Women's Access to in : Effects of Education, Residence, Lineage and Self-Determination

John Boateng* and Constance Flanagan

Department of Agricultural and Extension Education, The Pennsylvania State University, University Park Campus, State College, PA

ABSTRACT: Women*» physical and psychological access to health care was analyzed using the 2009 Ghana Demographic and Health Survey (GDHS), a nationally representative .study for monitoring population and health in Ghana. Female respondents from the 2133 cases in the couple's data set were used in this study. Women's level of education was positively related to physical but not to psyvhologicul access to health care. Residing in an urban area was posi- tively related to both types of access. MatriHny consistently showed positive effects on physical acce.ss. In addition tu these demographic factors, both physical and psychological access were positively related to women^s self-determination, i.e., women\s right and ability to make reai choices about their lives including their health, fertility, sexuality, childcare and all areas where women are denied autonomy and dignity in their identities as women. Self-determination factors huth mediated the effects of background factors on access and added explanatory power to the models.

INTRODUCTION However, physical access represents only one access issue. Even when health care is This study exainined women's physical geographically close and transportation and psychological access to health care in availahle, there may be psychological bar- Ghana. It looked at factors including the riers such as women's knowledge and role of education, current residence in rural sense of her right to access care and her and urban contexts, self-determination and subordinate position in family decision gender role norms with respect to women's making. TTiis makes psychological access rejection of domestic violence as these to health care also an important issue, one impacted women'-s health aeeess in Ghana. which unfortunately has received little Much research in Ghana and the world has research attention. focused on women's physical access to health care (Arcury, Beîl, Snively, Smith, It is known from demographic studies Skcily, and Wetmore, 2006; Bour, 2003; that educated women are better than non Gething, Noor, Zurovac, Atkinson, and educated ones in accessing health care, Hays, 2004; Tsoka and Le Sueur, 2004). using contraception, reducing fertility and enjoying many reprtxluclive and child health outcomes (Benefo, 2006). Urban- ization in the nations of sub-Saharan Author's noie: This paper is based on the first author's PhD disseriaiion. Africa is also likely to increase access for •AddiK.ss tonespondence to: John Boaleng. do AEE two reasons: One is the increased number Departmeni, 201 Ferguson Building. Penn Stale University. Siaie College, PA 16802-2601. of health care systems and greater ease of

