Date post: | 30-Sep-2016 |
Category: |
Documents |
Upload: | monica-boyd |
View: | 212 times |
Download: | 0 times |
DEMOGRAPHY© Volume 10, Number 1 February 1973
OCCUPATIONAL MOBILITY ~ND FERTILITYIN METROPOLITAN LATIN AMERICA
Monica BoydDepartment of Sociology and Anthropology, Carleton University, Ottawa,. CanadaKlS586
Abstract-Thc relation between career mobility and reproductive behavioris examined for five cities of developing Latin American nations: Bogota,Columbia; San Jose, Costa Rica; Mexico City, Mexico; Panama City,Panama; and Caracas, Venezuela. The data are obtained from fertilitysurveys eonducted between September, 1963, and August, 1964, in theabove-named cities under the auspices of the Centro Latinoamerieano deDemografia (CELADE) , and the analysis is based on information from600-800 women per city who have been married only once and marriedten years or more. Career mobility is defined as an occupational ehangeof the husband between the dateof marriage and 1963-1964, based on theHall-Jones occupational scale. Reproductive behavior is operationalized as the number of live births. The eonclusion of the multiple classification analysis is that the process of career mobility in four out offive Latin American samples is not a significant factor in explainingdifferential reproductive behavior. These results are compatible with avariety of North American studies which stress that variation heretoforeaseribed to the process of mobility is an additive composite of past andpresent status effects. These findings are discussed; a theoretical andmethodological critique of mobility-fertility research is presented.
Research into the relationship betweeninter- or intra-generational mobility andfertility is part of the larger investigation by social scientists into the socialcorrelates of mobility. Two models ofmobility effects exist. Proponents of theinteraetive model have argued that thedynamics of the mobility process exerta unique depressing or augmenting effecton completed family size. Others havecontended that reproductive behavior ofthe mobile is not uniquely influencedby the mobility experience but is anadditive composite of the effects ofclass of origin and class of destination.Mueh of the past research into eithermodel is based on data from industrialized nations. It is the purpose of thispaper to re-examine the nature of therelationship between fertility and oc-
1
cupational mobility in five metropolises of developing Latin American nations. Such a study not only advaneesgeneral knowledge concerning the relation of fertility and social mobility butalso documents similarities or differenceswith respect to correlates of mobilitywithin a comparative framework.
THEORETICAL MODEL
Although early studies interpreted therelationship between fertility and socialmobility in terms of inherited subfecundity by diverse social groups, thisperspective is neither adopted nor supported by later studies (see Blau andDuncan, 1967, p. 367). Rather, fertilitydifferences by mobility status have beenattributed to social-psychological factors. Based upon the assumptions that
2 DEMOGRAPHY, volume 10, number 1, February 1973
childbearing is inimical to social mobility and that fertility is inversely relatedto social class, upward social mobilityis hypothesized to be directly related tofertility planning and inversely relatedto size of planned family, both relationships persisting within otherwise homogeneous social groups (Westoff, 1953, p.31). A related interpretation (Tien,1961, p. 298) of the relation betweenmobility and reproductive behaviorssuggests:
1. Holding past social status constant,family size is inversely related to presentsocial status.
2. Holding present social status constant, family size is inversely related topast social status.
As discussed in the literature, the relationship between fertility and mobilityis hypothesized for either form of socialmobility, inter-generational or intragenerational. An examination of the literature further reveals that the processof social mobility often implicitly isconsidered to have a unique depressingor augmenting effect on fertility. Blauand Duncan (1967, pp. 415-416) suggest that social mobility may be disruptive of integrative social ties with theresult that mobile couples either restriettheir fertility or have more children asa means of reintegration. Besides therationale offered by Blau and Duncan,other considerations suggest why familysize variation is associated with theprocess of achieving a given social statusabove and beyond that of the consequences of past and present positions.The birth and raising of children requires time, money and energy, whichcould otherwise be spent in improvingone's social standing (Westoff et al.,1961, p. 237). Thus a pervasive successorientation and resultant mobility is associated with a reduction in fertility(Westoff, 1953, p. 30). Although thisargument most clearly applies to casesof upward mobility, it can be logicallyextended to instances of downward mo-
bility. Presumably couples experiencingdownward mobility are motivated eitherto regain their lost social position or tohalt their social demise, and they expressthis motivation by restricting theirfertility.
Although mobility-fertility studies predominantly test for reduced fertility ofmobile couples, further thought also suggests that fertility of the downwardlymobile also may be the result of the selectivity of downward mobility for highfertility couples. The latter must copewith large families and thus lack thetime, money and effort to maintain theirprevious social standing. Relatively highfertility of upwardly mobile couples ismore difficult to explain although aplausible argument would be that upwardly mobile persons have chosen toinvest their social gains in more children.Such an investment would be likelywhere mobile couples come from, or enterinto, reference and membership groupswhere high fertility is stressed as an important, if not conspicuously consumptive, value.
The above arguments are used to invoke a model of behavior in which theprocess of social mobility has a uniqueeffect of either increasing or depressingfertility. Since the resulting behaviorsuggests a statistical interaction betweensocial mobility and fertility, it is termedthe "interactive" model of mobility effects. This model is implicit in much ofthe research which categorizes coupleson the basis of their mobility experience(up, static, or down), and then attributesthe differences in fertility between categories to social mobility.
In contrast to this model, it is alsoargued that variations in fertility according to some category of mobilityexperience do not reflect the process ofmobility but only the additive influenceof past and present statuses. Averagefertility of the mobile couples lies somewhere between the averages observed fornon-mobile couples at status levels of
Occupational Mobility
origin and destination. Thus, the dataare seen to fit a model of additive effects.
