{IHIIHIUHHWI

$
}

  

WI

1

31% ‘IHHIHHIWHIHWWII!

 

 

LIBRARY

M'Chlgan 5!: ate
University

 

 

 

This is to certify that the

thesis entitled

‘ ‘ Viki 1:32} P‘i/j ix, W
L (,3 (31*;

- W?
{E'Q’L‘LLL ( (13.0,, E) 7" m 7.1!

eresented by

Fwd/ii LU )K/igl/w Ff: 0.1L MM i A73

,VWP/

has been accepted towards fulfillment
of the requirements for

[44 ° H .5 degree in MQELB

 

I‘d/Z 2], \p; (172.71%? .-”kr-1°'t/Z//

A (Major professor

{I . 4 I I.
DateAéZdCéC/fi (1?,5‘4) // 5/?

 

0-7639 MS U is an Affirmative Action/Equal Opportunity Institution

 

 

 

)V1531,1 RETURNING MATERIALS:

Piace in book drop to
LJBRARJES remove this checkout from.
n your record. FINES will

 

 

- be charged if book is
returned after the date
stamped below.

 

 

 

 

 

THE IMPACT OF'POLYGYNY ON FERTILITY:

The Case of Somalia

By

Abdulkadir Mohamed Elmi

THESIS

Submitted to
Michigan State University
in partial Fulfilment of the requirements
For the degree of

MASTER OF ARTS

Department of Sociology

1988

ii WEN/l5

Abstract

This study focusses on the impact of polygyny on
Fertility. with reference to urban Somalia.

An attempt is made to investigate the impact of this
form of marital condition on fertility. Data collected for
a Family Health Survey in 1983 are used in this analysis.

in this study we have used Frequency tabulations.
"difference of means" tests. " difference of proportions"
tests. and analysis of variance.

Among younger wives (below age 30). there were no
significant differences ‘in fertility between those in
monogamous and polygynous unions. regardless of the number
of times the wives had been married. I

For women aged 30-49. poiygynously married women had a
significantly higher cumulative number of lifetime births
than did monogamousiy married women. if both groups had
been married only once but a significantly lower cumulative
number of lifetime births . if both groups had been married
more than once.

The implications are discussed.

Acknowledgments

The author wishes to express his sincere
appreciation to his advisor. Dr. Nan Johnson. for her kind
guidance. advice and support throughout his study at this
school. without her help this thesis would not have been
completed. His thanks. too. go to Dr. Allan Beagle and Dr.
Dan Clay for their helpful comments.

His sincere gratitude also goes to Ms. Anne Cross
for permitting him to use the data from the 1983 Somali
Family Health Survey.

He is also thankful to the' Central Statistical
Department. Somali Ministry of National Planning which
made Iit possible for him to study at Michigan State
University.

Finally. the author would like to express his
appreciation to his wife. Halima All. for her patience.
understanding and help. it is to her that he dedicates this

thesis.

Abstact
Acknowledgments
Table of Contents
introduction
Literature Review
Methodology
Findings
Conclusions

Bibliography

TABLE OF CONTENTS

Page

ii

IO
15
28

31

Introduction

Polygyny is a type of marriage in which a man has more
than one wife. It has been a relatively common form of
marriage in the second half of the twentieth century among
populations in Africa (Chanacka. 1980).

Polygyny is generally studied by anthropologists rather
than by sociologists. Thus. nearly all studies of African
polygyny have treated it as a traditional form of marriage
inevitably destined to disappear under the pressures of
urbanization. wage employment. Christianity and general
modernization (Ware.i979).

There are at least three advantages of polygyny to men.
The first one is economic: plural marriage means more
children per husband. and this increases the economic
production of the family which he heads. Secondly. a
polygynous man commands a higher status and a wider sphere
of influence in the community where he lives. And thirdly.
it is believed that a polygynous man enhances his chances
of having more sons (Namfua 1984).

Ware (1979) mentioned that in a society in which
polygyny is widely accepted, nearly all monogamous
marriages are perceived as potentially polygynous. and solo
wives take the prospect of additional wives into

account. Ware argued that polygyny promotes competition in

childbearing among cowives; thus. the greater the number
of cowives. the greater the number of children that are
likely to be desired by each wife. In some cases. wives in
polygynous marriages may have greater autonomy because they
have less invested in the marriage and because. in losing
part of their huspands’ economic and moral support. they
also gain independence. The luckiest woman is the senior
wife. who receives respect both from her husband and
from the society in general. who is in command of her
Juniors. who does less work. and whose children usually
receive preferential treatment (Here 1979). In addition. in
African societies where survival depends upon agricultural
prOduction. where such production is labor-intensive. and
where SUCh labor is usually supplied by women. polygyny can
provide an economy of scale in farming. In these many
ways. polygyny serves the interests of both men and women
and perhaps even of capitalists.

