modernization and consanguineousmarriage in iran pubs/a74.pdfmodernization. three decades ago,...

BENJAMIN P. GIVENS AND CHARLES HiRSCHMANUniversity ofWashington

Modernization and Consanguineous Marriage in Iran

Using data on 4,667 women from the 1976-77Iran Fertility Survey, we examine the trend andsocial correlates of consanguineous marriage inIran. Based upon William Goode’s theory ofmodernisation, we hypothesise a declining trendin consanguineous marriage over time and nega-tive relationships between consanguinity andmeasures of social status. Contrary to our expec-tations, there was a modest increase in the pro-portion of marriages between cousins in Iranfrom the 1940s to 1970s. Results from multivari-ate logistic regressions, however, indicate thatmany of the measures of individual social status

had the expected negative relationships with con-

sanguinity. Overall, the results of this analysissuggest that forces ofmodernization may be slow-ly eroding the social bases of consanguinity,while the increased availability of cousins maylead to an increase in consanguinity in the nearterm.

The 20th century, especially since World War II,has witnessed a transformation of social and eco-nomic life in countries around the world. Thesechanges, loosely linked under the rubric of mod-ernization. are hypothesized to have a very strongimpact on patterns of marriage and family life.

Center for Studies in Demography and Ecology, Departmentof Sociology. DK-40, University of Washington, Seattle. WA98195.

Key Words: consanguineous marriage, cousin marriage, Iran.modernization.

Three decades ago, William Goode (1963) pre-sented an encompassing theory of the impact ofindustrialization, and modernization more gener-ally, on family structure and conjugal relation-ships. His basic idea is that the nuclear family,and a set of associated attributes, are more adapt-ed to the high levels of geographic and social mo-bility of industrial society than are traditional

family structures.While Goode’s thesis has been frequently

challenged, there is really no alternative theory offamily change in the modem world. Moreover,many of the hypotheses that were explicit or im-plicit in Goode’s theoretical formulation havefound empirical support (Lee, 1987; Thorton &Fricke, 1987). In particular, there is widespreadevidence in many developing countries of trendstoward later age at marriage and more autonomyin the choice of marital partners (Durch, 1980;Rindfuss & Morgan. 1983). One of the less dis-cussed aspects of Goode’s theory was consan-guinity-the marriage between cousins or otherclose relatives. As with other aspects of tradition-al marriage patterns, Goode hypothesized a de-cline in consanguinity with modernization. In thisarticle, we examine the trend and social correlatesof consanguinity in Iran with retrospective datafrom the Iran Fertility Survey (IPS), a nationalfertility survey conducted in 1977.

DEFINING CONSANGUINITY

Consanguineous marriage appears as an esoteric

topic in research on marriage and kinship in West-

820 Journal of Marriage and the Family 56 (November 1994): 820-834

Consanguineous Marriage in Iran 821

em nations, but it is a central feature of familysystems in many pans of the world. A consan-guineous marriage is one in which the two part-ners have at least one ancestor in common, withthe ancestor being no more distant than a great-great grandparent (Hafez et al., 1983). For descen-dants who are of the same generation, a consan-guineous marriage would be between one personand a third cousin or a closer relative. Consan-guineous marriages are relatively common inmany areas of the world, including most countriesin the Middle East and South Asia. some countriesin sub-Saharan Africa, and (possibly) China (Bit-ties, Mason, Greene, & Rao, 1991). Because na-tional surveys of family and fertility rarely inquirewhether spouses were related prior to marriage,precise knowledge of the prevalence of consan-guinity is generally unavailable, but estimates ofconsanguinity from one-quarter to one-third of allmarriages are common in many countries.

The practice of consanguineous marriage is

generally considered to be an important issue inanthropological studies of family and kinship sys-tems (Goody, 1976; Holy, 1989; Levi-Strauss,1969; Murdock, 1949). A marriage betweencousins helps to conserve heritable assets withinthe extended family and also minimizes the riskof incompatibility between marriage partners andfamilies. While less studied in contemporary soci-ology, the question of consanguinity is relevant to

many aspects of comparative sociological theo-ries of family systems. The persistence of cousinmarriage and cultural preferences for marriagebetween cousins in many modernizing societiesraises questions about the impact of moderniza-tion on family structure. Perhaps the division be-tween traditional family forms and Goode’s mod-em "conjugal family" may not be as clear-cut asonce thought. The institution of consanguineousmarriage is linked to many other aspects of familylife, including the choice of marriage partner,marital relations, and nepotistic behavior withinthe kin network. Empirical evidence on these is-sues is modest, and the sociological implicationsof consanguinity are rarely incorporated into dis-cussions and interpretations of comparative mar-riage and family systems.

MODERNIZATION ANDCONSANGUINEOUS MARRIAGE

Goode (1963) presented a clear statement of theimpact of modernization on the incidence of con-sanguineous marriage as a consequence of the de-

cline of the institution of arranged marriage.Goode reasoned that in the Middle East, sub-Sa-haran Africa, India, and China, the incidence ofcousin marriage should decrease as individualsbegin to enjoy greater freedom in the process ofmate selection. In modernizing societies, individ-uals are no longer restricted to the choices offeredby an arranged marriage. Goode summarizes hisargument as follows: "It can be predicted that anymovement toward a conjugal family and freedomof courtship is likely to reduce the percentage ofmarriages between close relatives. The range ofpossible eligibles becomes much wider, and freersocial intercourse reduces the simple statisticalchance that a given individual will fall in lovewith and marry a close relative" (1963, pp.218-219).

The incidence of consanguineous marriageshould decrease as modernization progresses andan increasing number of individuals begin tofreely choose their mates from a broader pool ofpotential marriage partners. In his description ofchanging family systems in the Middle East,Goode observed that arranged marriage and theassociated system of paying a brideprice will dis-appear as the more educated groups begin to

adopt Western norms about freedom of matechoice. He also suggested that women with moreeducation and labor force participation will de-mand more freedom in the process of choosing amarriage partner. For Goode, then, the first indi-viduals to abandon arranged marriage will gener-ally be individuals with high social status-high-ly educated men and women and women in thelabor force.

The logic of Goode’s theory leads to clear em-pirical expectations as modernization spreads andan increasing number of individuals achieve high-er social status (more education, modern occupa-tional roles, etc.). Individuals with higher socialstatus will desire more freedom from traditionalsociety and will tend to marry later in life andhave fewer arranged marriages. As fewer individ-uals have arranged marriages, the theoretical ex-pectation is that the overall incidence of consan-guinity should decrease.