The theoretical basis for this secondmodel is found in reference group theoryand in continuing adult socialization. Inan ex post facta assessment of the behavior of mobile individuals, Blau(1956) notes that socially mobile individuals have been subject to the socialpsychological forces of past and currentmembership and reference groups. Rethen postulates an intermediate level ofconformity by the socially mobile between the non-mobile groups at theorigin level and at the destination level:
... mobile persons are not weIl integratedin either socia! cIass. Without extensiveand intimate social contacts, they do nothave sufficient opportunity for completeacculturation to the values and style of lifeof one group, nor do they continue toexperience the full impact of the socialconstraints of the other. But both groupsexert some influence over mobile individuals since they have, or have had, socialcontact with members of both.... Hence,their behaviour is expected to be intermediate between that of the two nonmobile cIasses (Blau, 1956, p. 291).
Antecedent support for this proposi-tion may be found in Berent's (1952)analysis of 1949 British data. Whensocial origin is held constant, those couples who had experienced upward socialmobility had the smallest families on theaverage, those who had moved down hadthe largest, and those with static families had on the average an intermediatesize. Conversely, when persons of thesame current occupational status werecompared, those who had moved up hadlarger families than those who had remained static or had moved down. Berent interprets these different findings asreflecting both the acquisition of the fertility characteristics of the class of destination by the mobile persons and themaintenance of the family building habits of the class into which they wereborn.
3
As an extension of the arguments advanced by Blau (1956) and the researehconducted by Berent (1952), Duncan(1966, pp. 90-95) has re-examined Berent's study on occupational mobilityand fertility in England, using a modelwhich assumes only additive effects oforigin and destination elasses on fertility. Data from the American 1962Occupational Changes in a Generation(OCG) survey have also been examined(Blau and Duncan, 1967, pp. 367-399).Both studies conclude that there is nomobility effect underlying the fertilityof mobile couples other than the additivecombination of the fertility patterns ofthe two origin and destination occupational levels. Similar conclusions arereached with respect to occupational mobility and membership in voluntary associations (Vorwaller, 1970), kinship involvement (Aiken and Goldberg, 1969),work satisfaction (Laslett, 1971), andprejudice (Hodge and Treiman, 1966).
Although the above investigationshave confirmed the fit of data to an additive model of past and present statuseffects, there is a growing discussionasto whether the findings necessarily represent confirmation of the underlyingbody of theory. Methodologically, thereis the possibility that mobility effectsexist which cancel out and leave a zerointeraction term (Blalock, 1967a, 1967b;Hope, 1971). Hope suggests that whendistance moved and direction moved areincluded in a mobility model, mobilityeffects are observed in Berent's (1952)data, which were reanalyzed by Duncan(1966) and previously found to fit anadditive model. How serious is the failure to consider direction and distancemoved remains to be answered by futuremethodological replications of Hope's(1971) investigation. The mobility effects are small and, ifthe intent of aninvestigation is to develop a predictivemodel, the additive model is preferredbecause of its simplicity (Blalock,1967b).
Prior to the 1960's, social scientistsinvestigating the relation between mobility and fertility generally found differential reproductive experience by mobility status. The explanation given wasalmost always that based on the interactive model, namely, that because of thehigh costs of producing and raisingchildren andJor because of the disruptionof integrative social ties, social mobilityhad a unique effect upon fertility. Thisorientation may be seen in the earlyinterpretations of inter-generational mobility studies conducted by Baltzell(1953) in the United States and by Bresard and Girard (see Scott, 1958; Westoff, 1953) in France. This approach isalso implicit in the examination ofinter-generational mobility and fertilityby Kantner and Kiser (1954) and intragenerational mobility and fertility byRiemer and Kiser (1954) based on datafrom the Indianapolis survey, Hutchinson's (1961) investigation of inter-generationalrriobility and fertility in Brazilis the most recent study which finds
SoCIAL MOBILITY
FERTILITY RESEARCH
4 DEMOGRAPHY, volume 10, number 1, February 1973
Related to the above concern is the significant variation in mean family sizemobility-fertility-specific argument that by mobility experience categories.the processes which augment or decrease A partial replication of the Indianapfertility according to the interactive olis Study, the Family Growth in Metromodel may operate simultaneously, and, politan America studies (Featherman,in cancelling each other out, confirm the 1970; Westoff et al., 1963; Westoff et al.,additive model. However, with one ex- 1961), also tests the general hypothesisception (Riemer and Kiser, 1954, pp, that mobility experience is inversely re19~195) the assumption of most studies, lated to size of family and directlyincluding this one, is that mobility has related tofamily planning success withinthe effect of either increasing or decreas- otherwise homogeneous categories. Ining the aggregate fertility of mobile this case, results concerning assoeiationcouples. The latter argument clearly is between mobility aspirations and expericongruent with a macro-sociological ori- ence and reproductive behavior are inentation toward prediction although in- conclusive. Similar inconclusive findingsvestigations into causa I mechanisms of with respect to the relationship betweenmobility clearly are needed to provide reproductive behavior and social mobilgreater specification of the theoretical ity in Australia, Great Britain andframeworks utilized in mobility-fertility the United States are also noted byresearch (Blalock, 1967a; Miller, 1971). Boggs (1957), Goldberg (1959), Perrucci
(1967), Scott (1958), and Tien (1965).Excluding Goldberg's (1959)' investigation, which is based on the Detroit AreaStudy, the remaining four studies are allbased upon various professional occupational groups.
