Vous êtes sur la page 1sur 15

Age at First Marriage in Peninsular Malaysia*

BARBARA VON ELM


International Center for Research on Women**

CHARLES HIRSCHMAN
Duke University***

The average age at marriage among women in Peninsular Malaysia has risen dramatically from 18.5 years in 1947 to 22.3 years in 1970 (based on census data calculations of the "singulate mean age at marriage"). This paper examines the socioeconomic determinants of the average age at first marriage among women age 25-44 in
1966-1967 who were interviewed in a cross-sectional fertility survey of currently married women in Peninsular Malaysia. Substantial differentials in age at first marriage
are associated with ethnicity, years of formal schooling, and premarital work experience, while lesser differences are observed /or social and geographic origins.
Ethnic differences in age at first marriage remain wide even after statistically controlling,fvr other socioeconomic background variables. A life-cycle model of the effects of social background variables on age at first marriage is estimated using dummy variable regression analysis. Post-primary schooling and working before marriage are the strongest variables that delay age at first marriage. These patterns are
common to all ethnic communities. As more women participate in higher education
and in employment, age at first marriage will probably be further delayed.
There are relatively few empirical generalizations about the social and economic determinants of age at marriage. It is generally believed that rural tradition fosters early marriage, while urbanization and other forces of
modernization lead to marriage postponement. Extant data, however, have not always
shown this to be true. The most obvious
anomaly is the "Western European marriage
pattern," which indicates that prior to the Industrial Revolution, women commonly married in their mid- to late twenties (Gaskin,
1978; Hajnal, 1965). Furthermore, there is
We thank the National Family Planning Board of
Malaysia and the Department of Statistics of Malaysia
for permission to use the 1966-1967 West Malaysian
Family Survey data. Professor James Palmore kindly
provided a copy of the data tape and made constructive
comments on an earlier draft of this paper. We also
thank Teresa Dark for typing the paper.
International Center for Research on Women, 1010
16th Street, N.W., Washington, D.C. 20036.

Department of Sociology, Duke University, Durham, North Carolina 27706.

November 1979

Vol 41
(November):877-891.

little evidence which shows homogeneity in


the patterns or trends in the timing of nuptiality among less developed non-Western
societies.
As Dixon (1971) and others (Mitchell,
1971; Lesthaeghe, 1974; Salaff, 1976) have
noted, there are a variety of institutional factors, both at the societal and familial level,
that influence marital patterns. Among such
factors are the relative numbers of eligible
single men and women at the appropriate
ages, the familial structure in which young
couples may be supported through the extended household, and the relative status of
single women, including their opportunities
for higher education and paid employment.
While these factors are undoubtedly important, the specification of the key determinants in any single society may vary considerably. Moreover, factors which may be important at a single point in time may not be
those which account for trends over a period
of time. A brief look at variations in age at
first marriage and in the proportions marrying in younger age groups within the same

JOURNAL OF MARRIAGE AND THE FAMILY

877

region shows the tremendous diversity that


needs to be explained.
In recent decades there has been a general
rise in the average age at first marriage in
Asia, though the initial age levels and rate of
change have varied considerably from
country to country. Table 1 presents the proportions of women who first married at ages
15-19 and 20-24, as well as the "singulate
mean age at marriage" (for discussion of the
"singulate mean age at marriage" and
method of computation, see Hajnal, 1953;
Shryock, and Seigel, 1973:295) for selected
Asian countries. The most notable observation is that early marriage (under 20 years)
has not been a universal norm in any of these
countries, during this century at least. For example, the singulate mean age at first marriage was 18 years in Sri Lanka in 1901 and in
Peninsular Malaysia in 1947; even then only
one-half of the women aged 15-19 were married in these countries at these dates. In the
Philippines and Thailand, where generally
less than 20 percent of the women were married in their teenage years, there have been

only modest upward trends in the average age


at first marriage since the late 1940s.
In contrast, the declines in the proportions
married among young women in Sri Lanka,
Taiwan, and Malaysia have been substantial
in recent years. The current situation in these
countries, with 85 to 90 percent of the 15-19
year old women and about 50 percent of the
20-24 year old women still single, is fairly
comparable to contemporary Western levels
of nuptiality. Of course, these Asian countries are not necessarily representative of all
less-developed areas. In many other Third
World countries, very young marriage remains all but universal. However, these data
do suggest that there is considerable variation
in the levels and trends in age at first marriage in several Asian countries, irrespective
of socioecpnomic development.
Since there is no certitude about the comparative patterns of age at first marriage, nor
of the key determinants which affect nuptiality trends and transitions, a basic strategy
for more intensive research is needed. The
recent studies of Coale (1971) and Dixon

TABLE 1. TRENDS IN PERCENTAGES MARRIED AND MEAN AGE AT FIRST MARRIAGE OF WOMEN
FOR SELECTED ASIAN COUNTRIES__________________
Percent Marrieda
Singulate Mean Age
Ages 15-19
Ages 20-24
at First Marriage
Philippines (Ever-Married)
1903

1939
1948
1960
1970

26.4
19.7
14.9
12.7
10.8

66.7
63.8
59:3
55.7
49.7

20.9
21.9
22.1
22.3
22.8

51.9
24.7
24.3
15.0
10.5

79.0
70.6
67.5
58.7
46.9

20.7
20.9
22.1
23.5

17.9
12.5
17.5

64.2
56.4
57.9

21.1
21.6
21.9

12.5
5.9

44.4

42.2
37.0
16.1

86.7
78.6
57:0:

Sri Lanka (Ever-Married)

1901
1946
1953

1963
1971
Thailand (Currently Married)
1947

1960
1970

18.1

Taiwan (Currently Married)

1961
1974

60.9

Peninsular Malaysia (Ever-Married)

1947
1957
1970

18.5
19.4

22.3
^n two countries, data on currently married, rather than ever-married, are presented. At young ages, differences between currently married and ever-married are neghgible.
Sources: Concepi6n and Smith, 1977:20; Fernando, 1975:185; Arnold et at., 1977:8-9; Freedman et al
1977:13; Hirschman and Fernandez, 1978.