Several empirical studies have investigated thesocial correlates of consanguinity in the MiddleEast and South Asia and have reported findingsthat confirm Goode’s thesis. Rao and Inbaraj(1977), in a study of a district in Tamil Nadu,South India, found that urban women and men inhigh-status occupations were less likely to haveconsanguineous marriages than other women and

822 Journal of Marriage and the Family

men. Venugophal Rao and Murty (1984) foundnegative relationships between the incidence ofconsanguinity and both urban residence and so-cioeconomic status in the state of Andhra Pradeshin South India. In a study of consanguineous mar-riage in Saudi Arabia, Saedi-Wong, Al-Frayh,and Wong (1989) reported that men and womenin Riyadh with more education were less likely to

many their relatives than were persons with lesseducation. In a sample of households in Beirut,Khiat (1988) found that husband’s occupationalstatus had a negative relationship with consan-guinity.

CONSANGUINEOUS MARRIAGE IN IRAN

Historical sources indicate that marriage betweenclose relatives has been traditionally practiced inIran (Slotkin, 1949). Several anthropological ac-counts report a strong Iranian preference for mar-riage to first cousins. While Iranians observe theprohibitions against close relative marriage men-tioned in the Koran (e.g., a man’s mother, daugh-ters, sisters, aunts, nieces, and some of his wife’srelatives), a man is allowed to marry any of hisparallel or cross-cousins (Spooner, 1965; Tapper& Tapper, 1988). Unlike many other populationsin the Middle East, Iranians state no specific pref-erence for a man marrying his father’s brother’sdaughter (Tapper & Tapper, 1988). A general "bi-lateral" preference for marriage with close kin hasbeen observed in many regions of Iran (Barth,1961; Bradburd, 1984; Spooner, 1965; Tapper,1979).

Although based on very small samples, severalethnographic studies have reported the incidenceof consanguineous marriage among Iranian com-munities. Bradburd (1984) found that 37% of 108marriages among the Komachi nomads were con-sanguineous, with 29% of those marriages beingwith first cousins. Tapper (1979) found that 27%of 89 marriages among the Shahsevan nomadswere with first cousins, with an additional 31% ofall marriages being with distant agnates. Fischer(1978) found that 29% of the 381 marriages heinvestigated in the city of Yazd were consan-guineous, with 24% of all marriages being withfirst cousins.

Other studies of consanguineous marriage inIran have been based on larger samples. Naderi(1979), in a study of over 9,500 pregnancies inthe city of Shiraz, found that 24.5% of the birthsin his sample came from consanguineous unions.Behnam and Amani (1974), citing figures from

the 1966 census of Iran. reported that the inci-dence of consanguineous marriage ranged from25.1% in Tehran to 32.8% in some rural areas.Neither of these studies were able to examine re-lationships between socioeconomic characteris-tics and consanguinity.

MODERNIZATION IN IRAN

The development of the petroleum economy inIran led to a tremendous expansion of state rev-enues during the 1950s, 1960s, and 1970s-theera of the so-called "white revolution" or revolu-tion from above. While much of the new wealthgenerated by the rapid economic growth duringthis period went to unproductive uses, there wasalso a very substantial expansion of educationaland new occupational opportunities (see the timeseries statistics for Iran in Mitchell, 1982).

To provide a basic description of social changeand modernization in Iran in the decades leadingup to the late 1970s, Table shows the distribu-tions of educational attainment and other back-ground characteristics for several birth cohorts ofIranian men and women. These data are drawnfrom the 1976-77 Iranian Fertility Survey (IPS),the data source used in the subsequent analysis ofconsanguinity. The IFS samples of women andmen in these age groups (20-29, 30-39, and40-49 in 1977), however, are not fully representa-tive of successive birth cohorts in the Iranian pop-ulation. The IFS sample included only marriedwomen and their husbands. Furthermore, in orderto keep consistency with the subsequent analysis,we have restricted the sample to women who weremarried only once. However, the distributionsbased on the total IFS sample are almost identicalto the ones presented in Table 1. A more seriouslimitation is that unmarried persons are not includ-ed in the sample. For women, this is only a prob-lem for the youngest age group, where a smallsegment of the most modem women (those whohave delayed marriage for a few years) are missed.The sample of men is more biased, given that de-layed marriage is more common for men thanwomen. The computed mean years of schooling inTable is an underestimate of the true mean be-cause the top IFS educational category of 13 yearsof education or more was coded as 13.

Even with these limitations of data. Tableshows a modest trend toward modernization inIran. Although 70% of (married) women, betweenthe ages of 20 and 29, had no formal education in1977, this figure was a marked decline from those


TABLE 1. INDICATORS OF SOCIAL CHANGE AND MODERNIZATION IN IRAN BY AGE COHORT IN 1977

Wives

20-29 30-39 40-44

EducationNone ’-’;

1-5 years "-’6 years7-11 years12 or more years

Mean (in years)Urban originsWorked before first marriageWorked in nonagricultural jobs before first marriageWorked outside of the family before first marriage

EducationNone1-5 years6 years7-11 years12 or more years

Mean (in years)Urban origins

70%8975

2.1430%28%17%12%

20-29

45%192098

3.6828%

80%6544

1.4320%26%13%10%

Husbands

30-39

50%17156

13

3.6634%

88%54

2

0.7012%25%11%9%

40-49

67%121056

2.1629%

Note: Percentages rounded to whole numbers; some columns may not add to 100.

for women aged between 30 and 39 and 40 and49 (80% and 88%, respectively). The proportionof women with urban origins increased from 12%to 30% across the three cohorts. Intercohortchanges in the prevalence of employment prior tomarriage were minor, but the changes were in theexpected direction.

While the husbands between the ages of 30and 39 in 1977 in Table appear to be more"modem" (in terms of education and urban ori-gins) than husbands aged between 40 and 49, me20-29 age cohort appears to be more "traditional"than the 30-39 age cohort. The reason for thisanomaly is because the IPS sample of marriedmen aged from 20 to 29 is probably more ruraland less modem than the entire cohort of men atthis age (more’modem men in their twenties aremore likely to have postponed marriage).

In spite of these inconsistencies, our overallreading of Table 1 is of a slow but real trend to-wards modernization that should lead to a declinein traditional patterns of marriage and familystructure.

MODEL AND HYPOTHESES

Our adaption of Goode’s model of modernizationand consanguinity is presented in Figure 1. InGoode’s framework, arranged marriage is the keyvariable that mediates the effects of social changeand the social status of brides and grooms on con-sanguineous marriage. While arranged marriageis not measured in the IPS, there are many otherIPS variables that can be used to test hypothesesderived from Goode’s theory.