A plausible explanation for the inconclusive findings of recent social mobilityfertility investigations is that such findings reflect the societal conditions withinwhich mobility and reproductive behavior occur. All of the above studies arebased on data collected in developed societies, which are characterized by"fluid" or "open" stratification systemsand by shrinking variability in fertilityby socioeconomic status. The literature(Germani, 1966; Kessin, 1971; Treiman, 1970) suggests that mobility inthese modern societies may not necessarily have disruptive effects. Insuch societies, mobility rates are high,and mobility becomes institutionalizedthrough participation in the educationalsystem. Since mobility is no longer ascostly, nor the exception, it is not associated with drastic behavioral modification on the part of individuals concerned. Support for this contention isgiven by several investigators of social
OccupationaI Mobility
mobility and fertility. Boggs (1957) argues that to the extent that social mobility has become an accepted and expectedpart of an industrial society it no Iongerelicits special reproductive behaviorfrom the individuals involved. Both Tien(1965) and Perrucci (1967) stress theimportance of the institutionalization ofmobility through educational attainmentwhich in turn acts as a socializing agent.Tien's (1965) study of Australian professors reveals that the occupationalachievements of the respondents are related to their educational attainmentsand that marriage was postponed untilafter the attainment of professionalqualification andjor until after the actual commencement of the occupationalcareer. Thus the struggle to improve ormaintain status is virtually over beforethe university professors assume anyfamily responsibility of their own. Subsequent fertility performance only reflects present membership in the university profession (Tien, 1965, p. 162). Inher study of career and inter-generational occupational mobility of American engineers, Perrucci (1967) alsonotes that inter-generational mobility isachieved largely through education andgenerally precedes reproductive behavior.
In addition to the "open" stratificationsystem with its resultant implicationsfor mobility, another feature of modernsocieties is the increasing lack of variability in fertility by socioeconomic indicators. This lack of variation partlyreflects the nature of the stratificationsystem. Treiman (1970, p. 226) notesthat one of the consequences of a highrate of mobility is increasing behavioralheterogeneity within stratum and decreasing between-strata differences. Another factor is the widespread use ofcontraceptives in industrialized nationsand the resulting relative homogeneity offertility performance (Tien, 1965).
In contrast, developing or industrializing societies display not only higher
5
levels of fertility, but also differentialfertility by socioeconomic criteria. Suchfertility behavior reflects both higherfamily size norms and differential accessto, and use of, contraceptives as compared with modern nations. It also mayreflect high degrees of discontinuity between strata in the form of clear cleavages and gross behavioral differencesbetween strata (see Germani, 1966).
The importance for reproductive behavior of the stratifieation system indeveloping soeieties is emphasized further by the suggestion that the disruptive eonsequenees of mobility are mostlikely to be found in traditional, static,and elass-homogeneous soeieties, whiehare charaeterized by low rates of personal mobility (Germani, 1966; Kessin,1971; Treiman, 1970). In sueh soeieties,mobility mechanisms are not institutionalized by edueation, and individual mobility is more the exeeption than therule. Germani (1966) suggests that thesoeially mobile individuals consequentlyreeeive little soeial support and that considerable strain will be indueed whiehin turn is assoeiated with behavioralmodifieation. A slight rewording of Treiman's (1970) hypothesis coneerning theindividual consequenees of mobilitysummarizes the argument: the lower therate of soeial mobility of a soeiety, thegreater the likelihood that sociallymobile individuals will experienee disruptive strain resulting in pathologiealresponses; that is, the greater the likelihood of interaction effects (Treiman,1970, p. 228).
An examination of the relation between mobility and fertility using datafrom developing nations thus appearsstrategie. This eontention gains furthersupport in view of the failure of mostinvestigators of mobility-fertility to rigorously test for the existenee of the interaetive model of behavior in whichmobility is seen to exert a unique effeetupon fertility above and beyond thatattributable to past and present soeial
6 DEMOGRAPHY, volume 10, number 1, February 1973
status. Not until Duncan's (1966) reexamination of Berent's data was thereany attempt to increase precision in theexplanation of differential reproductivebehavior by mobility experience categories. For the most part, in subsequentinvestigations into the effect of mobilityon other dependent variables such asmembership in voluntary associations,rates of kinship visiting, work satisfaction, and prejudice, the data are observed to fit an additive model of mobility effects in which variations in thedependent variable are a function of pastand present occupational class. However,these findings may reflect the nature ofthe stratification system and thus theconditions in which mobility occurs. Ifit is assumed that developing nations arecharacterized by lower rates of mobility,the additive model of status effects mayhold in modern industrialized nationsbut not in Third World nations.
With these observations in mind, thisstudy investigates the existence of differential fertility by mobility experiencecategories and examines the fit of thedata to an additive model of occupational effects using data from Bogota,Colombia; San Jose, Costa Rica; Mexico City, Mexico; Panama City, Panama; and Caracas, Venezuela.
METHODOLOGY AND
ANALYTICAL PROCEDURES
The data to be used in the investigation of relation between social mobilityand fertility are obtained from fertilitysurveys conducted between September,1963, and August, 1964, in tbe abovementioned cities under the auspices ofthe Centro Latinoamericano de Demografia (CELADE) , which is a regionalresearch and training center of theUnited Nations. In keeping with the previous social mobility-Iertility research(Hutebinson, 1961; Kantner and Kiser,1954; Riemer and Kiser, 1954; Ti,en,
1965), the sub-sample selected for analysis consists of currently married womenwho have been married once for tenyears or more. Such a selection permitsa partial control for dura tion of marriage with respect to fertility variationsand with respect to changes in socialstatus. (An analysis on . those womenwho are native-born currently married,and married only once, to native-bornspouses for ten or more years and defined as fecund may be found in Boyd[1971]. The results of this analysis ona more restricted sampie are consistentwith those reported in this paper.)