While polygyny increases the number of children that
would otherwise be born per husband. its effect on the
fertility of wives is equivocal. Some researches have
concluded that polygyny suppresses fertility; others. that
polygyny enhances fertility; and others. that polygyny has
no effect on fertility. The followingsection reviews the
literature on this matter in search of a resolution to

these inconsistencies.

Literature Review

Much controversy has surrounded the possible effect of
polygyny on the fertility of women . Muhsam (1956) found
that among the Arab Bedouin in Israel. the number of
children born to polygynous wives was smaller than to
monogamous wives. Similarly. lower fertility has been
reported for polygynous than for monogamous wives in the
World Fertility Surveys of Ghana and Kenya and in a census
of nine villages in the Upper Bambara Chiefdom. Eastern
Province. of Sierra Leone (International Statistical
Institute 19838; Isaac. 1980; Mosley. Warner. and Becker.
1982). The frequency with which lower fertility has been
reported in polygynous unions has led to the hypothesis
that polygyny causes lower fertility. Causal mechanisms
freduehtly cited are: a lower frequency of sexual
intercourse. a higher divorce rate. and a closer observance
of taboos against sexual intercourse during lactation for
polygynous than for monogamous wives (DorJahn. 1958:
Isaac. 1980: Lorimer. 1954; Olusanya. 1971).

Evidence favoring an antinatalist effect of polygyny on
fertility is not ironclad. however. Culwick and Culwick
(1939) found no appreciable differences in the number of
children born to the wives of monogamous and bigamous

husbands in Tanganyika (how Tanzania). They concluded that

the rate of coital activity per wife was not greatly lower
for bigamists. In addition. the World Fertility Survey of
Cameroon found no significant difference in fertility
levels between the wives of polygynists and wives of
monogamists (International Statistical Institute. 1983b).
Unfortunately, the WFS study in Cameroon did not compare
the fertility of first wives in polygynous unions with that
of junior wives in polygynous unions nor with the fertility
of wives in monogamy.

Such comparisons proved important in a study of
nineteenth-century Mormons in Utah (Bean and Mineau. 1986).
The average number of live births per wife was much greater
for first wives than for later wives in polygynous unions
and mucn greater for first wives in polygynous unions than
for wives in monogamous unions. These findings for the
polygynous .American wives were particularly impressive
because. unlike their African counterparts. they did not
have high rates of illiteracy which would impair the
completeness of reported numbers of children ever born
alive to them. The lower fertility of the later wives
compared to the first wife in polygynous unions was partly
but not wholly explicable by the older ages at marriage of
the later wives and their bridegrooms and by the shorter
durations of sucn unions before widowhood. However. the

much higher fertility of the first wives in polygynous

unions as compared to the other two groups of wives
suggested that polygyny could have a positive effect on the
fertility of certain wives.

Wife rank was also important among the Mehde of
Sierra Leone (Isaac. 1980). A polygynous wife averaged 0.7
fewer children than did her monogamous counterpart. but
this relationsnip was affected by the number of her
previous marriages. Among once-married women. wives of rank
one in polygynous unions ("big wives") had a mean of 1.28
more live births than did their counterparts in monogamous
unions and had higher fertility than did once-married
cowives of rank two or more. Isaac (1980:311) reasoned that
these senior-most wives in polygynous unions may have
performed less strenuous work and consumed better diets due
to the presence of subordinate cowives. and thus may have
gained a reprOductive advantage over other once-married
wives in monogamous and polygynous unions. However. twice-
married wives had mucn higher fertility if they were in
monogamous than in polygynous unions (4.2 vs 3.1 mean live
births per fertile wife. respectively). Isaac (1980: 303)
proposed that a divorced or widowed woman of proven high
fertility was propably more likely to marry a monogamous
second husband. especially if he could not afford the
brideprice of a virgin. could not afford to become a

polygynist. and did not wish to risk marrying a sterile

woman. In contrast. a divorced or widowed woman with few or
no children would be quite likely to remarry as a junior
wife in a polygynous union. particularly when her second
huspand already had children by other cowives (Isaac. 1980:
304). Such a man would be less concerned about siring
children and would welcome the additional labor of a new
cowife. Consequently. this study emphasized the importance
of controlling the effects of wife's rank in a polygynous
union and the number of her previous marriages on the
polygyny-fertility relationship.

The effect of polygyny on fertility seems to differ not
only from one study to another in different countries but
from one stUdy to another in the same country. Olusanya
(1971) found in his study of Western Nigeria that wives in
polygynous unions had slightly more children than did wives
in monogamous unions. but he claimed this was largely a
function of the older ages and longer durations of marriage
for the former group. On the other hand. Ukaegbu (1977)
found in his study of Eastern Nigeria that wives in
polygynous unions had significantly fewer children on
average than did those in monogamous unions. He interpreted
this relationsnlp as resulting from the wider age
difference between husbands and wives in polygynous
marriages. Chojnacka’s (1980) study of six Nigerian