56 Ghanaian Women's Access to Health Care 57 transportation to them than would be an urban area (Dodoo, 1993). Further- available in rural areas. The second is the more, in an urban area, a woman is likely likely change in community norms and to be exposed to a broader range of infor- support networks associated with the mation sources which should benefit her move from a rural to urban area, what knowledge about access to health care. might be termed a social capital argu- The body of literature does not make it ment. Compared with the gender norms clear whether differences in health access ol" more traditional social networks in between urban and rural women and edu- rural areas, some of which may compro- cated and less educated women may have mise women's health (such as male abu- underlying causes such as differences in sive attitudes and behaviors), the more their self determination or in their control heterogeneous networks in urban areas over certain resources. should be associated with more endorse- The current study addresses such ment of women's self determination and issues. We look at the impact of demo- her right to decide about health care graphic factors including women's educa- needs. tion, age, parity, current residence (rural/ Research shows that patient socioeco- urban) and lineage type (matrilineal or nomic status (SES) affects pro- patrilineai) on physical and psychological files for health status and satisfaction access to health care. In addition, we (Franks and Fiscella, 2002). Similarly assess the role of self-determination fac- women with higher socioeconomic status tors on access issues. may gain in reproductive self determina- Potts (1997) emphasized that females tion, financial autonomy over their repro- with the highest social status gain access ductive and fatnily health decisions and to more resources and also achieve ascribed forms of status associated with greater reproductive success. According their environment and roles. This increased to Potts, once women have unconstrained self determination may well occur through access to a range of fertility-régulât i on the transformation of traditional attitudes options (including safe abortion), family to modern ones including such things as size falls in all groups and in all societies. smaller family size and wotnen's rights. In sucb a context, he notes, social success Women's self-determination also is likely tends to be associated with the accumula- to be affected by changing social support tion of material wealth, rather than with networks and the attitudes about women's having more children. The argument that self determination encouraged in those development causes fertility decline does networks. Likewise, traditional norms not hold because people cannot make associated with gender and power which choices alxiut family size without realistic favor male dominance, even to tbe point access to fertility-regulation technologies. of endorsing domestic violence and the Such access is historically recent and subordination of women's health deci- remains geographically limited. Where sions, should change as women enjoy the access to fertility regulation is con- benefits of education and modernization strained. Potts, (1997) affirms, the richer generally associated with urban living. and tnore educated are usually better able Women's autonomy is known to be posi- than the less privileged to surmount the tively related to education and is fre- barriers between them and tbe needed quently assumed to increase if she lives in technologies. Hence, there is a cotnmon 58 Boateng and Flanagan inverse relationship between income and the effects of specific demographic vari- family size. Much has been written about ables on health access. Alternatively, if the proximate determinants underlying the self-determination variables add fertility decline in Ghana and other devel- explanatory power but the effects of the oping countries (DeRose, 2003; DeRose demographic variables on outcomes are and Dodoo, 2(X)2; Dodoo, 1995; DeRose not reduced, we can conclude that the and Ezek, 2005; Tawiah, 1984). However, models are additive rather than mediating. little is known about the soc i o-cu litoral fac- tors that may be related to changes in women's altitudes and roles in household BACKGROUND AND CONTEXT decision making which also tnay impact Ghana is located in West Africa and their health access. It is possible that differ- borders the Gulf of Guinea, between Cole ences in self-determination factors associ- d'Ivoire and in the southern end. ated with living in urban areas mediate the The capital city, Accra, is situated on the effects of urbanization and education on Atlantic coast. There are over one hundred health access. Alternatively, women's self different ethnic groups and languages determination may have additive effects spoken in Ghana. Some of the major eth- over and above the effects of various nic groups are Akan, Ga/Adangbe, Ewe demographic factors. Thi.s study will test and Dagbani. these alternative explanations. Gender disparities, which plague most First, we look at the relationships of the developing world, also pose prob- between demographic factors (in particu- lems in Ghana. For example, Ghanaian lar, education level, rural/urban residence, women lag hchind men in educational age, number of children under six years attainment. The Ghana Living Standards (parity), lineage type) on self-determina- Survey (GLSS 4) shows that 44.1 percent tion factors that affect women's health of women compared to 21.1 percent of access decisions. Here we test whether men had no formal education (Ghana Sta- women's education, age, parity, residence tistical Service, 2000). Similarly, the and lineage affect her self-determination Ghana Demographic and Health Survey (that is, her say in purchase decisions, say of 2003, shows that nearly 50 percent of in family matters and her attitudes women have no schooling, 30 percent towards domestic violence). Second, we have secondary (both junior and senior examine ihe impact of several demo- secondary) and only 2 percent have graphic indicators (education level, age, higher education. Factors such as poverty, parity, current residence, patrilineal and early marriage, and teenage pregnancy matrilineal lineage) and self-determina- prevent females from continuing their tion on health care access outcomes. education to the tertiary level. As a conse- Here, we test whether, in addition to the quence, the majority of women who have demographic variables, women's self neither higher education nor marketable determination explains additional vari- skills are unable to obtain jobs in the for- ance in her health care access outcomes. mal sectors. Therefore women, as well as If the effects of demographic variables on men, migrate to find greener pastures in the outcomes are reduced by the inclusion the urban areas. of (he self-determination variables, we Although ihere are differences based on can conclude that these factors mediate ethnicity and on rural or urban residence. Ghanaian Women's Access to Health Care most Ghanaian women face common falls ill over a fortnight period of which health problems. Chronic health prob- approximately 67% has to stop their usual lems, , pregnancy, disability and activities. When ill, 8% of those ill even cancer are some examples of Ihe receive no treatment and 50% resort to health concerns that afflict women. Gha- self-medication, 22% consult doctors or naian women share certain health care pharmacists, while 15% consult nurse or concerns common for women in develop- medical assistants and 5% rely on tradi- ing countries. Access to health care and tional health care providers. information about family planning are Similarly, Agyepong (1999) show in high on those lists (AbouZahr, 1997; her study of tbe Dangme West district of Furuta and Salway, 2006; Lindstrom and the Greater Accra region that with an esti- Munoz-Franco, 2006). mated 1997 mid-year population of 103, Geographical accessibility to health 210 and divided into four smaller admin- services has a direct bearing on the utili- istrative units each with population rang- zation of these services {Arcury et al., ing from 20,000 to 30,000, the public 2006; Bour, 2004). In rural areas of health sector health services consisted of Ghana, approximately 40% of the popula- four rural health centers with about 10-25 tion has to walk 15 km to receive medical staff each and five community clinics attention. As a resull of the distance to Ihe staffed by 2 to 3 community health nurses health posts, 48% of rural children are not each. With no hospitals or laboratory ser- vaccinated, 62% of children below 5 vices, tbe overall nurse: population ratio years of age do not receive postnatal care was 6: IO,(MM) with one sub-district having and 30% of women do not receive any only 4: 10.000 and tbe two sub-districts pre-natal care (Michaels, Schumann, and closest to Accra, the nation's capital hav- Kliavkov, 2(K)1). Geographical accessi- ing as many as 8:10,000 (Agyepong. bility affects women in two ways: The 1999). detrimental impact on their health and the World Health Organization's Statisti- time and effort spent in traveling, accom- cal Information System (WHOSIS) on panying or carrying their children to the Ghana released in 2(X)7 shows that there health posts. was a total physician number of 3,240 in On the national radar (including both 2004 with density per 1000 rural and urban areas), Van den Boom, population of 0.15 in 2004). The total Nsowah-Nuamah, and Overbosch, (undated) number of nurses in Ghana according to show that, Ghana has a fair number of WHOSIS was 15,797 in 2004 with nurse facilities (I per 10, 0(K) inhabitants and densily per 10(X) population of 0.74 per 140 square kilometers). D(Ktors are (2004). Similar low figures are reported scarce (i per 11,000 inhabitants). Both for midwives (number: 3,910; density per medical facilities and doctors are very 1000 population: 0.18 in 2004), dentists unevenly spread in favor of urbanized (number: 393; density per 1000 popula- regions. Almost a third of the Ghanaian tion: 0.02 in 2004) and pharmacists (num- population lives outside a 5-km radius of ber: 1, 388; density per I(K)O population: medical assistants and nurses while a 0.06 in 2004). quarter is more than !5 km away from a These bealtb statistics from Ghana doctor. Demand side figures indicate that contrasts clearly with statistics from about 25% of the Ghanaian population developed countries such as Sweden 60 Boateng and Flanagan