878

JOURNAL OF MARRIAGE AND THE FAMILY

November 1979

(1971) are instructive in the measurement of


nuptiality patterns and in the basic theoretical expectations for comparative research
with aggregate data. In this paper, a somewhat different approach is taken. We formulate and test a model of the determinants of
age at first marriage in Peninsular Malaysia.
Using individual level data from the 19661967 West Malaysian Family Survey (National Family Planning Board, 1968), a national survey of currently married women, it
is possible to measure the impact of social
origins, residence, education, and premarital
work experience on the age at which women
marry. Because Malaysia is a plural society
with three distinct ethnic groups, we are able
to study the variations in age at first marriage among these communities within a
single analytical framework. Such an intrasocietal type of analysis might well provide an
alternative model of research that could be
extended to other societies in future comparative research.

PENINSULAR MALAYSIA
Formerly known as Malaya and West Malaysia, Peninsular Malaysia is part of the nation of Malaysia. Malaysia was formed in
1963 with the union of Malaya (which had
been an independent nation since 1957), and
two former British colonies on BorneoSabah and Sarawak. Singapore was briefly
part of Malaysia from 1963 to 1965. While
geographically large, Sabah and Sarawak are
sparsely populated, and Peninsular Malaysias population of 8.8 million comprised
almost 85 percent of the population of the entire country in 1970 (Department of Statistics, 1977:269). Because of data limitations,
only Peninsular Malaysia is included in our
study.
With the exception of the city-state of Singapore, Malaysia is the most affluent nation in
Southeast Asia, measured in conventional
GNP terms (World Bank, 1976:498). This
relative prosperity rests in part on the export
of primary products, rubber, tin and oil
palm. but also on growing industrial and
commercial sectors.
In terms of population, the most salient
characteristic of Malaysia is her plural or
multi-ethnic composition. Slightly over onehalf of the population is composed of Malays,
over one-third is of Chinese descent, and ap-

November 1979

proximately one-tenth is of Indian origin.


The Chinese and Indian populations (and a
share of Malays as well) are the descendants
of immigrants who arrived in the late 19th
and early 20th centuries. In spite of the fact
that the majority of Chinese and Indians are
second or third generation Malaysians, there
has been relatively little socioeconomic and
cultural assimilation among the ethnic communities (Hirschman, 1975). In general, the
Chinese are the most urbanized group and
have a more diversified occupational structure than do either the Indian or Malay communities. On most measures of socioeconomic levels, Indians rank between the Chinese and Malay populations. In short, these
three ethnic communities, with their differing
cultural and socioeconomic characteristics,
provide a unique opportunity for a comparative analysis of marriage patterns within the
context of one country.

TRENDS IN AGE AT FIRST MARRIAGE


IN PENINSULAR MALAYSIA
The 1966-1967 West Malaysian Family
Survey (National Family Planning Board,
1968), the primary data source for this study,
is one of the few Malaysian surveys with a
direct measure of age at first marriage for a
representative sample of women. However,
the 1947, 1957, and 1970 censuses do provide us with an indirect measure of age at first
marriage, since they have information on the
percentages of ever-married women by age
for each major ethnic group (see Table 2).
These data also allow for the estimation of the
"singulate mean age at marriage" figures
(Hajnal, 1953; Shryock and Seigel, 1973:
295). The singulate mean age at marriage is
simply a cross-sectional summary measure of
the percentages of the population married in
each age-group for a particular year.
From 1947 to 1957, there was only a
modest trend towards later age at marriage
among Malays and Indians (.5 of a year or
less increase in the singulate mean age at first
marriage), but there was a significant change
of 1.6 years among young Chinese women.
Age at first marriage had always been somewhat later among the Chinese than among
the Malays or Indians. For instance, less than
one-out-of-five Chinese women aged 15-19
was married in 1947, while more than onehalf of the Malays and Indians in this age

JOURNAL OF MARRIAGE AND THE FAMILY

879

TABLE 2. PERCENTAGE OF WOMEN EVER-MARRIED BY AGE GROUP AND SINGULATE MEAN AGE
AT FIRST MARRIAGE, BY ETHNIC COMMUNITY: PENINSULAR MALAYSIA, 1947,1957, AND 1970
Years
15-19
20-24
25-29
30-34
35-39
40-44
45-49
(15-49)

1947
42.2
86.7
96.1
97.6
96.5
97.8
97.5
(84.7)
18.5

1957
Total
37.0
78.6
94.4
97.9
98.5
98.5

98.6

1970

1947

1957

1970

Malay

16.1
57.0
86.2
94.4
96.7
98.1
98.7
(67.5)
22.3

59.2
93.4
97.9
98.7
95.9
98.9
98.6
(89.8)
17.4

54.1
90.6
97.6
98.8
99.2
99.4
99.4
(87.6)
17.9

22.7
67.6
91.3
96.7
98.1
98.9
99.3
(73.1)
21.1.

(80.6)
19.4
Mean Age at First Marriage
Indian
Chinese
17.0
53.2
52.3
6.0
10.3
17.6
15-19
63.0
90.6
40.3
93.1
56.8
73.9
20-24
88.3
97.5
78.6
98.2
88.6
92.1
25-29
96.1
98.8
98.9
90.5
95.8
96.2
30-34
99.5
97.9
99.0
94.3
97.3
96.7
35-39
98.6
99.5
98.6
97.4
96.6
96.4
40-44
99.0
99.4
98.4
97.6
97.5
96.3
45-49
(67.7)
(88.1)
(89.2)
(68.3)
(59.0)
(76.3)
(15-49)
21.7
17.6
17.9
22.1
24.2
20.5
Mean Age at First Marriage
Sources: Del Tufo, 1949:204-259; Department of Statistics. Federation of Malaya. 1960:72-76; Department
of Statistics, Malaysia, 1977:355-359.

