 25.4 3.8 14.0 Bei 16 21 46 2 1.0 1. 17.7 18. 6.44 5.34 10.1 22.6 3.2 4.4 Bei 2001 22 52 1 1.2 2.3 17.6 1.0 5.88 5.08 10.0 22.6 6.2 8.8 Liberi 186 11 60 0.08 1.0 2. 16. 17.3 2.31 6.10 3.5 10. 2..1 Cmeroo 11 21 67 0.71 0.8 4.8 14.4 17.3 6.03 5.76 5.5 23.1 1.3 6.6 Cmeroo 18 1 73 0 1.5 5. 15.2 17.8 5.82 4.53 5.1 27.2 2.5 10.0 CAR 15 0 0.23 1.1 2.1 15.8 16. 5.27 5.07. 15.0 4.0 3.4 Key 18 4 2 0.72 3.3 4.6 16.6 17.6 5.25 6.76 16.6 27.5 14.3 18.4 Key 13 5 1 8 4.1 5.5 17.4 18.3 4.27 5.41 27.2 33.0 25.2 27.6 Key 18 5 2 4 5.4 6.5 17.8 18.5 4.60 4.70 32.0 37.8 24.8 31.0 Mozmbique 17 17 55 6 1.4 2.2 15. 17.0 4.54 5.41 4.7 7.4 3.3 6.8 Tzi 12 31 55 0.52 18.0 4.1 17.1 3.7 5.70 6.44 11.4 12.7 8.0 8.4 Tzi 16 31 57 0.57 4.5 4.8 17.4 18.3 5.20 5.5 21.8 20.4 17.3 14.6 Tzi 1 33 54 0.55 4.7 5.0 17.5 18.1 4.64 5.75 27.4 27.3 22.6 17.8 Ugd 188 10 0 0.2 2. 2.8 16.0 16. 7.22 7.47 5. 5.5 3.7 2.7 Ugd 15 12 85 0.21 3.4 3.4 16.8 17.0 7.15 6.83 17.6 13.8 10.1 7.4 Mlwi 2000 13 85 0.30 2.2 4.2 17.4 17.5 7.26 6.24 22.5 2.0 17. 25.2 18
Figure 1. TFRs by religio 8.00 7.00 6.00 5.00 TFR 4.00 3.00 2.00 1.00 0.00 DHS Coutry/Yer M uslims Christis Figure 2. Muslim/Christi TFR rtio by coutry TFR 1.6 1.4 1.2 1 0.8 TFRm/TFRc 0.4 8 7 6 5 4 0.2 Coutry TFR 1
Figure 3. Muslim/Christi TFR rtio by Muslims' d Christis' reltive size 1.6 1.4 1.2 1 0.8 TFRm/TFRc 0.4 0.2 0 0.5 1 1.5 2 2.5 Shre Ms/Shre Cs 20
Tble 2. Muslim-Christi differeces, multivrite results Childre ever bor Coutry/DHS yer Poisso βs Durtio of lst birth itervl OLS βs Curret use of y method Odds rtios Curret use of moder methods Odds rtios Wts o more childre Odds rtios Togo 188 0.12 Togo 18 0.05 1.67-0.45 Cote d'ivoire 4 0.00-2.81 Cote d'ivoire 8 0.03-1.62 Burki Fso 14 0.00 0.58 Burki Fso 18-0.04 0.38 Gh 188 0.03 0.81 Gh 13 0.03 0.81 Gh 1 0.03 0.81 Nigeri 10-0.04 0.38 Nigeri 1-0.06 1.43 + Bei 16 0.03 0.57 Bei 2001 0.08 0.46 Liberi 186-0.04 + -1.81 Cmeroo 11-0.15 7 Cmeroo 18-0.04 CAR 15-0.07 2.07 1.1 3 0.55 0.32 0. 5 0.83 0.36 0.5 3 2.31 0.55 0.5 1 0.73 1 0.7 3 0.76 0.80 7 0.87 1.03 0. 6 1.08 0.51 0. 6 1.08 0.30 0. 6 1.08 0.3 7 0.87 0.5 0.4 5 0.33 0.55 0.4 3 0.0 0.32 0.4 0 0.4 2 0 2 0.5 2 0.50 + 1.4 + 0.3 1 0.41 0 5 1.70 + 0.5 Key 18-0.07 0.02 5 0.87 8 + Key 13-0.07 0.38 0.8 6 1.01 0.0 Key 18-0.04 3.70 0. 0 0.81 0 Mozmbique 17-0.14 1.76 0.8 5 8 0.46 0. Tzi 12-0.06 1.64 4 0.5 0.87 + Tzi 16-0.0 3.83 1.1 1.18 0.81 21
0 0. Tzi 1-0.10 5.62 2 1.16 0.78 + Ugd 88 0.01 1.0 1.0 6 1.50 1.1 Ugd 5 0.05 2.4 1.2 4 + 0.76 0.84 + 0.5 Mlwi 2000 0.03-1.62 1 0.73 0.57 Note: sigificce level p<.05, + p<.10 22
23 Figure 4. Curret use, y cotrceptive method, by religio 0 5 10 15 20 25 30 35 40 T o 8 8 T o g C o t C o t B u r B u r G h 8 G h G h N i g N i g B e B e 0 L i b 8 C m C m C A R K e 8 K e K e M o z T T T U g 8 U g M l 0 Percet usig Muslims Christis Figure 5. Curret use, moder cotrceptive method, by religio 0 5 10 15 20 25 30 35 40 T o 8 8 T o g C o t C o t B u r B u r G h 8 G h G h N i g N i g B e B e 0 L i b 8 C m C m C A R K e 8 K e K e M o z T T T U g 8 U g M l 0 Percet usig Muslims Christis