(Physician number: 29,122, density per lacks psychological access to care. In cer- 1000 population: 3.28 in 2002; Nurses tain cases women may refuse to access number: 90,758, density per 1000 popula- available health care because ibey do not tion; 10.24 in 2002; Midwivcs number: think they can be seen by female practi- 6.247; density per 1000 population: 0.70 tioners. In addition, as others have found, in 2002; Denttst number: 7,270, density women may not use healtb care facilities per 1000 population: 0.82 in 2002; Phar- because tbe health care systems do not macist number: 5,885, density per 1000 adequately address women's needs population: 0.66 in 2002) and United (Chao, 1999). Systems may offer inade- States of America (Physician number: quate gynecological services or contra- 730,801, density per 1000 population: ception and some providers fail to inform 2.56 in 2000; Nurses nutiiber: 2,669,603, households about women's specific density per 1000 population: 9.37 in health needs. 2000; Dentist nunitber: 463,663, density Sotne traditional practices also rein- per IOÍX) population: 1.63 in 2()()(); Pbar- force women's subordinate position add- tîiacist number: 249,642 density per 1000 population: 0.88 in 2000). ing to tbe psychological inhibitions tbat keep women from getting health informa- Based on work done in Kumasi in the tion tbat they need. For example, puberty Ashanti region, a relatively uitan region rituals among the Ewes of Southern compared to more remote areas such as Ghana include indoctrination of girls and Northern, Upper East, Upper West and women about tbe power of supernatural Brong Ahafo regions, Bour (2004) revealed forces to affect women's judgments. that Ghanaian females have a greater need When such indoctrination is internalized for health services tban males. Yet tbey do early in life, it becomes a "psychic bar- not utilize bealtb services as much. rier" from which women cannot easily Physical access is not the only health escape (Abotchie, 1997). access issue facing Gbanaian women. Even wben health care is geographically close and transportation available, there may be METHODS psychological barriers such as a woman's The data source for this study was tbe knowledge, her sense of ber right lo 2003 Ghana Demographic and Health access care, or her subordinate position in Survey, GDHS, of 5, 691 women ages family decision making (D'Atnbruoso, 15-49 and 5,015 men ages 15 - 59 from 2006). We know, for example, thai eco- 6, 251 households covering 412 sample nomic factors contribute to disparities in points throughout Ghana. Of the couple's access to care (Arhin-Tenkorang, 2(X)1). dalaset analyzed, consisting of 2133 cou- If gender norms require a woman to bave ples, only female respondents were con- a husband's permission before sbe can sidered. Tbe survey used a two-slage access bealth care - wbetber to spend the sample based on the Ghana 20(X) Popula- money on health care or to make the trip tion and Housing Census and was to the clinic - we argue in this paper that designed to produce separate estimates sbe lacks psychological access. Similarly, for key indicators for eacb of the ten if she must be accompanied by an escort regions in Ghana. against her will in order lo access health Data collection took place over a care, this also means that the woman three-month period, from late July to late Ghanaian Women's Access to Health Care 61

October, 2003 by the collaborative efforts coding. Table 2 shows descriptive infor- of tbe Ghana Statistical Service (GSS), mation based on the raw data and Table 3 the Noguchi Memorial Institute for Medi- shows the information for the recoded cal Research, and the Ghana Health Ser- (collapsed) variables. For categorical vice. Technical assistance was provided variables, the percentages in each cate- by ORC Macro through the MEASURE gory are presented. DHS program. Funding was provided by As the tables show, 66 percent of tbe U.S. Agency for International Devel- women aged 15^9, in the 2003 GDHS opment (USAID) and the Ghana Govern- couples dataset lived in rural areas. Hie ment (GSS, 2004). mean age of women in tbe sample was 32.33 with the youngest aged 15 and the SAMPLE CHARACTERISTICS oldest 49. On average, women had Summary and descriptive information between one and two children under six about the dependent and independent years old with six children as the largest variables used in this study is provided in family size reported. On average women Tables I, 2 and 3. The first table shows in the sample completed some primary tbe summary of the dependent and inde- schooling. Forty-three percent of women pendent variables describing how Ihey had no education, 18 percent had primary were recoded from the original GDHS education, 37 percent had secondary

TABLE 1 SUMMARY OF DEPENDENT AND tNtœpENM=.NT VARÍABLE.?