villages and towns snowed no systematic difference in

fertility between the two types of marriages; and what
differences existed were attributed to the higher
incidence of Childlessness in polygynous unions. A possible
reason for the disparate results among these three Nigerian
stUdies is the focus on different ethnic groups. Olusanya
studied the Yoruba of Western Nigeria. while Ukaegbu
analyzed the 100 of Eastern Nigeria. Chanacka’s data
covered a mixture of tribes in Eastern. Mid-Western. and
Western Nigeria. Because some ethnic groups are more
resistant to westernization and Christianity than other
ethnic groups. it is important to control the impact of
ethnicity on the polygyny-fertility relationship. In
addition. a common problem is that all three studies in
Nigeria assumed that all births to a given wife occurred in
her current union. Because divorce is somewhat common in
Nigeria. some births to a woman in a polygynous union may
have occurred in a previous marriage. In addition. divorce
or widowhooa .causes a fecund woman who remains celibate
between unions to lose reproductive time. If previously
married women are more likely to remarry as Junior wives in
a polygynous union. this would make it appear that polygyny
had depressed their fertility even when that depression may
have occurred prior to the initiation of the polygynous
union. Therefore. the impact of marital history (number of

previous marriages) should be controlled when comparing the

fertility of monogamous and polygynous wives.

The purpose of the present study is to reexamine the
polygyny-fertility hypothesis with recent survey data from
Somalia. While a few previous studies (e.g.. Lewis. 1962)
have examined nuptiality in the northern region of the
country. no recent study has been undertaken to compare the
relationship between fertility and form of marriage.
Because Somalia is almost entirely Muslim and because Islam
allows a husband to have up to four wives simultaneously
(if he can treat them all equally). polygyny has been
practiced in Somalia for a long time.

In Somalia. there are great psychological and economic
advantages to having children. The birth of a child is a
time for great reJoicing: and if it is a male child. his
parents receive special congratulations. A sheep or goat is
sacrificed to celebrate each birth. A large number of
children contributes in a very real way to the social power
and prestige of the family. and polygyny is a way for a man
to sire more Children than he could get with one wife only.

Children have an economic value to Somali parents.
Among nomads (who comprise 60% of the population). having
ten or more children is daily talk. The birth of each child
(especially if it is a boy) brings an extra pair of hands
to care for livestock and to fight outsiders. Children have

an economic value in the city. too. Since there is no old-

age pension scheme in Somalia. children are expected to
give financial support to their parents in old age.

While pronatalism should be strong in both monogamous
and polygynous unions. it shoud be stronger in the latter.
The economic value of children to their parents in Somalia
has already been discussed: and. indeed. polygyny is seen
as a way of strengthening the household economy through
providing for more women laborers and child laborers.
Polygyny provides economic advantages not only to husbands
but also to cowives; however. the cowife of rank order one
benefits disproportionately. As is the case of Sierra Leone
(isaac. 1980). the first wife ("big wife") in a polygynous
Somali union commands the labor and the respect of her
cowives and thus works less hard than they. Her privileged
status may have physiological consequences which will
enable her-to bear more children than her cowives. As such.
the fertility of cowives within polygynous unions should
vary inversely with the order in which each of the cowives
was married by their common husband. This inverse
relationship should also be abetted by an increasing age of
the polygynous husband at each of his subsequent marriages
and by a possible decreasing rate of coital activity per
cowife. In other words. if polygyny does have a positive
effect on the fertility of any cowife. this effect is most

likely to be manifest in the reproductive performance of

10

the first cowife. who would be found to have higher
fertility than her counterpart in a monogamous union.
Whether or not the fertility of cowives of rank order two
or more is also higher than that of wives of monogamous
unions is a moot question. Its answer depends on the
strength of the inverse effect of wife rank on fertility.
We reason that the conflicting results from previous
studies about the effect of polygyny on fertility stem from
a failure to control the impact of wife rank on fertility.
Unfortunately. our data set did not contain wife rank:
however. it did indicate the wife’s number of previous
marriages. Since most Somali husbands want their first wife
to be a virginal bride. the first wife is less likely to
have been previously wedded and less likely to be
subsequently divorced. On the other hand. cowives of rank
orders above one in a polygynous union are less likely than
the first wife to be virginal brides and are more likely
than she to have been previously wedded and to be
subsequently divorced. The purpose of the present
investigation is to explore how polygyny is related to the
fertility of Somali women and how that relationship varies

by their number of previous marriages.
METHODOLOGY

The data for this study are drawn from the Somali

ll

Family Health Survey (SFHS). This survey was conducted
between June and November. 1983 by the Family Health Unit
of the Ministry of Health. A total of 3016 ever married
women were interviewed in Mogadishu. Hargeisa. Kismaio.
Baidoa and Burao. These cities were selected because they
were the focus of a recently launched family health/family
planning project. A two stage sample was used. with
probability - proportional - to - size (PPS) selection at
the first stage. Areas known as WAAXDAs. ( medium sized
urban administrative units). formed the primary sampling
units (PSUs).