group were married. These ethnic differences


further widened by 1957, when the mean age
at first marriage for Chinese women rose to
22.1, while it remained below 18 years for
Malay and Indian women.
However, the 1957 to 1970 period was one
of declines in proportions married among all
ethnic communities. While the mean age at
first marriage rose more than two years for
Chinese women (to 24.2 in 1970), it increased
3.2 years for Malays and almost four years for
Indians. About 80 percent of all teenage
(15-19 years) Malay and Indian women and a
third of those in their early twenties (ages
20-24) were still single in 1970. While the proportions of single Malay and Indian women
were still considerably less than the prevailing
Chinese levels, the gap had narrowed considerably.
This dramatic increase in age at first marriage in Malaysia has been a key element in
the recent reduction of fertility in Malaysia
(Hirschman and Femandez, 1978). It would
be of considerable interest to understand the
social and economic determinants of this
nuptiality transition. However, the crosssectional survey data available for this study
do not really allow for an analysis of change.
But it does provide some assessment of the
main factors that are associated with age at
first marriage differentials between and within ethnic communities.
880

THE 1966-1967 WEST MALAYSIAN


FAMILY SURVEY
Prior to the launching of its program activities, the National Family Planning Board of
Malaysia, with technical assistance from the
Population Studies Center of the University of
Michigan and the Malaysian Department of
Statistics, conducted a benchmark survey
(1966-1967 West Malaysian Family Survey)
of fertility and family planning attitudes and
behavior among Malaysian married women.
Our study is a secondary analysis of these survey data, which had a sample of 5,457 currently married women between the ages of 15
and 44. While the sample was stratified to
overrepresent the largest cities, it can be
weighted to provide a representative national
sample, as is done in this study. (For additional details on the survey methods and sampling procedures, see the published report by
the National Family Planning Board, 1968.)
The survey has also been the source of a number of analyses of Malaysian fertility and
other studies (Palmore, 1969; Palmore and
Ariffin, 1969; Palmore, Klein, and Ariffin,
1970; Palmore, Hirsch, and Ariffin, 1971;
Hirschman, 1975). One question asked in the

1966-1967 West Malaysian Family Survey


was age at first marriage, which we use as our
dependent variable in this study. A variety of

social background variables was also collec-

JOURNAL OF MARRIAGE AND THE FAMILY

November 1979

TABLE 3. MEAN AGE AT FIRST MARRIAGE, PERCENTAGE MARRIED BEFORE AGE 18, AND PERCENTAGE MARRIED AT AGE 21 OR LATER OF
CURRENTLY MARRIED WOMEN, AGE 25-34 AND 35-44 IN 1966-1967 BY SELECTED BACKGROUND CHARACTERISTICS: PENINSULAR MALAYSIA,
1966-1967

Mean Age at Ma rriage

% Married

% Marrifd

atAf;e 17 or Beifore

atAf!e21 or Later

Total

Ethnic Community
Malay
Chinese
Indian

35^4

Total
(25-44)

25-34

68.6
19.2
55.4

69.0
24.1
45.9

68.8
21.1
51.6

63.7
47.9
31.0

63.8
51.9

25-34

35-44

16.2

20.8
17.6

16.5
20.3
18.0

16.3
20.6
17.8

16.6
18.5
19.6

16.8
17.8
19.4

16.6

16.6
19.9
20.3

16.8
19.3
19.4

16.7
19.7
20.0

16.6
17.8
17.8
18.7
21.1
22.8

17.0
18.1
18.4
19.2
21.1
22.8

16.8
17.9
17.9
18.8
21.1
22.8

41.8
5.7
6.8

16.8
18.2

17.0

18.6

17.8
18.5

16.9
18.0
18.5

62.4
50.6
42.1

17.6

17.5

17.5

54.1

%ofSaimple3

SamipleNb

Total

35-44 (25-44) 25-34

35-44

25-34

35-44

10.3
49.7
16.8

13.8
40.5
27.5

11.8
46.1
21.1

62.8
25.3
10.3

66.1
22.3
9.9

956
1164
329

669
818
226

11.9
28.7
38.1

15.9
26.8
32.9

13.6
27.9
35.9

57.1
19.5
16.7

57.7
18.5

37.3

63.8
49.5
33.6

17.0

815
750
754

582
459
552

63.6
30.8
25.8

63.2
39.8
35.0

63.4
34.2
29.7

11.7
42.9
50.4

16.7
32.0
34.5

13.8
38.9
43.6

72.8
15.2
11.9

74.7
12.5
12.3

1143
674
677

788
450
493

62.3
54.8
55.2

62.3
50.2

62.3
53.3
51.8
39.8

17.4
26.5
25.2
32.4

51.1
18.8
14.4

46.3
68.1

15.7
23.2
22:6
26.5
52.6
72.4

8.3
3.2
3.9

71.3
12.9
8:7
3.3
1.4
1.8

885
502
433
286
141
245

945
294
229
104
67
82

(25-44) 25-34

Fathers Occupation
Agricultural
Blue Collar
White Collar

18.2
19.5

Longest Place of Residence Before Marriage


Rural
Small Town

Town or City
Years of Schooling
None
1-3
4-5
6
7-8
9 or more

13.2

8.4

14.0
21.6
21.4
24.9
54.6
73.8

60.3
55.3

49.6

61.5
52.9
44.6

12.8
27.9
32.6

14.2
22:7
32.8

13.4
27.8
32.6

56.0
6.1
37.6

61.2
7.9
30.2

1345
144
1007

1171
102
459

56.5

55.1

21.1

20.9

21.1

100.0

100.0

2501

1743

43.9
32.5
22.1

9.7

Premarital Work Experience

None
Work Only at Home
Work Outside Home
Total Sample of Currently Married Women

^Weighted

to national population composition. The categories of each variable do not sum to 100% because others/dont know categories are not listed.
"Unweighted number of interviews. The figures for each variable do not sum to the total because others/dont know categories are not listed.