Dependent Variables 1. No problem physical access (0-3) • No protilem gelling money No problem = 0 0, 2 recoded as t • No problem regarding distance Small problem = 2 I recoded as 0 • No problem with transport Big problem = 1 2. No problem psychological access (0-3) • Knows where lo go for health care • Permission to access health care Independent Variables No education = 0 ] .2,3, Fecoded as t and 0 as 0 1. Demographic Primary = t • E-ducalion levet (0-1) Secondary = 2 • Age (I/R) Tertiary = ."î • Number of children aged 5/below (l/R) • Current residence (Rural- Urban (0-1 ) Rural = 2. urban = 1 2 recoded as 0. I as 1 • Lineage (patritineal/matrilineal) (0-1 ) Akan lineage = I Non Akan = 0 Akan lineage recoded as t and Non Akan as 0 from Husband wants more children (0-1) Yes = t. No = 0 the original DHS coding of the various ethnic groups in Ghana .Self-determination Woman alone = 1 1, 2, 3 recoded as t • I-inal say in family matters (0-2) Woman and hustiand = 2 • Final say in purchase decisions (0- 3) Wotnan and someone = 3 4, 5, 6 recoded as 0 • Rejected dotneslic violence (0 -5) Husband atone = 4 Someone = S Decision not made = 6 •^ t -- t-; 4 "^ Ö d 7

Ö Ö Ö 00 I —

, (N in O\ Tl; . -: M ,1; 00 in (-^ 00 I I I o — (N — Cl

iNoooiin „ ,:!ü'"- _J00Tr_J'SOCSg:o00 -^ I < I I o

f-j CT^r ^ r*^ ^ß ^ vo _: O' «O -ü- ^ '^ ^ r-' — ' ' o o

<*1 t- 00 '^ 00 "^^ S! ^ "^ íí^ I I I I o r.j _ • ^ ^ r'^ ^- ^

— O — ^riir-:i I I I lOO ~-' ^ trt —

__ r^ d O r-l ^ v-i oo \b I I I I

K G J >• Q a ai

S 2 a..5-=

62 ^ 00 (N o — ^ i/ TT I I I w fs r-

o^ f*i r*** o ^ d d d 00 o o — — — T —

<~^-1ÎN_;<^or^î^oouSr-^oo i loö

« -^ I I I o o

. n 00 (N r-^ I I I I o

CD ^ < 3 ^ oo r-i _^ (N o (S ""I r I

m >^ oo I I I I I I

I I I o o

lû ^ S «í --. O' — *'^r¡'^ — o — ^iri^ol I I lO lU O O 1* n ^— oB

Ï Ç o

IJ "O y 3 ro "^ Sí '^ 5

63 64 Boateng and Flanagan

(junior & senior secondary) and 2 per- psychological access to health care: a. cent had higher education. Four percent Woman knows where to go to access of women aged 15-49 are in the age health care, b. Woman does not require group 15-19 years, 37 percent are aged permission to access health care. These 20-29 years and 38 percent and 21 per- items were recoded from the original three cent are aged 30-39 years and 40-49 responses of no problem = 0, big problem years respectively. = 1 and small problem = 2 into result vari- ables where 0 = big problem and I = small MEASURES or no problem. These recoded items were Items were selected based on their face then summed to form a scale where 0 = validity and were recoded and summed to big problems with both and 2 = little or no form scales. The following summary of problem with either. constructs presents information about the items, response format, Cronbach's alpha INDEPENDENT VARIABLES (a measure of the scale's internal consis- The independent variables were classi- tency) where applicable and the numbers fied into two categories: demographic and of respondents with valid data. It should be self-determination variables. Demo- noted that all scales were created such that graphic variables included the respon- high scores reflected access and behaviors dent's educational level, age, number of that should promote women's health. children aged five years and below, cur- rent residence in rura! or urban area and DEPENDENT VARIABLES lineage (patrilineai or matrilineal). A No Problem with Physical access to Woman's perception about her husband's health care: Respondents answered yes fertility intention (husband wants more and no to three items measuring access to children) was included as a demographic health care: a. Getting medical help for variable. The need to include this variable self: Little or no problem getting money arose from the fact that, most Ghanaian needed for transport, b. Getting medical women, irrespective of their lineage help for self: Little or no problem regard- would make decisions based on what their ing distance (o health facility, c. Getting husband's position on an issue. This vari- medica! help for self: Little or no problem able enables us to capture the importance having to take transport. Three physical of men's role in decision making affect- access to health care items were recoded ing women's health access and behavior from the original three responses Ihat it such as family planning. was: no problem = 0, big problem = 1 The self-determination indicators were and small problem = 2 into a new vari- variables constructed from a set of items able with small or no problem = 1 and big tapping: a) whether a woman reported problem = 0. These recixled items were that she had a final say in family matters; then summed to form a scale where 0 = b) whether she reported that she had a big problems with all three issues and 3 = final say in purchase decisions; and c) little or no problem with any of the Ihree whether she rejected domestic violence as issues. an acceptable norm of men's behavior No problem with psychological access toward their spouses. to health care: Respondents answered Women's final say in family matters: yes and no to the two items measuring Two "who has the final say in family Ghanaian Women's Access to Health Care 65 matters" items (final say on visits by fam- recoded as 0 and yes responses as I. ily members and final say on food to cook Thus a higher score indicated that the each day) were recoded from the original respondent rejects domestic violence response format of: respondent alone = 1, under multiple circumstances. Don'l respondent and husband/partner = 2, know responses were recoded as miss- respondent and other person = 3, hus- ing. The five items were summed to band/partner alone = 4, someone else = 5 form a scale with values ranging from 0 and decision not made = 6. Original - 5 with higher scores indicating that responses of 4 or 5 (i.e., husband/partner the respondent is more likely to reject alone or someone else makes the deci- domestic violence/abuse. sion) were recoded as 0 and original responses of 1, 2, or 3 (indicating that the woman either makes decision alone or in RESULTS concert with another person) recoded as Zero order correlations between the 1. The original response of "6", i.e., deci- variables are presented in Table 4. As the sion not made, was recoded as missing table shows, age is inversely related to data in the new variable. The two items education and the number of children were then summed to form a scale (final aged five and below a woman has. Educa- say in family matters) with values ranging tion is positively associated with currently from 0-2. living in an urban area, with more auton- Women's final say in purchase deci- omy for women in handling family issues, sions: This variable was based on a similar with purchase decisions in the household, reeoding process of three "who has the and with assertiveness in rejecting domes- final say in purchasing" items (final say on tic violence and abuse. Finally, education how to spend money, on making large is positively associated with matrilineal household purchases, and on making smal! lineage as well as physical and psycho- purchases for daily use). The same collaps- logical access to health eare. Purchase ing of the responses as noted above was decision power is negatively correlated used. The three recoded items were then with family size and family size is posi- summed to form a seale (final say in pur- tively correlated with psychological chase decisions) with values from 0-3. access to health care. Besides having fewer young children, • Women's rejection of domestic vio- older women are more likely to reside in lence: Five items indicating that urban areas, have more purchasing power, respondent rejects domestic violence/ autonomy in family matters, are more abuse under multiple circumstances Hkely to reject domestic violence as a were used as originally coded in the norm, and to have physical and psycho- 2003 GDHS: I. wife rejects heating if logical access to health care. As expected, she goes out without telling husband 2. family size (number of children under six) wife rejects beating if she neglects chil- is negatively correlated with urban resi- dren 3. wife rejects beating if she dence, women's autonomy in family deci- argues with husband 4. wife rejects sions, physical and psychological health beating if she refuses to have sex with access. Tahle 4 shows that men's desire husband 5. wife rejects heating if she for more children is inversely related to hums the food. No responses were his educational level and the number of — p