The key exogenous variable in this model isthe respondent’s year of (first) marriage, which isour proxy for the time path of modernization thatis hypothesized to condition all the other vari-ables in the model. The primary impact of mod-ernization is to raise the status of women andtheir spouses, which in turn should reduce con-sanguinity. Time is allowed to have a direct im-pact on consanguinity in addition to the indirectpaths via the characteristics of brides and grooms.The respondent’s background, represented bywoman’s origin, woman’s education, andwoman’s work before marriage, is hypothesizedto influence consanguinity directly and indirectlythrough her age at marriage.


FIGURE 1. HYPOTHESIZED MODEL OF TIME AND SOCIAL BACKGROUND ON CONSANGUINITY

Husband’sBackground

The wife’s attributes may also affect consan-guinity indirectly via her husband’s characteris-tics (the logic of putting the wife’s characteristicsprior to the husband’s is an analytical simplifica-tion of the assortative mating process). Husband’sbackground, represented by his origin, education,and occupation, is hypothesized to influence con-sanguinity directly and indirectly throughwoman’s age at marriage. More "modem" hus-bands may be more likely to marry older women.All of these variables are hypothesized to affectconsanguinity either directly or indirectly throughsubsequent variables.

The hypotheses in Figure can be stated asfollows:

1. Women marrying in later decades willhave a lower prevalence of consanguinitythan women marrying in earlier decades.

The IFS includes women who were first mar-ried from the late 1930s to the year of the survey(1977). Assuming a slow (and probably irregular)process of modernization, we would expect thelevel of consanguinity in Iran to decrease overtime as an increasing number of individualsachieve higher social status and (presumably)have fewer arranged marriages. Modernizationmay have a direct influence on the practice ofconsanguineous marriage in Iran through culturaldiffusion. The diffusion of ideas about individualautonomy and the freedom of mate choice mightbe measured by the direct impact of period (year

of marriage) on consanguinity, net of the effectsof the intervening social status variables.

2. Women with an urban origin will have alower incidence of consanguinity thanwomen with a rural origin.3. Women whose husbands have an urbanorigin will have a lower incidence of con-

sanguinity than those married to men with arural origin.

The effect of urban/rural origin on consan-guinity is hypothesized to operate primarilythrough origin’s effect on education. There is a

strong correlation between urban origin and edu-cation for both men and women in the IFS.

4. Women with more education will have alower incidence of consanguinity thanwomen with little or no education.5. Women married to men with more edu-cation will have a lower incidence of con-sanguinity than women whose husbandshave little or no education.

Iranian women who stay in school longershould be more likely to enter the labor force anddelay marriage. Women who delay marriageshould be less likely to have an arranged marriageand therefore less likely to have a consanguineousmarriage. Also, women who receive more educa-tion may receive more exposure to Western ideasabout freedom of mate choice and women’semancipation from family control.


Men with more education in Iran are also moreexposed to modern ideas about independent mate

selection and consequently may view arrangedmarriage as undesirable. Educated men may alsohave greater opportunities for interaction withwomen who are nonrelatives than do men withless education.’Al-Thakeb (1985) suggested thisinterpretation as the reason why well-educatedmen in Kuwait were less likely to marry relativesthan less educated men.

6. Women who worked before marriagewill have a lower incidence of consanguini-ty than women who did not work.7. Women married to men who work inhigh-status occupations will have a lowerincidence of consanguinity than those mar-ried to men who work in low-status occu-pations.

Women in Iran who worked before marriageshould be more likely to delay marriage and to

avoid having a groom chosen for them. ApplyingAl-Thakeb’s (1985) reasoning to women, it is

plausible that working women have greater expo-sure or contact with unrelated men than dowomen who do not work. Men who work in high-status occupations may marry at an older age andalso be exposed to high-status women who havealso postponed marriage.

8. Women who marry at later ages willhave a lower incidence of consanguinitythan women who marry at younger ages.

In the absence of a variable measuring an ar-ranged marriage, woman’s age at First marriagebecomes the key intervening variable predictingconsanguinity. Age at marriage should mediate at

least some of the effects of the other variables onconsanguinity (see Figure 1). A woman who mar-ries relatively late (in the Iranian context, thismeans age 20 or older) may be able to avoid hav-ing her marriage arranged. This should increaseher chances of marrying a nonrelative.

DATA AND MEASUREMENT

The 1976-77 Iran Fertility Survey (IPS) consistedof a household survey and a survey of individualever-married women aged from 15 to 50. Using asampling frame from the 1976 Census, IPS house-holds were selected using a multistage probabilitysample designed to represent the entire country.From the original listing of 6,056 living quarters.5,730 households were successfully interviewedafter three call-backs to each identified housins

unit. From the sample of IFS households, 4,932eligible respondents, defined as ever-marriedwomen between the ages of 15 and 50, were inter-viewed. Nonresponses for 42 cases rendered an ef-fective sample size of 4,890 ever-married women.For this analysis, the sample was limited to 4,667ever-married women who were married only once.

The IFS was pan of the broader World Fertili-ty Survey Project, with data collection conductedby the Statistical Center of Iran in 1977; analysisand publication of the IFS data were postponedbecause of the Iranian revolution in 1979. Afterthe IFS data were released to the Shiraz Universi-ty Population Center in 1984, a descriptive reportof fertility based on IFS data was published byHoma Agha (1984), with consultation by AkbarAghajanian. The Iran Statistical Center has alsopublished a preliminary report on the survey(1986). The current research project, of whichthis article is a part, is based upon the collabora-tion of the Iranian Statistical Center and AkbarAghajanian. The primary focus of the project isthe analysis of Iranian fertility trends with IFSdata (Aghajanian, 1992; Aghajanian & Gross,1991; Aghajanian, Gross, & Lewis, 1993;Raftery, Lewis, & Aghajanian, 1993; Raftery,Lewis, Aghajanian, & Kahn, 1992). The IFS isthe first nationally representative survey that con-tains detailed information on fertility, contracep-tive knowledge and use, socioeconomic back-ground, and consanguineous marriage for womenin Iran. More recent national data on fertility andfamily patterns in Iran are not available.

The dependent variable of consanguinity isbased on a question in the individual IFS ques-tionnaire that asked the respondent if she was arelative of her current or most recent husband.The response categories to this question were:"near relative," "far relative," or "no relation."The Farsi terms used for the response categoriesin the original questionnaire have similar mean-ings to the English translations used here. Basedon discussions with knowledgeable experts, wethink that urban respondents probably interpreted"near relative" as being a first cousin and "far rel-ative" as being a second or third cousin or otherrelative. In the rural areas of Iran where kinshipterms, such as cousin, are often used as terms ofaddress, it is more difficult to assume a commonunderstanding of the degree of relatedness im-plied by near and far relatives.