" Prior to this first stage selection. several
decisions were made. First. the target sample size in terms
of eligible women (ever married women aged 15-49) was
determined for eacn city. These targets ranged from 800 to
950 respondents. Next. the number of households necessary
to preduce this target number of eligible women was
calculated using an average of 0.84 ever married women
aged 15-49 per household. a figure which was derived from
the 1980 Demographic Survey in Mogadishu. Bay and Lower
Shabelle. Finally. an average cluster size of 40 household
was used as the ultimate cluster sampling unit " (Ministry
of Health. 1985: 6).

" The frame used for selection of the PSUs consisted of

a list of WAAXDA along with a count of the households in

12

each. Because the PSUs were too large to interview
completely. eacn selected WAAXDA was subdivided into
roughly equal sized segments according to its measure of
size (l.e..the number of households divided by the cluster
size of 40). The segment operation was conducted by a team
of mappers who first located the boundaries of the WAAXDA.
divided it into many small chunks using recognizable
boundaries. made a quick count of the number of households
living in eacn Chunk. grouped chunks into the required
number of equal sized segments. selected one segment at
random and finally. sketched a map of the selected
segment. Thus the second stage sampling unit was a segment
or a cluster of approximately 40 households " (Ministry of
Health. 1985: 6).

In Mogadishu. the second stage sampling unit was the
TABELLA. a subdivision of the WAAXDA . Because it was
possible to get household counts for TABELLA from local
officials. the amount of segmenting and mapping was greatly
reduced . However. in about one-half of the PSUs in
Mogadishu. the TABELLA was still too large to serve as the
final cluster. A third stage . in which the TABELLA was
segmented. had to be introduced.

In all five cities.a household listing operation was
conducted in eacn selected segment immediately prior to

interviewing. in order to make the listing as current as

13

possible. However. the final sample size of eligible women
was substantially smaller than expected; instead of the
intended 4250 reapondents. the survey only covered 3016.
This discrepancy was due to a much lower average number of
eligible women per household than expected.The SFHS has
found that the eligible women per household was between
0.63 and 0.69 instead of 0.84 (this one was derived from
data for MogadiShu from the 1980 Demographic Survey ). It
was unclear. as to why this discrepancy occurred. One
theory was that the interviewers may have intentionally
pushed eligible women out of range. by estimating their
ages to be lower or higher than the cutoff of 15 and 49. or
by recording ever married women as single. so as to avoid
having to conduct the longer individual interview. Another
possibility was that younger ever married women were for
some reason systematically undercounted in the household
schedule. either because they were less likely to be at
home or because of some misclassification. such as divorced
women reporting themselves as never married. The response
rates were high. both at the household and individual
respondent level. In both cases. almost all rates were in
the nineties. indicating that interviewers made sufficient
call-back to reaCh selected respondents. The interviewers
of the survey were 23 young women. mostly nurses with the

Ministry of Health. who were recruited from the cities in

14

which they were to work.

All the information was collected by means of a
questionnaire. The questionnaire used in this survey was
based on the medal questionnaire developed by the
Contraceptive Prevalence Survey project (CPS). Unlike the
standard CPS practice of listing only eligible women in
each selected household. the SFHS utilized a very brief
household scnedule in which all members were .listed by
name. sex. and age. and those over age 12 were asked if
they had ever been married. Ever married women aged 15-49
were identified for an individual interview.

The individual questionnaire had five sections: 1)
background Characteristics of the respondents.such as
education. work status. and marital status: 2) fertility.
including questions on number of Children ever born. date
of last birth and other demographic questions 3 3)
fertility regulation: 4) availability: 5) sources of
information about family planning; 6) maternal and Child
health; 7) husband’s background and housing.

The aim of this study is to examine the relationship
between fertility and polygyny . The dependent variable to
be employed in this study is the average number of children
ever born to women (CEB). All respondents were asked a
series of questions about the number of children they had

given birth to in their lives. Those who ever had a live

15

birth (even if the baby lived for just a short time) were
asked about the number of male and female children who were
currently living with them. those who were living somewhere
else. and those who died. These numbers were summed by the
interviewer to find the total number of live births. Then
the interviewer asked: "Altogether. you have had [the
total] live births? There may have been a tendency for the
wives to underreport their live births. since the majority
of the wives were illiterate. Nevertheless. this
questionnaire was designed to minimize the degree of
underreporting live births.

The test variable in this study was type of current
marital union. Currently married women were asked whether
they had cowives or not. Those who had cowives were
considered to be in polygynous unions: otherwise. they were
considered to be in monogamous unions.

Because this question was not asked about previous
unions. we have necessarily excluded never married.
widowed. or divorced women from the present investigation
(N=354). 0f the 2662 women remaining in our analysis. 452

(= 16.98%) were in polygynous marriages (see Table 1).