Si ë u-1 y sa 5 o 1

^- tj- — Q r-i

00 o O o — 1

o t^ c: oo o " 1 r^ o o O o O ~- 1 1 1 1 1

— p — r-1 — p — I

p — r r

•o

2 ye £23 « < "- e ^ Ji s 2 _; -

66 Ghanaian Women's Access to Health Care 67 children under age six. Rural residence with high fertility desires anticipate being favors men's desire for more children. able to stop childbearing when they want Women's autonomy over family matters to, but they do not expect to be able to con- does not diminish men's desire for more tinue if their husbands want to stop. Those children. However, women's autonomy with low fertility desires do not anticipate over purchase decisions in the household being able to stop witbout their husbands' does diminish husband's desire for more consent (Derose & Dodoo, 2002). children. A husband's desire for more Women's expected influence appears lim- children is inversely related to women's ited to situations where their fertility rejection of domestic violence, her matri- desires conform to normative expectations. lineal lineage, and her physical access to These authors note that women's relative health care. A husband's desire for more power in cases of reprtnluctive conflict children is positively related to women's decreased among matrilineal ethnic groups psychological access to health care. over the decade with matriiineal women going from being more likely to use con- The table also shows that women in matri- traception even if their husband did not lineal groups have fewer children under the want to stop chiidbearing to no longer age of six. The fact that family size is posi- holding that sway. Patrilineal women, liveiy related to purchase power may mean however, became more likely to use con- that women with more say in purchasing also have a voice in determining family traception if either partner wanted to stop size. Some studies have shown that women childbearing and men's influence in cases have more say in reproductive decisions of conflict increased more than women's. than tuen. For example, in Peru, women's wages and education influenced family size All of the self-determination indicators more than men's (Schafgans, 1991). Also, are moderately and positively correlated Dodoo (1995) showed that women's inten- with one another. Women who have a tion to stop childbearing was a significant voice in family matters also report more predictor of contraceptive use in Ghatia while men's intentions were not. autonomy in purchasing power. As However, according to DeRose expected, matrilinea! lineage is associated (2003), in recent years men's relative with women having more self-determina- power in determining contraceptive use tion in family and purchase decisions. All has increased, largely due to men's educa- of the self-determination and social support tion. Specifieally, the magnitude of the indicators are higher in matrilineal than in effect of men's education on women's fer- tility decisions increased significantly patrilineal groups and women in matrilineal from 1988 to 1998 (DeRose and Ezeh, groups are more likely to contest domestic 2005), In fact, men's influence grew more violence as an acceptable norm. The self- strongly among matrilineal groups where determination variables are positively cor- women had a fair degree of reproductive related with the dependent variables. autonomy before the onset of rapid fertil- ity decline (DeRose and Ezeh, 2005). MULTIVARtATE ANALYSES Therefore, Ghanaian fertility decline may not be associated with women having Women's SeirDetemunation. In Table greater control over their reproduction but 5, the three multivariate OLS regressions rather asstx^iated more with men's declin- show how demographic factors influence ing fertility desires. women's self-determination. It is clear According to focus groups of young from the first of Ihe three models that Ghanaian men and women, young women women's educational level, age and «r—^ tn-