To explore the meaning of "near" and "far"relative of the respondent’s spouse, we cross-tab-ulated the dimension of relatedness (near relative,


far relative, no relation) by many other variablesin the data set. Overall, almost 40% of respon-dents in the IFS sample reported being married to

a relative, with about 24% to a near relative and15% to a far relative. Although the overall levelof consanguinity (the sum of the two categories)varied across levels of other variables, the divi-sion between near and far relatives was fairlyconsistent across subgroups of the population(roughly a 60-40 split). Accordingly, we groupedthe two categories of near and far together in adummy variable with all consanguineous mar-riages coded and all nonconsanguineous mar-riages coded 0. We use the term consanguinity in-terchangeably with cousin marriage.

To measure the trend in marital patterns, thecalendar year of the respondent’s first marriagewas constructed from several questions in the in-dividual questionnaire that measured the timingof first marriage of the respondent. Because onlyrespondents age 50 and below are included in the1977 IPS sample, the estimates of marriages forprior years are based on a restricted age range.For example, the IFS sample of women marriedin year 1950 are limited to those married at age23 or younger (50-[1977-1950]). Since most Ira-nian marriages are contracted before age 20, wefeel that the IFS can yield fairly reliable estimatesof consanguinity from the late 1940s to themid-1970s.

For women who were married more than once,the IFS has information on the current or most re-cent marriage but no information on the consan-guinity (and other characteristics) of the first hus-band. Women who were married more than onceare excluded from the following analyses becausemany of our independent variables relate to thecircumstances of the first marriage (year of firstmarriage, woman’s work before first marriage,and woman’s age at first marriage). Excludingthese women does not significantly reduce thesample size, since 94.7% of ever-married womenin the IFS were only married once at the time ofthe survey (N 4,667 women). The small numberof formerly married women (widowed or di-vorced) who were married only once are includedin the analysis. The level of consanguinity amongwomen in second or higher order marriages (whoare excluded from this analysis) is much lowerthan the sample of women married only once (thenumbers are 20% and 40%, respectively). Perhapsthe lower incidence of consanguinity among high-er order marriages is because such marriages areless likely to be arranged (Huzayyin, 1976).

THE TREND IN CONSANGUINITY

The first step of our analysis is to examine thetrend in consanguinity by respondent’s year offirst marriage. Figure 2 shows a simple 3-yearmoving average of the incidence of consanguinity(shown as the percentage of all marriages in theyear) from 1946 to 1976. As noted earlier, theIFS-based estimates of consanguinity for earlieryears are selective of young marriages (e.g., onlymarriages to women, age 20 or below, are mea-sured for 1946).

In spite of the selectivity of the sample for theearlier years in the series, the overall upward driftevident for most years from 1946 to 1976 is not

an artifact of measurement. The upper cutoff ofage 20 in 1946 only excluded a small handful ofwomen married in that year. The one subsequentperiod with little overall change was the 1960s. Incontrast, there were modest (about 5 percentagepoints) but real rises in consanguinity during the1950s and the 1970s. This finding is opposite towhat might be expected if modernization, repre-senting the central force of change in Goode’s hy-pothesis, were the dominant influence on chang-ing marriage and family patterns. In the balanceof this article, we explore patterns of consanguini-ty across social groups in Iranian society and in-quire whether the observed trend may be a func-tion of the changing socioeconomic characteris-tics of brides and grooms.

THE SOCIAL CORRELATES OF CONSANGUINITY

Table 2 shows the levels and trends in consan-guinity for four background characteristics of therespondents in the IFS: the woman’s origin, hereducation, her work experience before first mar-

riage, and her age at first marriage. The percent-age of consanguineous marriages is presented foreach category of the four independent variablesfor the total sample and by four marriage cohorts:the 1940s and before, the 1950s, the 1960s, andthe 1970s. These time periods are identified bythe respondent’s year of marriage. Sample sizesfor each category are reported in parentheses be-side each percentage.

The incidence of consanguinity is lower forwomen who grew up in urban areas relative tothose who were reared in rural areas (providingpositive support for hypothesis 2). Consanguinityhas increased across marriage cohorts for bothorigin groups, but the increase is greater for therural-origin women. The same general pattern of


FIGURE 2. TREND (3-YEAR MOVING AVERAGE) OF CONSANGUINEOUS MARRIAGE IN IRAN

40

30

20

10

01946 1950 1960

Year of Marriage

1970 1976

an inverse relationship between women’s statusand consanguinity holds for the other three vari-ables in Table 2 (hypotheses 4, 6, and 8 are sup-ported).

Relatively few Iranian women have any for-mal schooling, and the deficit is greatest for theoldest age group who were married in the 1940s.Women with some schooling were less likely tomarry their cousins than women without anyschooling. There has been a general increase inconsanguinity over time for women with noschooling and’for those with to 5 years ofschooling. For the relatively small segment ofwomen with’6’or more years of schooling, therehas not been an increase in consanguinity.

The work status before marriage variableshows that it is not formal employment, but thenature of work that makes a difference. Womenwho worked in agriculture or in production jobs("blue collar") were just as likely as women whodid not work to marry their cousins. For thesethree categories, there has been an increase (albeit

uneven) in consanguinity across marriage cohorts.The most common nonagricultural employmentfor Iranian women is carpet weaving. Womentypically work in their own homes in rural areasin quite traditional settings. It is only for the verysmall number of women working in- white collar-or service occupations ("modem sector" employ-ment) before marriage that the incidence of con-sanguinity decreases over time (only the last twodecades have a sufficient number of cases to re-port figures).

The incidence of consanguinity increases overtime for women who married at age 20 or earlier,while the reverse is true for the small numbers ofwomen who married at age 21 or older. In the1970s, the level of consanguinity among womenwho manied at age 21 or above was only half ofthe comparable figure for women who married atyounger ages.

The patterns of consanguinity by the charac-teristics of women generally support Goode’s the-sis that modem social roles lead to a decrease in


TABLE 2. TRENDS IN CONSANGUINITY BY CHARACTERISTICS OF WOMEN

OriginUrbanRural

EducationNo education1-5 years6 or more years

Work before marriageDid not workWhite collarAgricultureBlue collar

Age at first marriageLess than 1516-2021 or above

Total

Tot

MarriedRelative

34%42

423633

40224144

414126

40

al

1.0333.599

3.601327706

3.359138544567

2.5261,683338

4.667

1940searii

MarriedRelative

29%35

352831

34

3439

3433

34

andier

58509

5093226

4412

7149

4291353

1,323

195

MarriedRelative

27%39

392429

38

4032

39383238

Os

n

1511,031

1.0315992

89615

152115

67543773

1,468

196’

MarriedRelative

36%43

433635

41334244

44402641

Os

3641,103

1,104105257

1.09739159161

842510116

1,185

1970;

MarriedRelative

35%48

484332

45144449

474623

44

445872

873122325

84881153231

576601146

567

Note: n indicates total number-both married to relative and not married to relative-in subsample by characteristic.Women who married earlier than age 9 are excluded from this table; values for cells with less than 20 cases are not reported.

consanguinity. The only anomaly is the rise inconsanguinity from the 1950s to the 1970s forwomen in the low- and medium-status categories.The socioeconomic correlates of men and consan-guinity, as shown in Table 3, show a more com-plex pattern.