Figure 6. Muslim/Christi curret cotrceptio odds rtios (udjusted), by totl cotrceptive prevlece, y method 1.4 1.2 Muslim/Christi odds rtios 1 0.8 0.4 0.2 0 Totl curret cotrceptive prevlece (y method) 1.6 Figure 7. Muslim/Christi curret cotrceptio odds rtios (udjusted), by totl cotrceptive prevlece, moder methods 1.4 Muslim/Christi odds rtios 1.2 1 0.8 0.4 0.2 0 Totl cotrceptive prevlece (moder methods) 24
1.4 Figure 8. Muslim/Christi curret cotrceptio odds rtios (udjusted), by percetge of Muslims i the popultio, y method Muslim/Christi odds rtios 1.2 1 0.8 0.4 0.2 0 0 10 20 30 40 50 60 Percetge of Muslims i the popultio 1.6 Figure. Muslim/Christi curret cotrceptio odds rtios (udjusted), by percetge of Muslims i the popultio, moder method Muslim/Christi odds rtio 1.4 1.2 1.0 0.8 0.4 0.2 0.0 0 10 20 30 40 50 60 Percetge of Muslims i the popultio 25
Muslim/Christi odds rtios 1.6 1.4 1.2 1.0 0.8 0.4 0.2 0.0 Figure 10. Muslim/Christi curret cotrceptio odds rtios (udjusted), by idex of dissimilrity i the popultio, y method Idex of Dissimilrity 1.6 Figure 10b. Muslim/Christi curret cotrceptio odds rtios (udjusted), by idex of dissimilrity i the popultio,moder method Muslim/Christi odds rtios 1.4 1.2 1.0 0.8 0.4 0.2 0.0 Idex of Dissimilrity 26
Figure 11. Muslim/Christi curret cotrceptio odds rtios (djusted), y method 1.4 1.2 Muslim/Christi odds rtios 1 0.8 0.4 0.2 Tog88 Tog8 Cote4 Cote8 Burk4 Burk8 Gh88 Gh3 Gh Nigeri0 Nigeri Bei6 Bei01 Lib86 Cm1 Cm8 CAR 15 Key8 Key3 Key8 Mozm7 Tz2 Tz6 Tz Ug 88 Ugd5 Mlwi00 p<.05 p<.1 p>.1 2.5 Figure 12. Muslim/Christi curret cotrceptio odds rtios (djusted), moder method Muslim/Christi odds rtios 2.0 1.5 1.0 0.5 0.0 Tog88 Tog8 Cote4 Cote8 Burk4 Burk8 Gh88 Gh3 Gh Nigeri0 Nigeri Bei6 Bei01 Lib86 Cm1 Cm8 CAR 15 Key8 Key3 Key8 Mozm7 Tz2 Tz6 Tz Ug 88 Ugd5 Mlwi00 p>.1 p<.05 p<.1 27
8.0 Figure 13. Muslim totl fertility rte by curret cotrceptio, moder methods 7.5 7.0 Totl fertility rte 6.5 6.0 5.5 5.0 4.5 4.0 0 5 10 15 20 25 30 35 Moder cotrceptive prevlce, Muslims West/cetrl Afric South/Est fric West/Cetrl Afric sout/est Afric All Afric 8.0 Figure 14. Christi totl fertility rte by curret cotrceptio, moder methods 7.5 7.0 Totl fertility rte 6.5 6.0 5.5 5.0 4.5 4.0 0 5 10 15 20 25 30 35 Moder cotrceptive prevlce, Christis Sout/Est Afric West/Cetrl Afric South/Est Afric West/Cetrl Afric All Afric 28
1.6 Figure 15. Odds rtios (djusted), Muslim/Christis, preferece for o more childre 1.4 Muslim/Christi odds rtios 1.2 1 0.8 0.4 0.2 0 Mlwi00 Ugd5 Ug 88 Tz Tz6 Tz2 Mozm7 Key8 Key3 Key8 CAR 15 Cm8 Cm1 Liberi86 Bei01 Bei6 Nigeri Nigeri0 Gh Gh3 Gh88 Burki8 Burki4 Cote8 Cote4 Togo8 Togo88 p<.05 p<.1 p>.1 2