oo t^ O* r~- 00 r-•"l• S>o ^o 3>o —o

ri ^o o — N o 0 »'I

UJ 'z

O\NO OON

T* o — ^ M —"O• r~ —- Qo (o^ —ON

— r~ IO rj r-l o

68 Ghanaian Women's Access to Health Care 69 matrilineal lineage are important determi- who are under six years are positive pre- nants of whether she has a say in purchase dictors, although this sel of variables decisions in ber household. The older and explains just under 4 percent of the vari- more educated the woman is, and having ance. The two dependent variables - physi- a matrilineal lineage, the more likely she cal and psychological access to health would have purchase power at home. care are dichotomous variables and thus These demograpbic factors explain nearly logistic regressions were run. 13 percent of the variance in women having a say in purchase decisions. Women's Physical Access to Health Similarly, a women's assertive attitudes Care. In Table 6, tbe baseline model about rejecting domestic violence and abuse examines the impact of women's demo- is very mucb influenced by ber educational graphic characteristics including lineage level, age, current urban residence, and on Iheir physical access to health care. matrilineal lineage. The higher ber educa- Table 6 shows the positive effects of edu- tion level, tbe older sbe is, residing in a city, cation, urban residence malrilineal and and having a matrilineal lineage, tbe more lineage on women's physical access to assertive she would likely be about rejecting bealtb care. A woman's education level in domestic violence and abuse. This set of Gbana very much detennines her occupa- variables explains 12 percent of the variance tion and socio-economic status. Therefore in women's attitudes towards domestic vio- women with higher educational levels lence as an acceptable norm. will most likely have access to health care Wilh respect lo women's autonomy over through the leverage they have over income, family matters, a woman's educational transportation, and residential proximity lo level and the number of children sbe bas health care facilities. Similarly, matrilineal

TABLE 6 ODDS RATIOS ANH ERRORS i-t« LOOETHC REGWXSIONS pRrnir ti PirvsitAi. Actrj-s FOR GHANAIAN WOMEN

Mosm 1 MODEL 2

IJtMiXmAPKÏ (B) Odds Ratio S.E. IBI Odds Raiio S.K.

Conslant -1.05 .351*** .238 -1.26 .283*** .247 Education level (0-1) .835 2.4!*** .107 .775 2.17*** .110 Age (!/R) .007 1.01 .006 .006 1.01 .006 Number of children aged 5/beiow (l/k) .026 1.03 .065 .036 1.04 .065 Currenl residence (Rural-Urban 0-1) 1.528 4.61*** .121 1.4% 4.462*** .122 Lineage (patrilineal/mntrilineal) (0-1) .285 1.33** .110 .235 1.264* .114 Self Detemiinaiion Final say. family maiters (0-2) .012 1.013 .060 Final say. purchase det:ision(0-3) .004 LOI .027 Rejects domeslic violence (0-5) .091 1.095*** .247 Model Summaries N 2133 2133 Improvement X^ (-2 Log likelihood) 2564.02 2547.01 Degree of freedom 1 1 Nagelkerke R^ .221 .227 Significance .000 .000

Signiflcante levels: •"p

TABLE 7 Ot)DS RATIOS AND STANDARD ERRORS FOR LoGtsTic REGRESSIONS PREDICITNG PSYCHOLOGICAL ACCESS

MODEL 1 MüiMü.2

[JEMOORAPHV (B) Odds Raiio S.E. Odds Ratio S,F,.

Constant 1.009 2.74*** .302 .844 2.33*** .311 Hducation level (0-1) .132 1.14 .145 .109 1.12 .015 Age (I/R) .013 1.02* .008 .013 1,01 .008 Number of children aged 5/below p/n, .124 1.13 .167 .134 1.14 .085 Current residence (Rural- Urban) .698 2.01*** .147 .679 1.97*** ,153 Lineage (path 1 ineal/matrilineal) (0-1) -.032 0.97 .302 -.056 .95 .101 Self Determination Final say, family matters (0-2) .192 1.21* .080 Final say, purchase (lecision(0-3) .190 1.27*' .035 Rejects domestic violence (0-5) .094 1.10** .311 Model Sununaries N 2133 2133 Improvement X^ (-2 Log likelihood) 1716.72 1699.47 Degree of freedom 1 1 Nagelkerke R^ ,022 .033 Significance .000 .000

Significance levels: -"p < 0.01)1; "p < O.ÜI: •?< 0.05: 'p < 0,1. Note: Values in brackets indicate how respective independent variables were cixled.