In Table 3, the incidence of consanguinity ispresented by husband’s origin, education, current

occupation, and age at current marriage. Note thatthe unit of analysis remains the women respon-dents (the sample of men is composed of hus-bands currently married to women from age 15 to

50). The last two independent variables in Table 3are not exact parallels of the respondent’s charac-teristics. While husband’s current occupation istemporally subsequent to marriage, current posi-tion is probably a good proxy for his occupationalstatus at the time of marriage (assuming relativelylittle occupational mobility for men). Becausedata were not collected on husband’s marital his-tory, we only have husband’s age at current mar-riage, not necessarily his first marriage.

The incidence of consanguinity is much lowerfor women with urban-origin husbands (confirm-ing hypothesis 3). The incidence of consanguinityfor the urban-origin sample of husbands has re-mained roughly stable over time, while the inci-dence of consanguinity for women who marriedmen with rural origins has risen steadily over thedecades.

For husband’s education, the expected inverserelationship is not found (contradicting hypothe-sis 5). The incidence of consanguinity is roughlythe same for all three levels of husband’s educa-tion and there have been increases over time forall levels of husband’s education. This unexpect-ed pattern is addressed more fully in the multi-variate analysis section and in the concluding sec-tion of the article.

The expected relationship between higher so-cial status and lower consanguinity is evident forhusband’s occupation. There are much.higher lev-els of cousin marriage for women married to menin agricultural and "blue collar" occupations (pro-duction, transport, and unskilled labor). While theoverall level of consanguinity is lower for womenmarried to men in higher status occupations (pro-fessional, administrative, clerical, and sales), thetrend is not linear. For some white collar occupa-tions, there was a rise in consanguinity in the1960s (relative to the 1950s) and then a decreasein the 1970s. There were steady increases in con-sanguinity across all decades for women marriedto men in the lower occupational ranks (agricul-ture and unskilled labor). The incidence of con-sanguinity is lower for women whose husbandsmarried at 26 or older. There are fluctuations inconsanguinity over time across husband’s age cat-egories that defy a simple interpretation.


TABLE 3. TRENDS IN CONSANGUINITY BY CHARACTERISTICS OF HUSBANDS

OriginUrbanRural

Total

MarriedRelative

31%44

n

1.3763.238

1940s aearlie;

MarriedRelative

28%36

nd

165399

1950s

MarriedRelative

31%40

n

311869

19(

MarriedRelative

33%45

i0

n

4481,011

19-

MarrieRelativ

31%50

TO;

de n

426889

EducationNo education1-5 years6 or more years

OccupationProf. & tech.Admin. & manag.ClericalSalesServiceAgricultureBlue collar

414238

35303132414540

2,753669

1,198

26291188399255

1.7331,607

343531

28293833

4305281

198146841

247160

394134

34282030434335

830152200

53254012561

532328

434237

40234137494540

843215403

96356111776

508539

474440

32332931385145

585232498

9321697871

402552

Age at current maniaLess than 1516-2021 or above

Total

ge444938

40

126879

3,478

4.667

473534

34

30137367

1.323

484436

38

40216911

1,468

445638

41

39232

1.1931,185

5341

44

17294

1,007

567

Note: indicates total number-both married to relative and not married to relative-in subsample by characteristic.Men who married earlier than age 9 are excluded from this table; values for cells with less than 20 cases are not reported.

If the women in the IPS sample are dividedalong the dimension of tradition and modernity, ageneral conclusion can be drawn from Tables 2and 3. From the 1940s to the mid-1970s, it ap-pears that the incidence of consanguinity in-creased for women in the more traditional cate-

gories, while consanguinity remained stable ordecreased for more modem women. A few excep-tions to this general trend can be found in Tables2 and 3. The two main exceptions are womenwith urban origins and women with highly edu-cated husbands. The incidence of.consanguinityfor these two groups of women has generally in-creased over time.

MULTIVARIATE ANALYSIS

We now turn to a multivariate analysis of the hy-potheses presented in Figure 1. With consanguini-ty measured as a dichotomous dependent variable

(nonconsanguineous marriage 0, consanguine-ous marriage 1), logistic regression is used to

estimate the impact of the modernization vari-ables on the natural log of the odds of a consan-guineous marriage (Morgan & Teachman, 1988)based on the IFS sample of 4,667 women whohave been married only once.

Prior to the regression analysis, all independentvariables were examined to determine if statisti-cally significant or substantively important nonlin-ear relationships with consanguinity were ob-served. This evaluation was based on the observedbivariate relationships between consanguinity andeach value of the independent variables. Based onthese preliminary analyses, each independent vari-able was receded to best represent the observednonlinear relationship. The final coding of hus-band’s education was based on an OLS regressionwhere consanguinity was the response variableand dummies for each year of husband’s educa-tion were included with all of the other indepen-dent variables. (The final codings for the indepen-dent variables, as well as means and standard de-viations, are available upon request.)A potential problem in multivariate analysis is

high intercorrelations among the independentvariables in the equations. Although the largesample size of the IFS increases the stability ofthe estimates even with highly intercorrelated in-dependent variables, we examined all bivariatecorrelation coefficients to detect potential prob-lems (results available from authors upon re-quest). The only potential problems are the inter-correlations of education and rural/urban originfor both husbands and wives (0.51 and 0.89, re-


spectively). Rural women and men were concen-trated in the lowest rungs of the educational clas-sification. We have examined the patterns of ef-fects from the multivariate equations estimatedhere and those from detailed cross-tabulations ofconsanguinity and these variables. The findingsare broadly similar, although there are some sub-tle interactions that are not evident in the multi-variate models. These variations are noted in theappropriate findings below.

The recursive model in Figure can be repre-sented by a set of four structural (logistic regres-sion) equations. The first model includes onlyyear of marriage, which is our proxy for modern-ization. The total effect of each independent vari-able on the logit of the probability of consanguin-ity can be decomposed into direct and indirect ef-fects (Raftery & Hout, 1985). Here, our interest isin how the effect of modernization is mediatedthrough wife’s characteristics (Model 2), hus-band’s characteristics (Model 3), and the wife’sage at marriage (Model 4).