Findings

In this multivariate analysis. we examine the

relationShip between polygyny and fertility after

I6

controlling a number of other relevant factors. One such
factor is age. Exact timekeeping has no great cultural
meaning in Somalia. since everyone is going to die anyway.
Therefore. many women were unable to state their exact age
in completed years of life. In such cases. the interviewers
were instructed to guess the exact age as Closely as
possible. The interviewer who felt that the respondent did
not know her age or did not tell
the truth about her age was instructed to ask more
questions. The interviewer asked the respondent's age at
first conabitation and the age of the eldest child. if
relevant. For example. if a woman said that she was 18
years at first conabitation. and if her eldest child was
reported as 8 years. then the interviewer would assign her
an age of 27 years: the age at first cohabitation plus the‘
age of the eldest child plus one year of pregnancy.

Table 1 shows that the Somali urban wives in polygynous
unions (17% of the total sample) displayed significant
demographic differences from their counterparts in
monogamous unions. For example. the average age of the wife
at first marital conabitation was earlier for wives in
polygynous unions than for those in monogamous unions.
Among women under age 30. the average ages at first marital
cohabitation were 16.75 and 17.21 years for wives in

polygynous and monogamous unions. respectively (Table I).

17

Table 1. Characteristics os Soaall Urban Hives by Age and Type
of Harriage.

 

Descriptive Variable

Age at Interviev =15-29 yrs

Age at intervlev =30-49 yrs

 

Polygynous lonogalpus t Polygynous lonogalous t
Hives lives lives Hives
Average nunber of
Children ever born 2.35 2.18 1.13 5.61 5.65 -.11
Average age of
Uife at interviev 23.68 yrs 22.81 yrs 3.35"' 36.30 yrs 34.91 yrs 4.01"'
Average age of vife
at first cohabitation 16.15 yrs 11.21 yrs -2.39" 11.30 yrs 11.81 yrs -2.21'
Percent carried lore
than once 30.4 1 15.3 1. 5.21m 49.2 1 25.3 1 1.469'"
Percent childless 19.6 1 20.80 1 -.315 1.4 1 4.6 1 1.866'
Percent vlves illiterate 69.1 1 62.2 1 1.816' 83.1 1 83.6 1 .038
Percent vives eaployed 11.0 1 14.3 1 .964 23.3 1 18.4 1 1.815'
Percent vlth husbands present 85.1 1 19.4 1 1.839' 16.1 1 16.3 1 .138
Percent ever using
contraception 0.5 1 1.3 1 -1.000 0.8 1 2.2 1 -1.56
I 194 1156 258 1054

 

' p< 0.05. one-tailed test
" p( 0.01. one-tailed test
"' p( 0.001. one-tailed test

18

To ascertain whether this relationship was statistically
significant. we employed a "difference of means" test (see
BlaIOCK, 1972:220-228). This required us to assume that the
wives in polygynous and monogamous unions represented two
independent random samples (N = 194 and 1156 respectively)
whose mean ages at first cohabitation were equal. Separate
computations of the test statistic (t statistic) can be
made depending on whether the variances for the two groups
are assumed to be equal or unequal. Blalock (1972: 227)
wrote that when the sample sizes are very different (as is
the case in our study) it becomes more difficult to test
the model assuming inequality of variances because the

proper number of degrees of freedom must be approximated.
For this reason. we preferred to assume that the variances
in age at first conabitation for women in polygynous and
monogamous unions were equal. The t statistic computed
under this assumption. -2.39 (see column 3. table 1) was
statistically significant for alpha less than .01 (one-
tailed test). Although wives at ages 30-49 had begun
marital cohabitation at somewhat older ages than had wives
aged 15-29. earlier conabitation by the ones in polygynous
unions was again opserved and was statistically significant
(t = -2.27. alpha less than 0.05 for the one-tailed test)
Since earlier conabitation is associated with higher

fertility. the impact of age at first cohabitation would

l9

favor higher fertility among women in polygynous marriages.

On the other hand. a younger average age at first
cohabitation in Somalia is associated with a higher rate of
marital dissolution. The percentage of wives aged 15-29
who had been married more than once was 30.4% for those in
polygynous unions but only 15.3% for those in monogamous
unions. in order to examine whether these proportions were
significantly different from each other. we employed a
"difference in proportions" test (Blalock. 1972: 228-230).
The t statistic of 5.21 was statistically significant for
alpha less than 0.001 (Table 1). Likewise. for wives aged
30-49. the proportion of those who had been married more
than once was 49.2% in polygynous unions but only 25.3% in
monogamous unions. a difference that was statistically
significant (t = 7.469. alpha less than 0.001). Marital
disruption can reduce the length of time that a woman faces
the risk of pregnancy and can thereby have a permanent
dampening effect on her lifetime fertility. This dampening
effect would.cet§ris paribus, tend to reduce the fertility
of wives in polygynous unions below that of wives in
monogamous unions. As such. the effect of younger age at
first marital conabitation in raising the fertility of
wives in polygynous unions is tempered by subjecting these
same wives to greater risks of marital dissolution.