care because a woman who has control education on her and her family's health over purchase decisions would not need (Akin, 2œ5; Gisselman, 2006; Onah, permission from her spouse to seek medi- Ikeoko. and lloabachie, 2005). There may cal help. In summary, one might expect be different reasons underlying the effects women with greater autonomy in manag- of education on different outcomes. In the ing money and having assertive attitudes case of physical access, education may about rejecting domestic abuse to have simply reflect the fact that educated better psychological health access than women have better incomes which allows those who do not. them to afford access to heallh care or transportation. Like education, urban living provided consistent positive results DISCUSSION AND CONCLUSION for physical and psychological access. There was a consistent educational The positive effects on women's health of effect on physical access to health care. urban living are consistent with other Contrary to our belief, there was no sig- work (Coast, 2006). nificant effect of women's education on Unlike education and urban living, psychological access to health care. This there were mixed results for matriiiny. It is surprising insofar as education should showed consistent positive effects on enhance women's sense that they have a physical access and responsible sexual right to health care. With the exception of behavior. Women in matrilineal groups psychological access, the results arc con- control their own money and no husband sistent with a large body of scholarship has any right to dictate how a woman showing the positive effects of women's should spend her earnings. It could be that 72 Boateng and Flanagan women's control of wealth makes it possible hul". Ghanaian society assigns different for them to access health care. This inter- roles to men and women, and role expec- pretation is consistent with previous tations are reflected in tbese outcomes. research indicating that women in matri- As expected, matriliny predicted lineal groups have greater autonomy in participation in purchase decisions and fertility decisions (Dodoo, 1995; DeRose, rejection of domestic violence. If a 2003), although recent work suggests that woman controls money and has absolute husbands' decisions are also contributing authority over decisions aboul bow to to lower fertility (DeRose and Ezeh, spend ber earnings, she should be in a bet- 2005). The study consistently shows the ter position to assert ber will and reject importance of education in strengthening abuse from a partner who may be finan- women's say in purchase decisions, rejec- cially dependent on ber. tion of domestic abuse, and say in family The study provided partial support for decisions. This observation is in line with mediation in some cases for education findings from previous studies showing tbat and in other cases for urban effects on education enables women to be empow- pbysical and psychological access to ered and enhances their opportunities to improve their healtb (Tawiah, 1984; healtb care. The effects of education on Sandilord, Monlegegro and Sanchez, 1995). pbysical health were reduced with the Whereas urban women were more introduction of self determination and likely lo reject domestic violence, tbey social supports. Similarly, the effects of were no more likely to report that they urban living on psycbologicai access were had a say in family or purchase decisions. reduced when self determination was Tbis was quite unexpected as one would introduced. There also was evidence that think tbat living in an urban area would be the self determination variables added related to greater autonomy and thus explanatory power in tbe models. With increased leverage in purcbase decisions respect to physical access, the self deter- and in other family matters. However, mination variables explained more of the since the items lap family decisions con- variance in the outcomes than did the cerning visits by family and what food is demographic variables alone. cooked, the lack of urban effects may be less surprising. Urban women do not ACKNOWLEDGEMENTS report more autonomy in purchasing This paper is based on the PhU thesis of ihe power. This shows that urban living does first author. The authors wish lo thank the edilor not make the woman tbe bead of the fam- and anonymous reviewers for Iheir constructive ily. This observation is in line with a Gha- reviews. We also thank tbe iollowing people for naian proverb Ibat says that "when a their guidance: Drs. Les Gallay, Tena St. Pierre, woman buys a gun, it is stored in a man's Tasha Snyder, Edgar Yoder and Dr Francis Dodoo.

REFERENCES

AHOTCMIE, C. 1997. Social control in traditional Molhcrhood Consultation in Sri Lanka, 18-23 southern Kweland of Ghana: Relevance for mod- October, em crime prevention. Accra: Ghana Universities AGyF.K»NG. I. A. 1999. Relonning health service Press. delivery at district level in Ghana: the perspec- AB()II7AHR. C. t997. Improved Access to Quality tive of a di.striüt medical officer. Health Policy Maternal Heaiit» Services. Presentation at Safe and Planning; 14(1): 59-69. Ghanaian Women's Access to Health Care 73