In addition to the coefficients. Table 4 presentsthe L-square (the deviance) and the model chi-square for each of the logistic regression equa-tions. The L-square is analogous to the residual

sum of squares in OLS regression, while themodel chi-square reflects the improvement theequation makes over a model with no predictors(Morgan & Teachman, 1988; Weisburg, 1985). Ineach equation, dummy variables were included torepresent the missing data categories for each in-dependent variable. Each "missing data" dummyvariable was set to 0 when the relevant indepen-dent variable had a nonmissing value, and set towhen the relevant independent variable had amissing value. The coefficients for themissing-variable dummies are not reported inTable 4. The dashes in cells of Table 4 representthe omitted category for a set of nominal cate-

gories (the coefficients for the other categories inthe set are deviations from the omitted category).

The coefficient for trend (based on thewoman’s year of marriage) is positive and signifi-cant in all the equations. This result confirms thefinding from Figure that the incidence of con-sanguinity has increased over time in Iran (exceptfor a few small modem groups; see Tables 2 and3). Interestingly, the impact of modernization onconsanguinity is largely direct. Goode’s theory isthat modernization raises the status of the popula-tion which then leads to a decline in arranged

TABLE 4. ESTIMATED COEFFICIENTS FROM LOGISTIC REGRESSION MODELS

Independent Variables

InterceptWoman’s year of marriageWoman’s origin (urban)Woman’s education0-3 years4-8 years9-11 years12 or more years

Woman’s work before marriageWhite collarBlue collarAgricultureDid not work

Husband’s origin (urban)Husband’s educationNo education1-6 years7-11 years12 or more years

Husband’s occupationWhite collarBlue collarAgriculture

Woman’s age at marriage (age S 21)L-squareModel chi-square

*Chi-square significant at the .05 level. Chi-square significant at the .01 level.

Model

-.7512**

.0139**

6241.42

21.53**

Model 2

-.7666**.0187**

-.0935

-.3713.0524

-.5400*

-.6475*-.0056-.0599

6184.34

78.61

Model 3

-5528**.0156**.0074

-.2728.1070

-.5226

-.6330*-.0457-.1724

-.4617**

.1730*

.1814

.5850**

-.3533**-.1274

6129.52133.44**

Model 4

-5565-*.1068**

-.0059

-.2716.1066

-.4151

-.4445-.0395-.1667

-4578**

.1705*

.1897

.5982**

-3480**-.1197

-6005**6108.31

154.65**


marriages and to a decline of traditional familyforms such as consanguinity. Only the inclusionof husband’s characteristics in model 3 mediatesa fraction of the impact of year of marriage.

Net of other variables, respondent’s urban/

rural origin is not a significant predictor of con-sanguinity (the educational variable overlapsheavily with rural/urban origin). There is a signif-icant direct effect of the highest education catego-ry on lowered consanguinity (there are only non-significant differences across the lower levels ofeducation). The impact of the highest level of fe-male education parallels the effect of womenworking in high-status white collar occupations.About one-third of the impact of both high-statusroles (education and premarital work experience)is mediated by a later age at marriage, but the ma-jority of the impact of modem female roles onconsanguinity is direct.

The impact of husband’s statuses on consan-guinity is peculiar. As observed earlier in the bi-variate tables, the impact of urban origins and awhite collar occupation is negative. But the net

impact of husband’s education is positive, withhighly educated men more likely to be married to

cousins than men with lower levels of education.Finally, woman’s age at marriage has the expect-ed effect: Women who married after the age of 20are much less likely to marry a relative. Althoughwe have no direct evidence, we suspect that a

woman’s age at marriage is probably highly cor-related with whether the marriage was arranged.

One of the difficulties in interpreting the logis-tic regression coefficients is the coding of the de-pendent variable (the log of the odds ratio). Toprovide a more conventional expression of the co-efficients, Table 5 shows the impact of the inde-pendent variables in the Table 4 equations interms of the probability of a consanguineous mar-riage. The effects of a unit change in an indepen-dent variable in a logistic regression depends onthe value of the independent variable as well as

the other independent variable in the equation(see DeMaris. 1990). In Table 5, all of the inde-pendent variables are set near or at their averagevalues. The transformation of the coefficients wascomputed only for variables that are significant atthe .05 level or better in Table 4.

The year coefficient in the first model of Table5 shows that consanguinity rose by 3.4 percentagepoints every 10 years of the era (assuming amonotonic linear trend). This linear summary ofthe bivariate relationship is only an approxima-tion of the actual trend presented in Figure 2. Thereal value of the multivariate analysis, however,is to test how the increasing trend in consanguini-ty is mediated by the rising level of social andeconomic characteristics of brides and groomsover time. Model 2 (in Table 5) shows an increasein the year coefficient from Model 1. This meansthat the underlying rise in consanguinity is actual-ly greater than the observed trend. The modern-ization of the socioeconomic characteristics ofwomen in more recent marital cohorts has partial-ly suppressed the underlying trend of rising con-sanguinity. The same effect is evident with the in-clusion of female age at marriage in Model 4.What has been the impetus behind the rise in con-sanguinity over this period if rising levels of fe-male status have worked to lower consanguinity?A partial answer is given in Model 3 with the in-clusion of husband’s characteristics. The im-provement in the status of men (educational at-

tainment) does explain a significant part of therise in consanguinity over time, but most of therising trend in consanguinity is unrelated to anyof the variables in this analysis.

The importance of modem female roles in low-ering the probability of marrying a cousin is illus-trated with the total effects of 12 to 14 percentagepoints for higher education and premarital modemsector employment in Model 2 of Table 5. Thetransformed coefficients in Table 5 show thatwomen who marry men with 12 or more years of

TABLE 5. EFFECT OF A UNIT CHANGE W AN INDEPENDENT VARIABLE ON THE PROBABILITY OF CONSANGUINITY

Independent Variables

Woman’s year of marriageWoman’s education (12 or more years)Woman’s work before marriage (white collar)Husband’s originHusband’s education1-6 years12 or more years

Husband’s occupation (white collar)

Woman’s age at marriage

Model

.0034

Model 2

.0046-.1241

-.1462

Model 3

.0039

-.1485-.1106

.0432

.1446

-.0856

Model 4

.0042

-.1103

.0426

.1475

-.0847

-.1423


TABLE 6. CONSANGUINITY BY WIFE’S AGE AT FIRST MARRIAGE AND HUSBAND’S AGE AT CURRENT MARRIAGE

Husband’s Age at Current Marriage

Woman’s Age at

First Marriage

<1515-1718-2021-2526+

<15

MarriedRelative

50%52

50237

0

15-17

MarriedRelative

43%39

150661252

18-20

MarriedRelative

52%4844

n

34525473152

21-25

MarriedRelative

43%424526

n

695752270879

26-

MarriedRelative

35%35372519

^

n

54061530116148

Note: Percentages not reported for cells where < 20; women and men with age at marriage less than 9 are excludedfrom the table.

schooling are almost 15 percentage points morelikely to have a consanguineous marriage than menwith no education. This is in contrast to the bivari-ate finding of a weak relationship between hus-band’s education and consanguinity in Table 3. Acloser examination of detailed cross-tabulationsshowed that the positive impact of husband’s edu-cation on consanguinity is largely confined to therural sector (where the number of highly educatedmen is fairly few).