Source: 1966-1967 West Malaysian Family Survey (National Family Planning Board, 1968).

considerable variance about the mean; over


one-half of the women in the sample married
before age 18, while one-fifth married at age
21 or later. However, there is ho evidence of
any overall trend when comparing women
aged 35-44 with the younger cohort of women
aged 25-34. The selection bias in a crosssectional sample of currently married women,
as discussed earlier, masks the trend that we
have observed from census data.
There are, however, strong associations between age at first marriage and the ethnic
background variables shown in Table 3. Of
those Malaysian women aged 25-44 in
1966-1967, the Chinese evidenced by far the
highest mean age at first marriage; marrying
almost three years later, on the average, than
Indian women, and more than four years
later than Malay women. Put another way,
while almost one-half of the Chinese women
married after the age of 21, less than 20 percent of the Indians and only 10 percent of the
Malays did so.
The socioeconomic status of family origin,
as measured by fathers occupation when
respondent was twelve years old, also exerted
a strong influence on age at first marriage.
Two-thirds of the daughters of fanners married below age 18, whereas only one-third of
the daughters of white collar workers did so.
The effect of a blue-collar family origin was
VARIABLES AFFECTING AGE
intermediate between the effects of farm and
AT FIRST MARRIAGE
white collar origins.
In this section we describe the associations
Another variable of interest is that of ruralbetween several independent variables and urban origins. One question in the survey inage at first marriage among our sample of quired about the type of "place of longest
currently married women older than 25 years. residence prior to marriage," Presumably,
Table 3 presents three summary measures this place was the locale of adolescent socialiof age at first marriage: the mean, the per- zation and education which exposed the
centage marrying below age 18, and the per- respondent tocommunity norms about adult
centage marrying at 21 years or older by se- behavior. Women who grew up in either
lected social background variables for women small towns or in larger towns or cities were
aged 25-34 and 35-44. For each independent much more likely to postpone marriage than
variable in Table 3, the residual categories those who grew up in rural areas. Although
"other" and "unknown" are not listed, the difference in age at first marriage bealthough these cases are included in the tween those from small towns and those from
"Total" figures if data concerning their age towns or cities was minimal, it was in the exat first marriage were reported.
pected direction. In contrast to the lack of
For the total sample, the mean age at first inter-cohort change in the effects of other
marriage is 17.5 years, considerably lower variables, there was a substantial increment
than the singulate mean age at first marriage in the age of first marriage for those women
for either 1957 or 1970. (To some extent this from urban areas between the 35-44 and
lower estimate may reflect the absence of 25-34 age groups. Perhaps the rural-urban
young women who marry late from the gradiant has widened for these cohorts of
sample of currently married women.) There is women.

ted in this survey, including social origins,


longest place of residence prior to marriage,
education, and premarital work experience.
These variables allow for a rather comprehensive analysis of the sociostructural influences upon age at first marriage in Peninsular Malaysia.
There is an intrinsic bias in studying age at
first marriage in a sample restricted to currently married women. Young women who
have married are included in the sample
while those who have postponed marriage are
not._Tp_ reduce the effects of this bias, our
study limits the sample to married women
who were older than age 25 at the time of the
survey. By age 25, about 90 percent of the
Malay and Indian women, and almost 80 percent of the Chinese women were married in
1970. Another potential problem with the
sample is that formerly-married women ((. e.
those women who were divorced, separated,
or widowed) at the time of the survey were excluded, while women who had remarried were
included. We suspect that the association
between age and first marriage and marital
disruption is not large enough to affect our
results although we have no independent evidence on this.
have no independent evidence on this.

882

JOURNAL OF MARRIAGE AND THE FAMILY

November 1979

Of all the variables in this table, years of


education has by far the strongest effect. The
difference between no schooling and a few
years of primary education is modest but real,
resulting in about one year of marital postponement. The completion of primary school
(through the sixth grade) delays marriage
almost another year. But entry into lower
secondary education (7th and 8th grades) has
the strongest effect; the mean age at first.
marriage increases another two to three
years. Another way to express these differences is to consider the percentage who marry
at age 21 or later. Of those women with primary schooling or less, only 25 percent
married at age 21 or later. But 50 percent of
the women with seven or eight years of schooling and almost 75 percent of the women with
nine or more years of schooling postponed
marriage until after age 21. The distribution
of the sample by education shows that only a
small fraction of women received any postprimary schooling (less than 8 percent of
women aged 25-34 and less than 4 percent of
women aged 35-44) in 1966-1967. But as education becomes more widely available to the
female population, its impact should grow.
The last independent variable in Table 3 is
premarital work experience. Women who
have worked are subdivided into two categories. those who worked only at home and
those who worked outside the home (also includes women who worked both at home and
outside the home). Work is defined according to the standard labor-force concept which
includes unpaid family workers in family
farms or enterprises. Housework and care for
small children at home is not part of this conventional measure.
Theoretically, work experience can postpone marriage of young women for two reasons. Families whose daughters are economically active may not be so eager to lose them
from the parental household. Additionally,
young women may wish to maintain their independence for a while before settling down
to traditional domestic roles, and work provides them with the necessary resources to
remain single. The differences in age at first
marriage by premarital work experience are
in the expected direction, but of a fairly
modest magnitude, since only a 1.6 year difference in mean age at first marriage separates those who never worked from those who
worked outside the home.

November 1979

EXPLAINING ETHNIC DIFFERENCES


The considerable ethnic gap in average age
at first marriage may have a variety of explanations. One interpretation is that Malays
are more likely to live in rural areas, have
agricultural origins and have fewer educational opportunities than the Chinese and
Indians, and that differences in marriage age
simply reflect ethnic differences in these
background variables. Another interpretation is that differences in ethnic cultures or
value orientations are the primary reason for
earlier Malay marriage, regardless of structural differentials. To address this issue,
Table 4 shows the same three age at first
marriage indicators (mean, percentage of
women below age 18, and percentage of
women 21 years or older) for the same selected background variables in Table 3, separately for Malay, Chinese, and Indian
women, aged 25-44 in 1966-1967. Data presented in this table allows us to address two
questions: (1) what are the ethnic differentials, if any, within the same categories of
each independent variable? and (2) do the
relative effects of these variables differ across
ethnic groups?
In general, .ethnic differences remain at
almost their original level, even when holding
background constant. Chinese women who
are daughters of farmers marry almost four
years later than do Malay women with agricultural origins. Within white-collar families,
the Chinese-Malay gap is narrowed, but still
remains substantial with a three-year differential. In fact, only when women attain postprimary education, do ethnic differences become negligible. For all other background
variables, including residence prior to marriage and work experience, Chinese women
marry substantially later (three to four years)
than Malay women (with Indians occupying
an intermediate position).
Addressing the second question raised
above, the effects of background variables on
age at first marriage are attenuated, but still
remain important within ethnic communities.
For instance, the difference between being
the daughter of a farmer or a white-collar
worker is about three years in mean age at
first marriage overall (Table 3), but only
about one year within each ethnic community (Table 4). The effects of education on
age at first marriage are still substantial with-