In this research, fertility is operationalized as the average number of livebirths as a rough approximation of completed family size. Movement betweensocial status positions is defined asmovement between occupational categories by the husband. While it would bedesirable to examine the relations between fertility and both intra- and intergenerational mobility, the high rate ofitem-nonresponse (18-37 percent) forthe husband's father's occupation prevents analysis for the latter type of mobility. Thus, the specific plan of theanalysis is to examine the relation between fertility and intra-generational(career) mobility of the husband. Thenecessary data are obtained fromCELADE survey questions which askedthe female respondent the main occupation of her husband at the time ofthe interview and his occupation atmarriage.
The source for the ranking of upward,downward, and static occupational mobility status is the Hall-Jones (1950)scale, which was used by CELADE incoding occupations. This prestige scale,which has been used by Glass and Hall(1954) in Britain, has been verified andused by Hutchinson (1957, 1961) inBrazil, and consists of the followingoccupational categories:
OccupationalGroups
Occupational Mobility
In keeping with the analysis conductedby Berent (1952) and areanalysis ofBerent's data by Duncan (1966) thefollowing analysis is conducted on thebasis of collapsed classification of thesecategories consisting of four occupational groups, detailed in the right sideof the above chart.
It should be noted that the use of occupation as a single index of socialstatus and change of social position isbased on the premise that social classis a general term which includes socioeconomic status as weIl as a way oflife. While other indices of social statuswould be desirable, there is a paucityof research concerning stratificationvariables in Latin America (Germani,1963; Iutaka, 1965), and occupationalposition remains one of the frequentlyused indices of social position in LatinAmerica.
The mobility model tested in this research is one which assumes that reproductive behavior for any combinationof past and present occupation is aweighted avetage of the additive effectsof the two origin and destination variables used to define mobility. The term"effect" is used to denote a relationshipbetween variables of occupation and fertility; it does not connote causality. Thestatistical technique for inquiring intothe effects of mobility is the multiple clas-
Occupational Status
1. Professional and high }administrative
2. Managerial and executive3. Inspectional supervisory 1
and other non-manual(higher grade)
4. Inspectional supervisory jand other non-manual
(Iower grade)5. Skilled manual and
routine non-manual6. Semi-skilled and
unskilled manual
I
II
III
IV
7
sification analysis (MCA) which is a formof dummy variable regression analysis.The statistical model of the multipleclassification analysis is based on aleastsquares approximation of a set of normalequations, and it assumes additivity ofrow and column effects. The procedureyields a set of expected values that reflect the net effect of a given social characteristic after the impact of otherspecified factors have been removedstatistically. In this analysis, adjustments are made for intercorrelation between past and present occupationalposition, and the hypothetical averagenumber of live births is calculated foreach cell (Andrews, Morgan, and Sonquist, 1967; Blau and Duncan, 1967;Duncan, 1966).
CAREER MOBILITY AND
NUMBER OF LIVE BIRTHS
The effects of husband's present oecupation in 1963-64 and occupation atmarriage on average number of livebirths are presented in Tables 1 and 2.The F-test for the multiple R, computedon the basis of dummy variable regression, indicates that the husband's pastand present occupational positions aresignificantly associated with variationsin reproductive behavior for all fiveLatin American cities considered (Table1, column 2). The multiple R (Table 1,column 1), however, also indicates therelative unimportance of the joint effectsof past and present occupational positionin accounting for variation in reproductive behavior. The explained varianceranges from 3.3 percent in Bogota, Colo~bia, to 7.5 percent in San Jose, CostaRlCa.
Table 1 also presents the solutionsto the additive multiple classificationmodel, discussed above. The model isrepresented by the equation:
f i ; = f + a, + b, + eii'
8 DEMOGRAPHY, volume 10, number I, February 1973
TABLE l.-Dummy Variable Multiple Regression Coefficients, F-Tests, and Net Effects of Husband's Occupation at Marriage and Husband's Current Occupation for Mean Number of Live
Births of Women in Five Latin American Cities
GrandMean MultipleNumber Regres- Net Effectsof Live sion Co- Past Occupation Present Occupation
City Births efficient F Test I I! II! IV I I! II! IV
Bogota 5.11 .1831 3.93a .20 -.11 -.13 .38 -1.13 .11 .10 .57San Jose 4.89 .2743 9.17b -1.15 -.45 .15 .90 -.20 -.42 .27 .33Mexico
City 5.26 .2557 9.60b -.46 -.37 -.02 .48 -.59 -.30 .22 .57Panama
City 3.95 .2012 3.74 a -.73 .14 -.02 .39 -.19 -.62 -.06 .81Caracas 4.44 .2116 4.73a -.22 -.06 .06 .62 -.57 -.59 .40 .20
a - p < .05h - P < .01
where Yij is the mean score for the cellcategory; Y is the grand mean for thetotal sample: a, is the effect on the wife'sfertility behavior due to her husband'smembership in the ith occupationalcategory at marriage; b, is the effect onwife's fertility behavior due to husband'smembership in the jth destination occupational category in 1963-64; and e., isthe difference between the observed andthe expected average number of livebirths, computed on the basis of theadditive effects of row and columncategories.