Another very important potential interaction isthe independent and joint effects of husband’sand wife’s characteristics on consanguinity. Theassumption of additivity in the multivariate modeldisguises the potential overlapping patterns of in-fluence. To illustrate this issue. Table 6 shows theincidence of consanguinity by husband’s age atcurrent marriage and wife’s age at first marriage.

In general, young brides and grooms are muchmore likely to marry their cousins. However,even for men and women marrying in the middleage range (18-20), the level of consanguinity isgenerally well over 40%. It is only for the oldestage categories (women above 21 and men above26) that the level of consanguinity declines. Formen who marry above age 26, there is a signifi-cantly lower level of consanguinity, regardless ofthe age of the bride. Perhaps the delay of mar-riage means selecting a spouse outside the tradi-tional kin networks. The delayed marriages mayalso be disproportionately composed of personswith the most modem characteristics. The oldercategory of men (above age 26) may include a

significant number of men who are entering theirsecond marriage, which may mean selecting a

spouse outside the kinship system. These specula-tions need to tested in future analyses.

CONCLUSIONS

The results of this study provide mixed supportfor Goode’s thesis that modernization and higher

social status will lead to a decline in consan-guineous marriage. Many of the key independentvariables, including all of the wife’s characteris-tics and two of the husband’s attributes (urbanorigins and occupation) have precisely the effectthat Goode predicted. Modernization, from thelate 1940s to the mid-1970s, has led to a gradualincrease in the proportions of women in Iran withmore education, working before marriage, and alater age at marriage. Women in these modemroles are less likely to marry their cousins thanare women in more traditional statuses.

However, there are several anomalies with re-sults in unexpected directions. In spite of the rapideconomic and social changes in Iranian societyfrom the 1950s to the 1970s, the overall incidenceof consanguinity appears to have increased overthis time period. More specifically, the incidenceappears to be increasing for the more traditionalgroups in society. The other unexpected finding isthat husband’s education seems to be only weaklyrelated to consanguinity (and positive in the ruralsector). This is in sharp contrast to wife’s educa-tion and husband’s urban/rural origins and occu-pation, for which higher status leads to lower con-sanguinity. These unexpected findings deservesome speculation even if we cannot provide satis-factory interpretations.

The time trend of consanguinity in other coun-ties is largely unknown. Khiat (1988) reportedthat consanguineous marriage has decreased overthe last several decades in Beirut, Lebanon.Khoury and Massad (1992), however, reported astable trend of consanguineous marriage in Jordanfrom 1900 to 1964. The trend in Iran may not beunusual, but more comparative data are needed toaddress the question.

One possible explanation for the rise in con-sanguinity, especially for rural and traditionalpopulations, may be increased availability of firstcousins and other relatives of mamagable ages.


The significant reduction of infant and child mor-tality in Iran from the 1940s to the 1960s certain-ly increased the pool of eligible cousins in thedecades prior to the 1977 IPS. In a context withconstant cultural preferences (cousin marriagebeing the ideal), the increasing availability ofcousins of marriageable ages would result in anincreasing incidence of consanguineous marriage.There may have been an increasing availability ofcousins in urban areas as well, but the effects ofavailability may have been offset by otherchanges in socioeconomic status and culturalchange resulting from modernization.

In response to an early draft of this article, ourcolleague, Akbar Aghajanian, suggested that cer-tain historical events in Iran may have been im-portant factors in increasing the desirability ofconsanguinity. The most important of such fac-tors was the land-reform program implemented inthe early 1960s (Lambton, 1969). Land reform al-lowed many landless families in rural areas ac-cess to small plots of land. These families mayhave used cousin marriage as a means to pooltheir newly acquired economic resources.

The lack of a negative impact of husband’s ed-ucation on consanguinity in Iran is unusual, butother studies in the Middle East do not reportconsistent results. Saedi-Wbng et al. (1989) founda negative relationship between husband’s educa-tion and consanguinity in Saudi Arabia. Hus-band’s education had no effect on consanguinityin Khiat’s (1988) study of Beirut. In a survey ofhouseholds in Kuwait, Al-Thakeb (1985) foundthat highly educated husbands were less likely tohave married their cousins than were men withless education. Khoury and Massad (1992)showed that a nonlinear relationship exists be-tween husband’s education and consanguinity inJordan. Men in Jordan with no education or a uni-versity education have the lowest incidence ofconsanguineous marriage, while men with prima-ry and secondary education have the highest inci-dence of consanguinity.

Men with higher education (as with women inthe same status) are likely to be exposed to theforces of modernization and have a wide pool ofeligible marriage partners to consider beyond theimmediate kin network. But families, as well asindividuals, have a strong stake in marriage selec-tion in Iranian society. A son with higher educa-tion may be a valuable family resource, and fami-lies may try to exercise greater control over the se-lection of his marriage partner (Khoury & Massad,1992). The families of educated men may use

cousin marriage as a way to maintain family prop-erty (Goody, 1976). This interpretation rests onthe assumption that families consider the selectionof spouses for highly educated sons differentlythan selection for highly educated daughters.

The empirical association between husband’sstatus and consanguinity in this analysis is alsoconfounded because of multicollinearity-thehigh correlations between two of the backgroundvariables. The positive effect of husband’s educa-tion on consanguinity (as reported in Table 4) islargely due to its joint association withrural/urban origins. There is a positive effect ofhusband’s education among men with rural ori-gins, but the number of highly educated men withrural origins is very few. Among men with urbanorigins, there is little effect of education on con-sanguinity. These complex patterns require moreanalysis than is possible here.

Finally, the rise in consanguinity and the an-omalous effect of husband’s education on theprobability of consanguinity raise questions aboutthe linear process of modernization and changingfamily structure. There are certainly systemicpressures in the directions that Goode predicted,but there may be other factors that have shaped theobserved trends. Forces of modernization in Iranduring this period may have been too tentative to

have had the impact that Goode predicted. Theimpact of the Iranian oil boom was evident in eco-nomic levels and government spending, but thechanges in the statuses of men and women overthis period were very modest. Perhaps a better testof Goode’s modernization thesis would occur in asociety that is undergoing a much greater shift insocial and economic levels of men and women. Afinal point to stress is that, during periods of socialchange, there may be differences between short-term and long-term responses in cultural normssuch as a preference for consanguinity.