The difference between wives in polygynous and

20

monogamous unions in regards to illiteracy. employment
status. and the physical presence of husbands depended on
the wives' age group. The "difference in proportions" test
indicated that among wives aged 15-29 years. those in
polygynous unions were more likely to be illiterate (t =
1.816. alpha less than 0.05. one-tailed test) and more
likely to have husbands present in the household (t =
1.839. alpha less than 0.05. one-tailed test) in comparison
to their counterparts in monogamous unions but no more
likely to be employed (t = 0.964. see Table 1). Among
wives aged 30—49. the difference in proportions test
suggested that wives in polygynous unions were no more
likely than their counterparts in monogamous unions to be

illiterate (t = 0.038) or to have husbands present (t

0.138) but much more likely to be working for pay (t

1.815. alpha less than 0.05. One-tailed test: see Table 1).
Because the differences in these three variables between
wives in polygynous and monogamous unions depended on age
group. we decided to conduct separate multivariate analyses
for women aged 15-29 and 30-49.

Table 1 shows that wives in polygynous unions were
somewhat older on average than those in monogamous unions.
Among wives under 30. the average age of wife at interview
was 23.68 years for wives in polygynous unions. and 22.81

years for wives in monogamous unions and the difference

21

between means was statistically significant (t = 3.35).
Among older wives (over 30 years). the average age of wife
at interview was 36.30 years for wives in polygynous unions
and 34.97 years for wives in monogamous unions and the
difference between means was statistically significant ( t
= 4.01).

Table 1 shows that younger woemn (below 30 years). in
polygynous unions had higher average number of children
ever born than their peers in monogamous unions. The
average numbers of children ever born were 2.35 and 2.18
children for wives in polygynous and monogamous unions.
respectively; but the difference between means was not
statistically significant ( t = 1.13). Among older wives
(over 30 years). wives in monogamous unions had higher
average number of children ever born than their peers in
polygynous unions. The average numbers of children ever
born were 5.61 and 5.65 children for wives in polygynous
and monogamous unions. respectively but the difference
between means was not statistically significant (t = -.17).

Among younger wives (below 30 years). the percent of
childless wives was 19.6% and 20.80% for wives in
polygynous and monogamous unions. respectively (Table 1).
but the difference between the proportions was not
statistically significant ( t = -.375). Among older wives

(over 30 years). the percent of childless wives was 7.4%

22

and 4.61 for wives in polygynous and monogamous unions.
respectively. and the difference of proportions was
statistically significant (1.866).

Also. table 1 shows that the percent ever using
contraception was not much higher among wives in monogamous
unions for both age groups. Among younger wives (below 30
years). the percentages ever using contraception were 0.5%
and 1.3% for wives in polygynous and monogamous unions.
respectively but the difference of proportions was not
statistically significant ( t = -1.000). Among older wives
(over 30 years). the percentages ever using contraception
were 0.8% and 2.2% for wives in polygynous and monogamous
unions. respectively but the difference of proportions was
not statistically significant ( t = -1.56).

Table 1 suggests that the small differences between
wives in polygynous and monogamous unions in average number
of children ever born could have resulted from sampling
error alone. Our literature review. however. implied that
significant differences between the two groups of wives
ought to be found when their numbers of previous marriages
were taken into account. To explore this possibility. we
turn now to table 2.

Table 2 shows that the number of children ever born
alive per wire varies with her marital history. Among

younger wives (below 30). wives with multiple marriages

23

Table 2. Average Nuaber of Children Ever Born Per Hife by Age of
Rife at interview. Type of Marriage. and number of lines

 

Harried.
Hives Aged 15 - 29 Yrs Hives Aged 30 - 49 Yrs
Polygynous Honogaaous Polygynous honogaaous
t t
Unions Unions Unions Unions

 

Harried only once

 

 

 

i 2.33 2.09 7.37 6.09 5.70 7.35
77 32 972 729 776

Harried lore Than Once
i 2.57 2.68 - 0.56 5.72 5.49 -7.09
77 59 777 727 267

 

24

have higher fertility than once married wives. regardless
of whether polygynous or monogamous unions are considered.
Younger polygynous wives who have been married only once.
had higher fertility than their peers in monogamous unions
who have been married only once. but the difference between
the two means was not statistically significant (t = 1.37).
Also Table 2 snows that younger polygynous wives who have
been married more than once. had lower fertility than their
peers in monogamous unions who have been married more than
once. but the difference between the two means was not
statistically significant (t = -0.56).

Among older wives (aged 30-49). once married wives had
higher fertility than their peers who had been married
only once. regardless of whether polygynous or monogamous
unions are considered. Older polygynous wives. who had
been married only once. had higher fertility than their
peers in monogamous unions who have been married only once.
but the difference was not statistically significant (t =
1.35). Table 2 also shows that older polygynous wives. who
had been married more than once. had lower fertility than
their peers in monogamous unions who had been married more
than once. but the difference was not statistically
significant (t = -1.09). However. Table 2 does not show how
these interrelationShips are affected by the other relevant

variables in Table 1. Thus we now turn to a multivariate

25

analysis.