JOURNAL OF MARRIAGE AND THE FAMILY

883

TABLE 4. MEAN AGE AT FIRST MARRIAGE, PERCENTAGE MARRIED BEFORE AGE 18, AND PERCENTAGE MARRIED AT AGE 21 OR LATER OF
CURRENTLY MARRIED WOMEN, AGE 25-44, BY ETHNIC COMMUNITY AND SELECTED BACKGROUND CHARACTERISTICS: PENINSULAR MALAYSIA, 1966-1967
Mean A.ge at Miimage

0
M

Fathers Occupation
Agricultural
Blue Collar
White Collar

>

Longest Place of
Residence Before

Marriage

Rural
Small Town

2>

Town or City
Years of Schooling
None

Tl

>
H

w
2>

r:
Z

I3

1-3
4-5
6
7-8
9 or more

% Married
at Age 17 or Belore

% Marrif;d
at Age 21 or Later

S ample N11

% of Samiple Popuilation3
Malay

Chinese

Indian

Malay

Chinese

ndian

18.3,
22.4
25.9

72.1
13.1
8.5

24.1
29.6
37.2

39.8
33.5
20.4

846
327
330

395
618
790

138
241
149

36.5
48.7

17.8
27.3

22,6

52.2

34.7

89.2
6.1
4.7

31.7
35.2
32.0

74.1
15.0
10.8

1116
233
226

488
679
803

256
191
105

51.6
56.1
58.4
80.6
29.9
13.4

11.1
3.4
15.8
12.5
47.6
54.9

35.9
48.6
49.8
47.1
65.7
81.6

17.7
22.9
11.3
8.4
35.8
68.3

67.1
12.6
12.6
5.1
1.5
0.8

42.1
25.6
10.7
10.5
4.5
6.5

53.3
19.4
13.5
3.8
3.3
5.9

837
227
312
124
60
54

764
473
254
225
104

158

216
90
96
38
39
64

27.7
20.7
13.5

58.3
*
44.1

7.6
16.6
19.2

34.0
58.9
58.1

12.9
*
29.3

62.6
7.8
29.1

49.7
7.5
42.8

50.1
1.4
48.3

1135
103
381

990
136
854

370
7
174

21.1

51.6

11.8

46.1

21.1

100.0

100.0

100.0

1626

1982

555

Malay

Chinese

50.4
52.6
51.7

10.4
13.0
20.6

38.7
47.5
46.7

28.9
19.5
15.2

51.8
51.0
51.4

10.6
21.1

71.4
80.8
61.3
54.6
5.0
11.1

30.5
16.6
17.9
14.4
8.4
2.1

18.5

72.4
64.0
62.1

17.8

68.8

CChinese Indian

Malay

Chinese

Indian

Malay

16.2
16.3
17.6

20.0
20.6

20,9

17.4
17.8
18.6

70.1
70.9
54.7

25.4
23.4
17.1

16.2
17.6
17.2

19.8
20.8
21.2

17.5
18.5
18.6

69.8
60.7
60.4

16.0
15.7
17.2
17.3
21.1
21.6

19,7
20.8
20.8
21.0
21.6
23.7

17.3
17.9
16.9
17.0
20.1
22.5

16.0
16.8
16.9

19.7
21.4
21.5

17.0

16.3

20.6

Indian

Premarital Work
Experience

None
Work Only at Home
Work Outside Home
Total Sample of
Currently Married

Women

<D

S"
^0
~i

v0

*Lessthan 10 cases.
Weighted to national population composition. The categories of each variable do no add to 100 percent because others/doint knoiw categoric! are not listed.
^Unweighted number of interviews. The figures for each variable do not sum to the otalbeoiuse others/dont know categolriesare not listed.
Source: 1966-1967 West Malayaan Family Survey (National Family Planning Board, 1968).

WOMEN,
TABLE 5 EFFECTS OF SOCIAL BACKGROUND ON AGE AT FIRST MARRIAGE OF MARRIED

AGES 25-44: PENINSULAR MALAYSIA, 1966-1967

Gross__________Net_________
Ethnic Community
Malay
Chinese
Indian

Others
Fathers Occupation
Agricultural
Blue Collar
White Collar
Not Reported

Unweighted (N)

-1.20
3.05
0.22

-0.84
2.29
-0.09

1621
1981
553
81

-0.88
0.67
1-98

-0.14
-0.02
0.34

1393
1208
1304
331

-0.82

-0.17
0.51
0.39

1924
1124

-0.43
0.25
2.47
3.43

1825

Longest Residence Prior to Marriage

Rural
Small Town
Town or City

2.12
2.41

1170
18

Not Reported
Educational Attainment

-0.74
0.53

None
Primary (1-6)

Lower Secondary (7-8)


9 or More Years
Not Reported

3.58

5.28

1846

208
327
30

Premarital Work Experience

""None------

2510
-0.52
-0:66
246
0.73
0.50
1465
0.72
1.00
Not Reported__________________ -_________^_______________"
Notes- Effects are expressed as deviations from the grand mean of 17.54 years.
the regression
Coefficients of the "Not Reported" categories are not listed here, but they were included m
Source: 1966^1967 West Malaysian Family Survey (National Family Planning Board, 1968).
Only at Home
Outside Home

in ethnic communities, though reduced somewhat. Working prior to marriage has a


stronger effect within the Chinese and Indian
communities than among Malay women. In
short, it appears that ethnicity (with the exception of post-primary education) is a much
more significant variable in affecting age at
first marriage than any of the structural vanables considered in this study,
One final test of the relative effects of
ethnicity and social background characteristics is presented in Table 5, which shows each
variables gross and net effects, i.e. controlling for the additive effects of all other vanables in the same regression equation. Effects
are expressed as deviations from the grand
mean of 17.54 years old at first marriage. The
net ethnic coefficients in Table 5 show the
ethnic differences after statistically holding
constant ethnic variations in fathers occupation, place of residence prior to marriage,
educational attainment and premarital work
experience among the sample of married
women. Controlling for all these vanables,
the Malay-Chinese difference in mean age at