NOTE

This research was supported by a grant from the Na-tional Institute of Child Health and Human Develop-ment (HD 26330, "A Socioeconomic Analysis of Fertil-ity in Iran," Charles Hirschman, Adrian Raftery, andAkbar Aghajanian, co-investigators). The authors grate-fully acknowledge the very helpful comments of AkbarAghajanian, Adrian Raftery, Steven Lewis, Pierre vanden Berghe, William Lavely, and the anonymous re-viewers on earlier versions of this paper. This paper ispart of a collaborative project of analysis of the IranFertility Survey between the Iranian Statistical Centerand Akbar Aghajanian.


REFERENCES

Agha, H. (1985). The study of fertility in Iran and its re-lationship with social and economic indicators: Ananalysis of data from the Iran fertility survey. Shiraz,Iran: Iran Statistical Center and Shiraz University.

Aghajanian, A. (1992). Population policy and contra-

ceptive use in Iran. Paper presented at the AnnualMeetings of the Southern Demographic Association.

Aghajanian, A., & Gross, A. B. (1991). Socioeconomicdeterminants of fertility in Iran. Paper presented atthe Annual Meetings of the Southern SociologicalSociety.

Aghajanian, A., Gross, A. B., & Lewis, S. M. (1993).Evaluation of Iran Fertility Survey (Working Paper93-2). Seattle: University of Washington, Center ForStudies in Demography and Ecology.

Al-Thakeb, F. (1985). The Arab family and modernity:Evidence from Kuwait. Current Anthropology, 26,575-580.

Banh, F. (1961). Nomads of South Persia. Boston: Lit-tle, Brown & Co.

Behnam, D., & Arnani, M. (1974). La population deL’lran. Paris: CICRED.

Bittles, A. H., Mason, W. M., Greene, J., & Rao, N. A.(1991). Reproductive behavior and health in consan-guineous marriages. Science, 252, 789-794.

Bradburd, D. (1984). The rules and the game: The prac-tice of marriage among the Komachi. American Eth-nologist. 11, 738-753.

DeMaris, A. (1990). Interpreting logistic regression re-sults: A critical commentary. Journal of Marriageand the Family, 52.271-277.

Durch, J. S. (1980). Nuptiality patterns in developingcountries: Implications for fertility (Reports on theWorld Fertility Survey No. 1). Washington, DC:Population Reference Bureau.

Fischer, M. (1978). On changing the concept and posi-tion of Persian women. In L. Beck & N. Keddie(Eds.), Women in the Muslim world (pp. 189-215).Cambridge, MA: Harvard University Press.

Goode, W. J. (1963). World revolution and family pat-terns. New York: Free Press.

Goody, J. (1976). Production and reproduction. Cam-bridge: Cambridge University Press.

Hafez, M., El-Tahan, H., Awadalla, M., El-Khayat, H.,Abdel-Gafar, A., & Ghoneim, M. (1983). Consan-guineous matings in the Egyptian population. Jour-nal of Medical Genetics, 20, 58-60.

Holy, L. (1989). Kinship, honour and solidarity: Cousinmarriage in the Middle East. New York: ManchesterUniversity Press.

Iran Statistical Center. (1986). Results of the Iran Fertil-ity Survey (Series No. 990). Tehran, Iran: Iran Statis-tical Center.

Khiat, M. (1988). Social correlates of consanguineousmarriages in Beirut: A population-based study.Human Biology, 60, 541-548.

Khoury, S. A., & Massad, D. (1992). Consanguineousmarriage in Jordan. American Journal of MedicalGenetics, 43,769-775.

Lambton, A. K. S. (1969). The Persian land reform,1962-66. Oxford: Clarendon.

Lee, G. (1987). Comparative perspectives. In M. B.Sussman & S. K. Steinmentz (Eds.), Handbook ofmarriage and the family (pp. 59-80). New York:Plenum.

Levi-Strauss, C. (1969). The elementary structures ofkinship (J. H. Bell, J. R. von Stunner, & R. Need-ham, Trans.). Boston: Beacon.

Mitchell, B. R. (1982). International historical statis-tics: Africa and Asia. New York: New York Univer-sity Press.

Morgan, S. P., & Teachman, J. D. (1988). Logistic re-gression: Description, examples, and comparisons.Journal of Marriage and the Family, 50, 929-936.

Murdock, G. P. (1949). Socia/ structure. New York:MacMillan.

Naderi, S. (1979). Congenital abnormalities in newbornsof consanguineous and nonconsanguineous parents.Obstetrics and Gynecology, 53, 195-199.

Raftery, A. E., & Hout, M. (1985). Does Irish educationapproach the meritocratic ideal? A logistic analysis.Economic and Social Review, 76, 115-140.

Raftery, A. E., Lewis, S. M.. & Aghajanian, A. (1993).Demand or ideation? Evidence from the Iranianmarital fertility decline. Unpublished paper. Univer-sity of Washington, Center for Studies in Demogra-phy and Ecology.

Raftery, A. E., Lewis, S. M., Aghajanian A., & Kahn,M. J. (1992). Event history modeling of World Fer-tility Survey data. Paper presented at the AnnualMeetings of the American Sociological Association.

Rao, P. S. S-, & Inbaraj, S. G. (1977). Inbreeding inTamil Nadu, South India. Social Biology, 24, 281-88.

Rindfiiss, R. R., & Morgan, S. P. (1983). Marriage, sex,and the first birth interval: The quiet revolution inAsia. Population and Development Review, 9, 259-278.

Saedi-Wong S., Al-Frayh, A. R., & Wong, H. Y. H.(1989). Socio-economic epidemiology of consan-guineous matings in the Saudi Arabian Population.Journal of Asian and African Studies, 24, 247-252.

Slotkin, J. S. (1947). On a possible lack of incest regula-tions in old Iran. American Anthropologist, 49, 612-617.

Spooner, B. (1965). Kinship and marriage in easternPersia. Sociologus (New Series), 15, 22-31.

Tapper, R. (1979). Pasture and politics. London: Aca-demic Press.

Tapper, R., & Tapper, N. (1988). Marriage, honour andresponsibility: Islamic and local models in theMediterranean and the Middle East. Paper presentedto the workshop on Islam Family Law in Theory andPractice, Cambridge, England.

Thonon, A., & Fricke, T. (1987). Social change and thefamily: Comparative perspectives from the West,China, and South Asia. Sociological Forum, 2, 746-779.

Venugophal Rao, V., & Muny, J. S. (1984). Selectionintensities and inbreeding among some caste groupsof Andhra Pradesh, India. Social Biology, 31, 114-119.

Weisburg, S. (1985). Applied linear regression (2nded.). New York: John Wiley.

modernization and consanguineousmarriage in iran pubs/a74.pdfmodernization. three decades ago,...

Documents