The net effects which are presented inTable 1 are the estimates of a, and bioThey are calculated from unstandardized dummy variable regression coefficients (Andrews, Morgan, and Sonquist,1967, pp. 115-117; Melicher, 1965, p.375), and represent the influence of pastand present occupational position onnumber of live births after the adjustments for the association between predictor variables (past and present occupation) have been made. They permitcomputation of the expected fertilityaverages which would be obtained on thebasis of an additive model of past andpresent occupational effects, according tothe following formula:
(NLB»,j = (NLB) + a, + bj.
Thus, the hypothetical or expected average number of live births for thosewomen in Bogota, Colombia, whose husbands held semi-skilled or unskilled jobsat marriage (occupational category IV)and skilled manual or routine manualjobs in 1963-64 (category 111) is calculated as follows:
((NLB»43 = 5.11 + .38 + .10
= 5.59.
The va lues calculated from the neteffects may be compared with the observed number of live births on a cell bycell basis, obtained by cross-classifyingpast and present occupational position ofthe husband (Table 2). As depicted inTable 2, cells below the major diagonalgive the actual average number of livebirths of women whose husbands haveexperienced upward career mobility fromthe time of marriage. Cells above themajor diagonal indicate the averagefertility of women married to men experiencing downward career mobility.Tbe deviations of the expected hypothetical average number of live birthsfrom the actual number also are presented in Table 2.
Where signs are negative, the observednumber of live births for a given cate-
Occupational Mobility 9
TABLE 2.-0bserved and Calculated Mean Nurnber of Live Births for Wornen in Five LatinArnerican Cities by Husband's Current Occupation and Occupation at Marriage
Husband 's Husband's Current OccupationOccupation Observed Mean Observed Minus
at Number of Live Births Ca1cu1ated MeanMarriage I II III IV Total I II III IV
Bogota, Co1ombiaI 4.08 8.50 5.75 0.00 4.28 -.11 3.07a .33a
(72) (2) (4) (0) (78)II 3.94 4.86 5.73 6.25 4.99 .07 -.25 .63 1.09a
(16) (97) (33) (4) (150)III 5.40 5.49 4.99 5.62 5.11 l.54a .39 -.10 .07
(5) (43) (265) (29) (342)IV 3.00 6.00 5.69 5.98 5.84 -1.36a .40a .10 -.08
(1) (4) (48) (64) (117)Total 4.12 5.12 5.16 5.89 5.11
(94) (146) (350) (97) (687)
San Jose, Costa RicaI 3.56 3.50 5.00 0.00 3.55 .01 .18a i.oi«
(50) (2) (1) (0) (53)II 4.00 3.95 6.00 5.50 4.12 -.25 -.08 l.29 .73a
(30) (133) (13) (2) (178)III 6.00 4.76 5.26 5.13 5.17 1.15a .13 -.05 -.24
(6) (71) (267) (13) (357)IV 0.00 5.50 6.00 6.14 6.04 •Ba -.05 -.03
(0) (6) (40) (49) (95)Total 3.88 4.26 5.38 5.92 4.89
(86) (212) (321) (64) (683)
Mexico City, MexicoI 4.10 6.00 12.00 0.00 4.23 -.11 1.50a 6.98a
(90) (2) (1) (0) (93)II 4.38 4.45 5.92 8.50 4.57 .08 -.14 .81 3.04a
(50) (142) (13) (2) (207)III 4.50 5.38 5.25 6.65 5.38 _.l6 a .43 -.22 .83
(6) (53) (193) (23) (275)IV 6.33 5.28 6.27 6.14 6.07 l.17a -.17 .30 -.18
(6) (32) (78) (139) (255)Total 4.30 4.79 5.59 6.24 5.26
(152) (229) (285) (164) (830)
Panama City, PanamaI 3.21 3.00 3.00 4.00 3.20 .18 .39a -.16a -.03a
(58) (3) (3) (1) (65)II 4.24 3.59 4.06 6.00 3.89 .34 .11 .03 i .io"
(17) (44) (18) (2) (81)III 3.88 3.67 4.03 4.83 4.07 .14 .36 .16 .09
(16) (15) (231) (24) (286)IV 3.00 2.83 3.78 4.49 4.13 _1.15a -.89a -.50 -.64
(6) (6) (27) (68) (107)Total 3.48 3.51 3.99 4.60 3.95
(97) (68) (279) (95) (539)
10 DEMOGRAPHY, volume 10, number 1, February 1973
TADLE 2.-0bserved and Calculated Mean Nurnber of Live Births for Wornen in Five LatinArnerican Cities by Husband's Current Occupation and Occupation at Marriage (Continued)
Husband's Husband's Current OccupationOccupation Observed Mean Observed Minus
at Number of Live Births Calculated MeanMarriage I II UI IV Total I II III IV
Caracas, VenezuelaI 3.76 1.50 4.00 2.00 3.68 .11 -2.13a -.62 a -2.42a
(68) (2) (2) (1) (73)II 3.58 3.73 6.00 4.33 3.89 -.23 -.06 1.21 -.25
(26) (90) (10) (3) (129)IU 3.50 4.08 4.72 4.82 4.63 -.3la .29 -.06 .24
(6) (40) (269) (22) (337)IV 0.00 3.50 5.67 5.22 5.30 -.98a .20 -.04
(0) (2) (21) (50) (73)Total 3.70 3.80 4.83 5.03 4.44
(100) (134) (302) (76) (612)
a - fewer than ten casesNote: Numbers in parentheses indicate number of cases.