A multivariate analysis of variance showed three
important factors which affected the number of children
ever born to wives under age 30 years (Table 3. panel 1).
Young women who had started living with their first
husbands at earlier ages or who were not working for pay at
the time of the survey had higher fertility than did other
other young women. Younger women who had ever used
contraception had borne a significantly larger mean number
of children (3.13) than had women who had never used
contraception (2.21). The reason for attempting
contraception in Somalia is most likely to stop fertility
rather than to time the births of children. and women with
a greater number of live births are more likely to decide
to stop. However. after the relationships between
cumulative fertility and these three variables had been
controlled. there was no significant residual difference in
the number of Children ever born to wives according to the
type of current marital union (f=0.213) or the number of
husbands they had ever wed (F=0.248).

Two factors showed a statistically significant
relatiOhShip to the number of children ever born to wives
aged 30-49 at the survey (Table 3. panel 2). As for the
younger wives. these older wives who had started

cohabitation with their first husbands at younger ages had

Table 3. Analysis of Variance in Number of Children Ever

26

Born to Urban Soaali Hives by Age of Hife at Survey

 

 

 

 

 

 

Source of Variation Sun of Degrees hean
Souares of F
freedon Souare
Hives Aged 15-29
Age of wonen at first
conabitation 1vrsl 565.949 1 565.949 116.191 ""
Literate 10=no:1=yesl 1.411 1 1.411 0.443
Employed il=yes:2=noi 32.428 1 32.428 10.130 ""
husband awav10=no:i=vesi 4.159 1 4.159 1.481
Ever contracept (1=yes:2=nol 12.381 1 12.381 3.868 "
Form of aarriage
11:001vgynv:2=nonoganyl 0.681 1 0.681 0.213
Number of tines wife married
il=oncez2=norel 0.194 1 0.194 0.248
interaction between
Fern of marriage and
3 of times wife narried 3.244 1 3.244 1.013
Explained 623.181 8 11.898 24.334 ""
residual 4200.024 1312 3.201
Total 4823.211 1320 3.654
Hives Aged 30-49
Age of wonen at first
cohabitation (yrsl 900.034 1 900.034 104.449 ""
Literate 10=no:1=yesi 4.014 1 4.014 0.466
anloyed ll=yes:2=nol 4.106 1 4.106 0.546
Husband away10=no:1=yesl 5.504 1 5.504 0.639
Ever contraceot (1=yes:2=noi 5.363 1 5.363 0.622
Fora of aarriage
11=oolygyny32=lonogalyl 0.900 1 0.900 0.104
Huaber of tiles wife aarried
1|=once:2:lorel 246.904 1 246.904 28.653 ""
interaction between
Fora of aarriage and
1 of tiles wife aarried 21.249 1 21.249 3.162 '
Explained 1201.322 8 150.165 11.421 ""
residual 10.986.638 1215 8.611
Total 12.181.960 1283 9.500
' p( 0.10
" o( 0.05
"' p( 0.01

"" o( 0.001

27

more children on average than their peers who had older
ages at first cohabitation (F=104.449. p < .001). In
addition. the type of the older wife’s marital union was
importantly associated with her cumulative fertility. but
this association was conditioned by her number of previous
marriages (F=3.162. p < .10). Polygynously wedded women had
a significantly higher cumulative number of lifetime births
than did monogamously wedded women if both groups had been
married only once but a significantly lower cumulative
number of lifetime births if both groups had been married
more than once (see table 2). These results support the
interpretation that polygyny is related to the cumulative
fertility of older wives (aged 30-49) but that the
relationsnip is conditioned by their number of previous
marriages. These results closely parallel the finding of
Isaac (1980) for the Mende tribe of Sierra Leone.

A question of importance to the family planning program
is why the ever - use of contraception would be
significantly related to the fertility of the younger wives
(aged 15-29) but not of the older wives (aged 30-49). A
possible answer is the recency of the start of the family
planning program. which was begun in 1981. only two years
before our 1983 survey. Because the younger wives had
significantly lower average fertility than the older ones

(Table 1). the relationship of contraception use to

28

fertility was probably more readily visible for the younger
wives. It is likely that future studies of the family
planning behavior in Somalia will show higher percentages
of wives ever using contraception than we found in our 1983
data and will show impacts of such usage on the fertility

both of older women and younger women.

Conclusions

This study has focussed on the effect of polygyny on
fertility. The data that were used in this study were from
the Somalia Family Health Survey which was aimed to
provide comprehensive data necessary for planning and
evaluating the Family Health and the Family Planning
programs.