November 1979

first marriage is reduced from 4.25 years to


3.13 years, the Malay-Indian difference from
1.44 to .75 years, and the Chinese-Indian difference from 2.83 to 2.20 years.
It is obvious that most of the ethnic differences in age at first marriage are not due to
differences in social background as measured
by these variables. Perhaps other structural
variables that more directly explain the differential exposure to "modernizing" influences or household-economic structures
would further reduce the ethnic differentials
in age at first marriage. But others might
argue that the basic differences are rooted in
cultural factors tied to religion and value
orientations on the appropriate roles for
young women. The question is one of great
significance, yet we are unable to resolve it
with the data at hand.

MODEL
^^SOCIOSTRUCTURAL
MARRIAGE
^
the preFrom the variables discussed
^g

pIRST

JOURNAL OF MARRIAGE AND THE FAMILY

in

885

FIGURE 1. LIFE-CYCLE MODEL OF SOCIAL BACKGROUND AND AGE AT FIRST MARRIAGE, PENINSULAR MALAYSIA, 1966-1967
ETHNICITY

EDUCATIONAL
.ATTAIIIMENT
None
Prinary (1-6)

Lower Secondary (7-8)


Nine

More Years

Not Reported

FATHERS OCCUPATION
White Collar
Blue Collar

Agricultural
Not Reported

PREMARITAL WORK
EXPERIENCE
PLACE OF LONGEST
RESIDENCE PRIOR
TO MARRIAGE
Rural Area
Small Town
Town
City

Not Reported

ceding section, it is possible to posit a causal


model of sociostructural influences on age at
first marriage. Following a life-cycle ordering, Figure 1 shows the paths of influence of
social background on age at first marriage.
Each variable is assumed to have a causal impact on all variables to its right (arrows are
omitted for the sake of clarity). The first variables in the model are those of ethnicity,
socioeconomic and spatial origins. Since ethnicity is so closely linked with the other variables, we use it as a control variable, running
separate models for Malays, Chinese and
Indians. Socioeconomic and spatial origins
are indexed by fathers occupation and type
of place of longest residence before marriage,
respectively. These variables are posited to
affect age at first marriage directly and indirectly through the intervening variables,
educational attainment and premarital work

experience.
Educational attainment is posited to have a
causal impact on marriage postponement,

886

net of the social origin variables. It seems


likely that because part of the effect of education is learned job skills or acquired aspirations, educational attainment could feasibly
result in delayed marriage. Thus, both direct
and indirect effects of education are hypothesized here. Work experience prior to marriage is also expected to delay marriage, net
of the other variables in the model.
The statistical method of analysis is
dummy variable regression or multiple classification analysis, with separate models being
run for each of the three stages of the model.
The results are presented in Table 6, separately for the three ethnic communities in
Peninsular Malaysia. For each variable, the
"Not Reported" category is included in the
analysis to avoid losing cases, but the coefficients are not listed, since they have no substantive meaning. Effects (regression coefficients) are expressed for each ethnic community as deviations from grand mean of age
at first marriage.

JOURNAL OP MARRIAGE AND THE FAMILY

November 1979

^
^^
2

a"

TABLE 6. EFFECTS OF SOCIAL BACKGROUND ON AGE AT FIRST MARRIAGE 01 MARRIED WOMEN AGES 25-44. SEPARATELY BY ETHNIC COMMUNITY: PENINSULAR MALAYSIA, 1966-1967

Model:

Fathers Occupation
Agricultural
BlueCoUar
White Collar
Not Reported

t-i

Longest Place of Residence Prior to

Marriage

>

^
2

g
SO
">

Rural Area

SmaUTown
Town or City
Not Reported
None
Primary (1-6 years)
Lower Secondary (7-8)
9 or more years
Not Reported

>
2

p
"<

-0.13 -0.19 -0.12


-0.10 -0.13 -0.06
1.03
0.71
0.81

-0.05 -0.05
0.56
0.51
0.70
0.22
0.27

-0.11

1.01

"->

21

(Mean

17.8 years)
3c
2

u,,^,gi,ted
(N)

(N)

843
327
329
122

-0.30
0.01 -0.11
-0.12 -0.06 -0.20
0.17 -0.12 0.08

395
617
790
179

-0.21 -0.09 -0.06


0.30 0.18
-0.03
0.59 -0.25 -0.01

137
241
148
27

1161
233

-0.73 -0.46 -0.52


0.14 0.21
0.17
0.50
0.25 0.24

487

0.07 -0.10
-0.20
0.07 0.50
0.57
0.60 -0.58 -0.03

254
191
105
3

-0.51 -0.48
-0.33 -0.28
2.45 2.44
5.19 4.51

214
224
39

-0.79
0.39
0.79

368
7
174
4

(N)

226

ia

679
803

I3

\1

-0.24 -0.29
0.17
0.27
4.42 4.39
4.72
4.48

834
661
60
54
12

-0.80 -0.77
0.22 0.27
0.92 0.75
2.97 2.52

1132
102
380
7

-0.88
0.88
0.87

763
952
104
158
4

64
12

Premarital Work Experience

None

-0.38
0.71
0.60

Only at Home
Outside Home

Not Reported

K^_______ _________________1.9

^^^

(Mean 20.59 years) unweighted

Educational Attainment

>
Z

16.34 years)
y
2t>

PI

^Tl

la

Indian

Chinese

Malay

(Mean

5.4

7.1_______

2.7

8.2

13:4

989
136
854
2

______1.4

11.7

15.2

Notes: Effects are expressed as deviations from the grand mean of each ethnic community.
Coefficients for "Not Reported" categories are not listed here, but they were included as part of the regression models.
SModel only includes fathers occupation and place of longest residence prior to marriage.
IModel 2 includes fathers occupation, place of longest residence prior to marriage, and educational attainment.
^odel 3 includes fathers occupation, place of longest residence prior to marriage, educational attainment, and premarital work experience.
Source: 1966-1967 West Malaysian Family Survey (National Family Planning Board, 1968).