Calculations based on net effects to the third decimal point.
gory is lower than predicted; where theyare positive, the observed fertility isgreater than that predicted on the basisof the additive model of occupationaleffects. Accordingly, these deviationsprovide a basis for testing the fit of thedata to the additive model. If the fertility of the mobile couples is not described by the additive model, the setof deviations will either overstate orunderstate the actual mean fertility formobility categories, and a patterning ofthese difference scores will be observedA danger with any test of the additivemodel is that the statistical assumptionsof normality and constant variance ofthe means are violated. Logarithmictransformations were performed. Sincethe conclusions were similar, the analysisbelow is based on untransformed numberof live births.
As indicated in Table 2, fertility in thefive Latin American cities clearly is inversely related to socioeconomic position. Table 2 also reveals a tendency ofthe additive model of occupational effects to underestimate the average number of live births for upwardly mobilecouples in Bogota, San Jose, and Mexico
City and for downwardly mobile coupIes in Bogota, San Jose, Mexico Cityand Panama City. For example, infive out of six upward mobility categories, wives in the Bogota sampie havea higher average number of live birthsthan predicted on the basis of the additive model. Conversely, for mobilecouples in Caracas, the additive modelslightly overestimates the average fertility.
The departure of the observed fromthe expected number of births also canbe assessed by the reduction of the datapresented in Table 2 according to thedirection of the mobility. The first panelof Table 3 presents the actual meannumber of live births by categories ofupward, static, and downward mobilityexperience. The second panel presentsthe values expected on the basis of anadditive model of past and present occupation effects; these values are generated by summing the expected meansmultiplied by the cell sizes and thendividing this estimated sum by the totalnumber in the group. The third panelshows the difference between the observed and the calculated means. The
Occupational Mobility 11
TADLE 3.-Collapsed Observed and Calculated Mean Nurnber of Live Births for Wornen inFive Latin Arnerican Cities by Husband's Career Occupational Mobility Experience
Husband's Mean Number of Live BirthsMobility Mexieo PanamaExperieuee Bogota San Jose City City Caraeas
Observed Mean Numberof Live Births
Up 5.35 5.01 5.45 3.75 4.24Statie 4.96 4.83 5.08 3.95 4.45Down 5.79 5.35 6.61 4.37 4.80
Calcu1ated MeanNumber of LiveBirths
Up 5.09 4.96 5.23 3.89 4.18Statie 5.09 4.87 5.26 3.92 4.48Down 5.33 4.88 5.49 4.26 4.58
Observed MinusCalcu1ated Meansa
Up .26 .05 .22 -.14 .06Statie -.13 -.04 -.17 .02 -.03Down .46 .68 1.12 .11 .22
Number of CasesUp 117 153 225 87 95Statie 498 499 564 401 477Down 72 31 41 51 40
a - Va1ues ea1eu1ated to three p1aces and may differ slight1y fromva1ues based on panels (1) and (2).
subsequent panel shows the number ofcases in each mobility category.
Table 3 shows that in four out of fivecities the additive model underestimatesthe fertility of the mobile couples andoverestimates that of the static couples.The mean number of live births by aggregate mobility categories most closelyapproximates the additive model for thePanama City and Caracas samples, Deviations of the calculated mean from theobserved tend to be higher for the remaining samples. On the whole, thelargest deviations appear in cases ofdownward mobility.
The differences between the observedand expected mean fertility in Tables 2
and 3 may be statistically assessed bythe respective use of the F-test for interaction and the t-test for differences between means. The latter test, however,is extremely conservative (Treiman,1965; Vorwaller, 1970). While it wouldbe desirable to use the F-test for interaction in a two-way analysis of varianceto ascertain if the deviations from theadditive model reflect sampling variability or support the interactive model, theoccurrence of empty cells prevents itsuse. Given the small cell frequencies forsome of the mobility categories, it further is difficult to make conclusions as tothe fit of the data in Table 2 to the additive model of occupational effects. For
DISCUSSION
Given the above findings, what comments can be made with respect to theeffects of career occupational mobility indiverse cultural settings? One conclusionis that the general association betweenmobility and fertility is similar acrossdifferent societies. An explanation for thefindings is found in what Moore (1965,p. 11) refers to as the "theory of structural constraints." His argument is thatan industrial system imposes certainorganizational and institutional requirements not only on the economy but alsoon many other aspects of society, including the behavior of its members. Thus,economic development is associated withthe processes of urbanization, industrialization, development of more complexand scientific technology, rising standards of living, increasing literacy, andincreasing real and perceived opportunity for advancement. The expansion ofthe mass media communication systemand the increase in transportation facilities link groups and individuals into
12 DEMOGRAPHY, volume 10, number 1, February 1973
the most part, the largest deviations are model. The static, or no mobility, cateassociated with cells which contain a gory was omitted to make the equationsmall number of observations and where soluble although the multiple regressionthe sampling errors for these categories coefficient reflects the influence of thisare bound to be large. third category. See Suits, [1957].) The
The dearth of respondents clearly re- increase is significant at the .005 levelfleets both the stratified nature of the for the Mexico City sample and at thesample and the conceptualization of the .10 level for the Bogota sample, Thus, inmobility problem. The data are derived four out of five samples, the data are notfrom a representative probability cluster observed to fit the interactive modelsample of women in each of the five better than the additive model at the .05cities (Miro, 1966) with the result that level of significance. The hypothesis thatthe samples contain a Iarge number of variations in fertility by mobility exwomen whose husbands are in the lower perience can be understood in terms ofoccupational categories. The conceptual- past and present statuses is supportedization of mobility as a deviant case for the Bogota, San Jose, Panama City,also compounds the problem of thinly and Caracas samples, The findings forpopulated cells. Use of a cross-classifled the Mexico City sample suggest the extable means that mobility is treated as istence of a mobility effect with rean exception to the rule; if the mobility spect to fertility variations beyendexperience is a one-step process, rather that attributable to past and presentthan two or three, then it is not unlikely occupations.to find that cells indicating mobilitybetween low strata and high strata arein fact thinly populated.