In this study we found that among younger wives (below
age 30). those in polygynous and monogamous unions had
highly similar cumulative fertility. once the effects of a
number of relevant variables had been controlled in a
multivariate analysis of variance. Among older wives
(aged 30-49). polygynously married women had significantly
lower fertility than their peers in monogamous unions if
both groups of wives had been married more than once but
significantly higher fertility than women in monogamous
unions if both groups of wives were living with their first

husbands. These effects of polygyny on the fertility of the

29

older wives (aged 30-49) but the absence of effects on the
younger wives probably means that the effects become
manifest only in the later years of childbearing. The
positive effects among the once—married older wives may
represent a reproouctive advantage of being a senior wife
in a polygynous union over being an only wife in a
monogamous union. The present study could not separate the
effects of wife rank from marital history. and this is a
question future researcn should undertake. However. the
negative effect of polygyny on fertility among older
remarried women may represent the disadvantages of rank as
a Junior wife. the time spent between marital unions. or
neither influence. future research could resolve these
questions. The mixed findings in previous studies of the
impact of polygyny on fertility may reflect only the fact
that the relationship is not simple.

The present study was limited to the five largest
cities in Somalia. The practice of polygyny may well be
more common in rural areas. since such places have had less
contact with westernization. Thus. future studies could
profitably explore the relationship between polygyny and
fertility in rural Somalia. As the present investigation
shows. future work should unravel how this relationship
might interact with the wife’s age and the number of her

previous marriages. Such new research can go beyond the

30

present one by distinguishing between the number of
children born to a woman in her current marriage and the
number born in previous marriages and by determining the
type (polygynous and monogamous) of previous marriages.
These details would allow us to specify to what extent a
polygynous union affects the number of children born to a
wife within it and to what extent her fertility affects the
type of union into which a woman remarries. Answers to
these questions will show how much polygyny may be speeding
or slowing the growth of the national population. since

most Somali households reside in rural areas.

 

 

31

BIBLIOGRAPHY

Bean. L. L.. and G. P. Mineau. 1986." The polygyny
fertility hypothesis: A Re-evaluation" Population
Studies 40: 67-81.

Blalock. Huoert n.. Jr. 1972. Social Statistics. New York:
McGraw-Hill 800k Company. second edition

Chanacka. H. 1980 " Polygyny and the rate of population
growth" Population Studies 34: 91—107

CulwiCK. A. T.. and Culwick. G. M. 1939 " A study of
population in Ulanga. Tanganyika Territory" Sociological
Review 30(4): 365-379

Oorjahn. V.R. 1958. " The factor of of polygyny in African
Demography" in w. Bascom and M. Herskovits . (eds)..
n i i gng_ gngngg in Afrigan letngsg Chicago.
University of Chicago Press.

International Statistical Institute. 1983 a. Ing ‘gflggg
fertility Survey; 1979-80. A Summary 9f Findings Report
No. 39 of the World Fertility Survey. Voorburg.
Netherlands.

International Statistical Institute. 1983 b. The Cameroon
Fertility _S_(_3rve_yi 1978. A Summary 9f Findings Report No.
41 of the World Fertility Survey. Voorburg. Netherlands.

Isaac. Barry L. 1980. " Female Fertility and Marital Form

Among the Mende of Rural Upper Bambara Chiefdom. Sierra

32

Leone" Ethnology 19: 297—313.
Lewis. I. M. 1962. " Marriage and the Family in North
Somaliiand". Kampala. Uganda. East African institute of

Social Research. East East African Studies No. 15.

 

 

Lorimer. F. 1954. Culture and Human Fertility United
Nations Educational. Scientific and Cultural
Organization. Paris.

Ministry of Health. I985.Fertility and Family Planning in

 

Urban Somalia: Results of the Somali Family Health Survey

 

 

in Five Cities. 1983 Mogadishu

Muhsam. H. V. 1956. "Fertility of polygamous marriages"
Population Studies 10 (1): 3-16.

Mosley. W.. Werner. Linda . and Becker. Stan . 1982." The
Dynamics of Birth Spacing and Marital fertility in
Kenya". World Fertility Survey Scientific Report No. 30.
Voorburg. Netherlands: International Statistical
Institute.

Ohadike. P. O. 1968." A demographic note on marriage. family
and family growth in Lagos. Nigeria" in J. Caldwell and C.
Okonjo (eds.). [fie Population gfi Tropical
Africa New York. Columbia university Press.pp.379-392.

Olusanya. P. O. 1971. " The problem of multiple causation in
population analysis with particular reference to the
polygamy-fertility hypothesis" Sociological Review

l9(2):l65-l78.

Namfua. P. P. 1984. " Polygyny in Tanzania: its

33

Determinants and Effect on Fertility" Thesis (Ph. 0.)
Johns Hopkins University. University Microfiims
International. Ann Arbor. Michigan.

Ukaegbu. A. 1977." Fertility of women in polygynous unions

in rural Eastern Nigeria" Journal of Marriage and Family

39: 397-404.

Ware. Helen. 1979. " Polygyny: Women’s Views in a

Transitional Society. Nigeria 1975" Journal of Marriage

and the Family 41: 185-195.

   

"'11711111711111!1171111111711111s