Fathers Occupation

Socioeconomic origins has only a modest


effect on marital postponement in the Malay
community. The mean difference in marital
age between women from agricultural families and those from higher status families
(white-collar) is less than 1.2 years (Model 1).
There is essentially no difference between
those with blue-collar origins and those

having an agricultural background. Adding


educational attainment and premarital work
experience into the equation further reduces
the gap between agricultural and white-collar
origins to about .9 year.
Among Chinese and Indian women, the
difference in total effects (Model 1) is even
smaller (.5 for Chinese and .8 for Indians)
and is almost nonexistent in the subsequent
models when education and work experience
are controlled. In fact, when education is
held constant, the net effect of white-collar
origins among Chinese and Indians appears
to reduce age at first marriage relative to the
other socioeconomic origin categories.
These data suggest that there is not a
strong inverse relationship between socioeconomic origins and age at first marriage in
Peninsular Malaysia. It is only important to a
modest extent among Malay women.
Place of Longest Residence
Prior to Marriage

We hypothesize that growing up in an


urban area would tend to postpone age at
marriage. This should occur through indirect
channels; urban areas provide greater opportunities for higher education and jobs which
might lead young women to delay marriage.
In a direct way, urban society might also convey new social norms that later marriage is an
acceptable behavior for young women (and
for their families). The evidence in Table 6
modestly supports this hypothesis.
While the effects of the three categories of
rural-urban background are not always
linear, there is a consistent finding that
women from rural origins do marry at a
younger age (about a years difference in all
three ethnic communities, see Model 1).
When education is introduced in Model 2,
the effects of geographical background are
reduced by more than half among Malay and
Chinese women, and even reversed among
Indian women. Access to jobs, as measured
888

by premarital work experience, does,not seem


to mediate any of the impact of residential

background.
In sum, we conclude that social origin, as
measured by fathers occupation, and place
of longest residence before marriage, have
real, but fairly modest effects on delaying age
at first marriage, and most of these effects are
indirect through educational attainment.
Educational Attainment

Years of education is by far the most significant variable in the model. Primary education (1-6 years) has only a slightly greater
delaying effect than no schooling. The difference, net of social origins, is only .4 and .2 of
a year for Malays and Indians, respectively.
For Chinese, however, primary schooling
does postpone marriage for about a year relative to women who have had no education at
all. But most of the effect of education is for
those with at least lower secondary schooling.
The effects are smallest for Chinese, but still
substantial. Deviations above the mean age at
first marriage are approximately four to five
years for Malay and Indian women with post-

primary schooling.
The effects of education are more than just
a postponement of marriage until schooling is
completed. For instance, women who have
only seven or eight years of schooling will
have completed their education by age 14 or
15 at the latest, yet most postpone their marriage several years beyond this age. Thus, it
seems that education has other consequences
that influence young women (and their families) to delay marriage. Part of the effect of
education is a mediation of socioeconomic
and residential background as noted in
earlier paragraphs. But the Substantial increase in variance explained for Model 2 compared to Model 1 indicates that most of the
effect of education on age at first marriage is
independent of social background.
Surprisingly, very little of the effect of education on age at first marriage is mediated by
premarital work experience. We would
expect that one of the basic consequences of
higher education would be to enhance employment skills and aspirations which would
tend to delay marriage. But almost all of the
effects of education are found to be directly
on age at first marriage, at least in this
model.

JOURNAL OF MARRIAGE AND THE FAMILY

November 1979

Premarital Work Experience


Although work experience prior to marriage does not seem to mediate any of the
effects of social origins or education on age at
first marriage, it does have a modest net
effect on the postponement of marriage.
Malay women who worked prior to marriage
married about a year later than women who
did not, and the comparable difference is
about one and one-half years for Chinese and
Indian women. It may be that work provides
psychological and financial alternatives to
early marriage. Perhaps, too, parents are less
willing to marry daughters at an early age if
they are augmenting the family income. The
distinction between work at home and outside the home was initially thought to be very
significant, yet it appears to be of no conse-

quence.
DISCUSSION AND CONCLUSIONS
It is probably a futile effort to attempt to
develop a general theory of the nuptiality
transition. While the general forces of change
in marriage patterns may be similar in different contexts and at different times, no
common set of factors such as industrialization or urbanization seems to be universally
associated with the timing of marriage. Even
within the less developed nations of Asia,
there is a diversity of levels and trends that
bears no obvious relationship to relative levels
of socioeconomic development.
An alternative comparative strategy of research is to proceed inductively, by empirical
investigation of trends in age at first marriage
and the socioeconomic correlates of age at
first marriage. From this perspective, one can
develop a causal model that can be empirically estimated and used to interpret the
relative effects of social background characteristics on age at first marriage. In this
study, we have examined the case of Peninsular Malaysia. An extension of this strategy
to other societies and a comparison of the
results might lead to a more cumulative tradition of comparative research.
Comparisons of proportions of married
women in the censuses of 1947, 1957, and
1970 show a remarkable trend towards delayed marriage in Peninsular Malaysia.
While only the Chinese population registered
a significant trend toward delayed marriage
among women from 1947 to 1957, the 1957 to