In spite of the existence of empty cellsand/or cells with a small number ofrespondents, the fit of data to the additive model can be assessed by testingthe fit of the data to the alternative interactive model. The differences betweenthe two models in the proportion of explained variance is statistically tested toascertain if the interactive model is thebetter predictor (Cohen, 1968). Usingthis technique, the multiple R's for theinteractive model are found to be .2004(Bogota, Colombia) ; .2790 (San Jose,Costa Rica) : .2803 (Mexico City, Mexico); .2048 (Panama City, Panama);and .2141 (Caracas, Venezuela).
A comparison of these figures withthose presented in Table 1 shows thatthe addltion of interaction terms forupwardly and downwardly mobile categories only slightly increases the proportion of variance explained. (Dummyvariables were created for upward anddownward categories and added to theregression equation for the additive
OccupationaI Mobility
larger social, political, and economicunits. Associated with these changesfrom a "traditional" society to a modernone is the emergence of a rationalisticvalue structure which stresses the individual's control over his environment,including his fertility and mobility. Excluding the Merican sample, the similarfindings of cross-national studies intomobility and fertility partially may reflect this impact of economic development and industrialization, espeeially forthe metropolitan populations which formthe basis for the CELADE data.
The argument concerning the importance of contextual factors can be usednot only to understand the similarity ofmobility-fertility findings between nations and cultures, but also to understand the deviations from this additivemodel, which are observed for the Mexico City data and which exist at a .10level of significance for the Bogota data.Typologies of economic and social development (Boyd, 1971; Vekemans andSegundo, 1963) suggest that the countries of Colombia and Mexico occupieda different and lower position with respect to development during the 1950'sand early 1960's than did the PanamaCanal Zone, Costa Rica and Venezuela.Although such typologies may be critieized because of the crudeness and variability of the social and economic indicators upon which they are based, theexplanation given is that the above results reflect. the correlates of economicdevelopment of the cities and of the respective countries from which the sampIes are drawn. Such an interpretationis congruent with the hypotheses ofGermani (1966) and Treiman (1970)concerning the association between economic development, stratification systems and behavioral correlates of mobility. The above findings with respect tothe Mexico City sampIe suggest futureattention should be turned to a moreexhaustive testing of Treiman's (1970)hypothesis.
13
PROBLEMS WITH SOCIAL
MOBILITY-FERTILITY RESEARCH
In addition to the explanation offeredby theories of social change, coneeptualand methodological reasons exist for thesimilar findings of social mobility-fertility investigations. Methodologically,the use of broad occupational classes inthis study and others probably reducesfertility differentials between mobile andnon-mobile groups. Also, the model examined is that of objective career mobility and ignores the conceptual multidimensionality of mobility. It is evidentfrom the literature that movement fromone occupational category to anotherconceptually is seen as involving changesin social affiliations and networks of participation. It is likewise assumed thatchange in the husband's occupation implies not only change in his participationnetworks but also in those of his wife.To the extent that these changes do notoccur, drastic behavioral modificationmay not be associated with the objectiveexperience of mobility. Recognition thatthese changes may not occur .is implicitin the discussions of mobility in modernstratification systems and in the stressplaced on the increasing heterogeneitywithin strata, but this recognition rarelyis incorporated into theory. In LatinAmerica, the family network is extremely important, and changes in occupation may not involve a severe disruption of ties to kin and friends. To theextent that mobile couples remain partof a larger family circle, which also generates a set of normative expectationsfor reproductive behavior, fertility onthe avetage would be expected to reflectan additive weighting of past and present occupational standing.