AKJN, M, S, 2(H)5, Education and fenility: A panel GETHtNCi P, W,. A, M. NtHIR. D, ZllROVAC. P, M, daia analysis for Middle Hastem countries. The AixtNSoN. and S. I, HAY. 2004. Empirical model- Journal of Developing Areas, 39(1). 56-69, ing of government health service use by children ARCURY. T, A,, R, A, BUL. B, M, SNIVELY. S, L, with fevers in Kenya, Acta Trópica 91:227-237 SMITH. A. H, SKEI,LY. and L, K, WrTwoRB, 2006, GHANA SiA-nsiUAi, ShRVich (GSS). NOCHKHI M^Mü- Complemcniary and alternative medicine use as RiAi. iNsntiiiT^: FOB Mttm AI RESRARCH health self-management: Rural older adults with (NMIMR). and ORC MACRO, 2004, Ghana diabeies. Journal of Geronlology: Social Sci- Detnographic and Health Survey 2003. Calver- ences, 61B, S62-S70, ton, Maryland: GSS. NMIMR, and ORC Macro, ARHrN-TwK(jRANti, D, 2001, for GHANA STAILSIICAI, SHRVKT. 2000, The Ghana Liv- the informal sector in Africa: Design features, ing Standards Survey. Round Four (GLSS 4) risk protection, and resource mobilization, CMti 1998/99: Data User's Guide, Unptihlished, Working Paper Series, Commission on Macro- GLSSKLMANN. M, D, 2006. The influence of maternal economics and Health, (Paper No, WG.Í:I), childhood and adulthood social cla.ss on the Camhridge. US, A. health of the infant. Social Science and BENFK), K. D, 2006. The community-level effects of Medicine. 63,1023-1033, women's education on reproductive behavior in LiNDSTRDM P, D,. and E. MUNOZ-FRANCO. 2006. rural Ghana. Demographic Research, Max Migration and maternal health services in rural Planck Institute for Demographic Research, Guatamala. Social Science and Medicine. Rostock. Germany. 14(20). pages 485-508. 63.706-721, BixAii Nil. J, K. 2006, Women's Health Access and MICHAELS D,. C, SCHUMANN, and H, KLIAVKOV, 2001, Behaviors: Effects of Urban Residence. Education. BUV Strategic Marketing Plan for Ghana, Self-Determination, and Social Supports in Africa, BUV Project. 17pp. Ghana, PhD dissertation, The Pennsylvania ONAH H, E,. L, C. IKFOKO. and G.C. lt,oABACHiE, Stale University, University Park Campus, State 2006. Factors associated with the use of mater- College. nity services in Enugu. southeastern Nigeria. BOUR. D, 2004, Water needs and women's health in Social Science and Medicine, 63.1870-1878, the Kumasi metropolitan area. Ghana, Health POTTS. M, 1997. Sex and birth rate: Human biology, and Place. 10:85-103, demographic change, and access to fertility reg- CHAO. S, 1999, Ghana: Women's role in improved ulation methods. Population and Development economic performance. Findings. No. 145, Review, 23(1) 1-39. CoASi. E. 2006, Local understandings of and SANDIFURI). P J,. C, M, MONTENEGRO, and G. SANCHEZ. responses to HIV: Rural-urban migrants in 1995, The itnpact of wotnen's literacy on child Tanzania. Social Science and Medicine. 63(4) health and its interaction with access to health ser- 1000-1010, vices. Population Studies. 49(1). 5-17, D'AMURÜOSO, L, 2006, Midwives altitudes to women ScHAFOANS. M. 1991, Fertility determinants in Peni: in labor in Ghana, IMMPACT id2l Research A quantity-quality analysis. In B, K. Herz and Highlight. University of Aberdeen. UK. S. R. Khandker (Hds,). Women's work, educa- . F, 2003, Marriage, lineage, relative spotisal lion anil family welfare in Peru (pp. J 71-203). power and fertility decline in Ghana, University World Bank discussion papers 116. Washington. of Maryland with funding from Maryland Popu- DC: The World Bank, lation Research Center and the National Insti- TAWIAH, E, O, 1984, Determinanis of Cumulative Fer- tutes of Health and Human Development. tility in Ghana. Demography 21(1) the Evidence, E. L. F.. and A, C, BZEH. 2005. Men's influ- Health Transilion Review. 3(2). pp 159-175. ence on the onset and progress of fertility TsoKA. J, M,. and D, LE SOEUR, 2004, Using GIS to decline in Ghana, 1988-98. Population Studies, measure geographical accessibility to primary 59(2)197-210. health care in rural South Africa, South Africa DRROSE. L. F., and F, N. A, DoDOu. 2002. Fenility Journal of Science. 100: 329-330, desires and perceptions of power in reproductive VAN REN BOOM, G. J. M,; N, N. N, NSOWAH-N|[AMAH. conflict in Ghana, Gender and Society. 16(1) and G,B, OVERHOSCH. (undated). Curative 53-73, health care utilization in Ghana: A multinomial DotxK). F. N. A.I995. Explaining Contraceptive Use analysis of equitable access opportunities. Differences: Do Men Play a Role? African Pop- Human Resource Development for Poverty ulation Studies/Etude de la Population Africaine Reduction and Household Food Security of the 10(1), Foundation Sécurité Alimentaire Durahle en FRANKS. P.. and K, FtsrEti.A, 2002. Effect of patient Afrique de L'C^uest Central (SADAOC. Quaga- socioeconomic status on physician profiles for dougou. Burkina Faso. JEL Classification: ILL; prevention, management, and diagnostic H5I;O55, testing costs, A/crf, Care 40(8):717-24 WHU STATISTICAL INFORMATION SY,STIZM (WHOSIS). FURUTA, M,. and S, SALWAY. 2006, Women's posi- 2007, World Health Statistics. Core Health tion within the household as a determinant of Indicators, Third Edition, Retrieved on 12/2/ maternal health care use in Nepal, International 2007 from: http://www.who,int/whosis/data- family planning perspectives 32:(I) 17-27, base/core_select_process. cfm