November 1979

1970 period saw major changes for all three


ethnic communities. The singulate mean age
at first marriage was above 21 years for
Malays and Indians in 1970 and above 24
years for Chinese women. It appears that
teenage marriage is becoming less common in
Peninsular Malaysia.
These trends revealed by examination of
successive cross-sectional census data are not
detected in our analysis of age at first marriage of currently married women interviewed
in the 1966-1967 West Malaysian Family Survey. This sample of married women aged
25-44 is underrepresentative of those who
marry late because of the selection criteria
(currently married women) or other sampling
factors. Nonetheless, there is considerable
variance in age at first marriage in the sample, and we have no reason to suspect that
associations of background variables and age
at first marriage are biased.
Of all the background variables that are
associated with age at first marriage, ethnicity is by far the strongest. Controlling for
other characteristics (socioeconomic and residential origins, educational attainment, and
premarital work experience), either individually or simultaneously, ethnic differences in
age at first marriage are only slightly attenuated. It may be possible that further refinement of the control variables, such as land
tenure or land holdings of farmers, and types
of work experience of young women would
reduce the Malay-Chinese-Indian differentials. But the differences are so substantial
that it seems safe to conclude that a good
share of ethnic variations in age at first marriage are due to differential cultural orientations about the appropriate timing of marriage. But the convergence of ethnic differentials over time (from 1957 to 1970) and the
similarity of age at first marriage for all
women with post-primary education (Table 4)
warns against any conclusions that ethnic differences will necessarily persist.
Among the three ethnic communities, a
comparison of the models of age at first marriage shows a rough comparability of effects
of background variables. Socioeconomic and
residential origins have a modest influence
which may delay marriage for a year or so for
women with the most advantaged background relative to those with the least. A substantial share of the effects of socioeconomic

JOURNAL OF MARRIAGE AND THE FAMILY

889

and residential background are mediated by Fernando, Dallas


1975 "Changing nuptiality patterns in Sri Lanka
higher education.
1961-1971." Population Studies 29 (July): 179Malaysian
of
the
Less than 10 percent
190.
women in this sample have advanced beyond Freedman, Ronald, Tze-Hwa Fan, Sou-Pen Wei, and
primary schooling. But for the small share Mary Beth Weinberger
1977 "Trends in fertility and in the effects of educawho did, marriage has been postponed for
tion on fertility in Taiwan, 1961-74." Studies in
several years beyond those who had no
Family Planning 8 (January):ll-18.
The
education.
schooling or only a primary
Gaskin, Katharine
effects of education on marital postponement
1978 "Age at first marriage in Europe before 1850:
are largely independent of social background
A summary of family reconstitution data."
Journal of Family History 3 (Spring):23-36.
and subsequent work experience. Although
our data are unable to test such a hypothesis, Hajnal.John
1953 "Age at marriage and proportions marrying."
we would posit that the recent expansion of
Population Studies 7 (2):111-136.
educational opportunity in Malaysia has been
1965 "European marriage patterns in perspective."
largely responsible for the trend towards a
Pp. 101-143 in David Glass and D. E. Eversley
later age at first marriage. Work prior to
(Eds.), Population in History. Chicago:Aldine
Publishing Company.
marriage has a positive effect on marital postponement of about one to one-and-a-half Hirschman, Charles
1975 Ethnic and Social Stratification in Peninsular
years, depending on the ethnic community.
Malaysia. Arnold and Caroline Rose MonoThe effect of work experience is completely
graph Series. Washington, D.C.:American
independent of social background and educaSociological Association.
tion.
Hirschman, Charles, and Dorothy Femandez
1978 "The decline of fertility in peninsular MalayThe models used in this analysis represent
sia." Unpublished manuscript. Duke Univeronly the beginnings of a comprehensive
sity.
With
in
analysis of age at marriage Malaysia.
Lesthaeghe, R.
only 7 to 15 percent of the variance explained,
1974 "The feasibility of controlling population
there is considerable room for further identigrowth through nuptiality and nuptiality polification of the social and economic determicies." Pp. 319-341 in International Population
Conference, Liege 1973 (vol. 3). International
nants of age at first marriage in Malaysia and
Union for the Scientific Study of Population.
elsewhere.
REFERENCES
Arnold, Fred, Robert Retherford, and Anuri Wanglee
1977 "The demographic situation in .Thailand."
Papers of the East-West Population Institute,
No. 45. Honolulu: East-West Center.
Coale, AnsleyJ.
1971 "Age patterns of marriage." Population
Studies, 25 (July): 193-214.
Conception, Mercedes, and Peter C. Smith
1977 "The demographic situation in the Philippines:
An assessment in 1977." Papers of the EastWest Population Institute, No. 44. Honolulu:

East-West Center.
Del Tufo, M. V.
1949 A Report on the 1947 Census of Population.
London:Crown Agents for the Colonies,

partment of Statistics.

Department of Statistics, Malaysia


1970 Population Census of Malaysia: General
Report (vols. and 2). Kuala Lumpur. Department of Statistics.
Dixon, Ruth
1971 "Explaining cross-cultural variations in age at
marriage and proportions never marrying."
Population Studies 25 (July):215-233.

890

1971

"Changes in fertility rates and family size in


response to changes in age at marriage, and the
trend away from arranged marriages, and increasing urbanization." Population Studies 25

(November):481-489.
National Family Planning Board
1968 Report on the West Malaysian Family Survey,
1966-1967. Kuala Lumpur, Malaysia:Author.
Palmore, James

1969 "Malaysia: The West Malaysian Family Survey, 1966-1967." Studies in Family Planning 40
(April):ll-20.
Palmore, James, Paul M. Hirsch, and Ariffin bin Marzuki
1971

"Interpersonal communication and the diffusion of family planning in West Malaysia."


Demography 8 (August) :411-425.

Malaya.

Department of Statistics, Federation of Malaya


1960 1957 Population Census of the Federation of
Malaya: Report No. 14. Kuala Lumpur: De1977

Mitchell, Robert E.

Palmore, James, Robert W. Klein, and Ariffin bin Marzuki


1970 "Differential participation in extended families: West Malaysia 1966-67." American Jour-

nal of Sociology 76 (November):375-398.


Palmore, James, and Ariffin bin Marzuki

1969 "Marriage patterns and cumulative fertility in


West Malaysia, 1966-1967." Demography 6

(November):383-401.
Salaff, Janet
1976 "The status of unmarried Hong Kong women

JOURNAL OF MARRIAGE AND THE FAMILY

November 1979

and the social factors contributing to their delayed status." Population Studies 30 (November):391-412.
Shryock, Henry S., and Jacob S. Siegel
1973 The Methods and Materials of Demography.

November 1979

U.S. Bureau of the Census. Washington, D.C.:


U.S. Government Printing Office.
World Bank
1976 World Tables, 1976. Baltimore: The Johns
Hopkins University Press.

JOURNAL OF MARRIAGE AND THE FAMILY

891