Social mobility-fertility theory alsomust be made more precise with respectto the nature of causality. The assumption .that children are continual hindrances to social mobility and the failureto consider timing of births in many
14 DEMOGRAPHY, volume 10, number 1, February 1973
studies preclude a causal interpretation Attempts to portion out variation in ferof the relationship between social mo- tility to past and present occupationsbility and reproductive behavior. It is and to the process of the mobility itselfnot altogether clear in most studies as to do represent an improvement over pastwhich variable within marriage, fertility research, where all too often fertilityor mobiIity, is cause and which is effect. variations by mobility category haveDoes social mobiIity, which presumably been attributed ad hoc to the processentails new membership in reference of mobiIity. However, when completedgroups, induce individuals to alter their fertility or number of live births afterreproductive behavior in order to match a suitable time period (typically tenthe norms of the new status level? Or, is years or more) is used as the indicatorlow fertility conducive to upward mo- of reproductive behavior and when thebility and is high fertility conducive to analysis is based on cross-sectional datadownward mobiIity? Relationships be- involving considerable time lag betweentween the two variables of mobility and points of reference, an investigation intoreproductive behavior also may be recip- the fit of the data to a model of eitherrocal to the extent that mobility may additive or interactive occupational efalter reproductive behavior, which in turn fects gives only minimal information.alters further desired or actual mobility For the most part, such an analysisexperience. To trace such causal net- simply reveals that the process of interworks requires careful consideration of generational or intra-generational movethe times when shifts in economic status ment does not appear to affect the averoccur, the social status when marriage age number of live births above andtakes place, and the timing of births beyond that whieh ean be understoodwithin marriage in relation to status simply in terms of the effeets of past andchanges (Berent, 1952; Riemer and present oecupations.Kiser, 1954). Hence, a major problem The problems of the statie analysis ofof mobility-fertility studies is that of the dynamic process are accompanied bystatie analysis of a dynamic process. The the use of number of live births as afailure to analyze the relationships be- measure of reproductive behavior. Untween births and mobility as a dynamic like membership in voluntary associaproeess is attributed not only to the lack tions or visits to kin, a woman can haveof theoretical specifications, but to the one or many children and having haduse of cross-sectional data which all too these children she can not decrease heroften lacks detailed information on mo- index of fertility. Thus, number of livebility aspirations and experienee and births is a unidireetional measure andtiming of reproduetive behavior. In an a fairly insensitive one if one is conanalysis based on cross-sectional data, sidering temporal variation associatedeonsiderable variation in - reproduetive with mobility. It appears that if the sobehavior by mobility experience will be cial mobility hypothesis is to be meanundetected if mobile and non-mobile cou- ingful the nature of causality must bepIes are similar with respeet to average more explieitly stated, a temporal analcompleted fertility but differ greatly in ysis must be made, and variables oftiming of birth and contraceptive usage. birth intervals must be ineluded. In
The need for more careful coneeptual- short, what is needed is a temporal studyization of causality and usage of longi- which utilizes birth intervals as indicatudinal, rather than eross-seetional, data tive of reproduetive behavior and whichsuggests one immediate problem with assesses the aspirations and oceupationalthe investigations of the fit of data toan status at the time of each pregnancy andadditive model of occupational effects. decisions concerning family sise made
Occupational Mobility
prior or after each pregnancy. Even thisapproach into the applicability of thesocial mobility-fertility hypothesis currently is limited. Perrucci (1967) in herstudy of approximately 1,000 engineering graduates from two west coast universities observed that there was no relationship between the correspondent'scareer mobility or social origin and average number of years between marriageand the birth of the first child. Similarly,Westoff et al. (1963) observed that thebirth of the third child in the FamilyGrowth in Metropolitan America studiescould not be predicted on the basis ofaspirations for mobility, and Featherman (1970) found no relation betweensubsequent fertility and achievementrelated motivation.
The above discussion suggests that, asit is currently formulated, social mobility-fertility theory is of dubious valuein understanding variations in fertility.Two possibilities await the would-be researcher into the problem. One approachis to continue to decompose the effectsof the constructed variable "mobility"to those attributable to process and toclasses of origin and destination. Thisapproach permits an arithmetical representation of an empirical theory (Rapaport, 1959) and introduces a degree ofquantitative simplicity into a rathercomplex area of investigation. The findings of this investigation as to the betterfit of Mexico City data to the interactive model suggest the continuation ofthis approach to data from developingnations as a means of testing Treiman's(1970) hypothesis.
Notwithstanding the findings for theMexico City sample, it seems clear thatmuch of what was originally discussedunder the rubric of the "mobility process" can now be handled adequately bythe notion of status effects. Thus, thesecond focus of future investigations intomobility and fertility is on causality.Temporal changes in actual andjor anticipated occupational statuses should be
15
studied to determine if such changes areassociated with alterations in reproduetive behavior. Such an approach redefines the problem to a career contingencymodel of behavior. lt represents a conceptual advancement in that it incorporates both aspects of status effects andprocess and offers greater specification ofcausality and timing.
ACKNOWLEDGMENTS
The above research was supported byan N.I.H. Graduate Training grant[5TOI GM01291]. The comments of JoelSmith, H. Y. Tien and an anonymousDemography referee on an earlier draftpresented at the 1972 P.A.A. annualmeeting in Toronto, Canada, are grateIully acknowledged and appreciated.Robert B. Hartford also gave generouslyof his time with respect to the CELADEdata analysis. The CELADE data wereobtained from the International Population Program, Cornell University withthe kind permission of J. M. Stycos, Director.
REFERENCES
Aiken, Michael, and David Goldberg. 1969. Social Mobility and Kinship : A Reexaminationof the Hypothesis. American Anthropologist7I :261-270.
Andrews, Frank M., James N. Morgan, andJohn A. Sonquist. 1967. Multiple Classification Analysis: AReport on a ComputerProgram for Multiple Regression Using Categorical Predictors. Ann Arbor: Survey Research Center, University of Michigan.
Baltzell, E. Digby. 1953. Social Mobility andFertility Within an Elite Group. MilbankMemorial Fund Quarterly 31:411-420.
Berent, Jerzy. 1952. Fertility and Social Mobility. Population Studies 5:244-260.
Blalock, H. M., Jr. 1967a. Status Inconsistencyand Interaction : Some Alternative Models.American Journal of Sociology 73:305-315.
--. 1967b.Status Inoonsistency, Social Mobility, Status Integration and Structural Effeets. American Sociological Review 32:790801.
Blau, Peter M. 1956. Social Mobility and Interpersonal Relations. American SociologicalReview 21:290-295.
16 DEMOGRAPHY, volume 10, number 1, February 1973
, and Otis Dudley Duncan. 1967. TheAmerican Occupational Structure. New York:John Wiley.
Boggs, Stephan T. 1957. Family Size and Social Mobility in a California Suburb. Eugenies Quarterly 4 :208-213.
Boyd, Monica. 1971. Occupational Mobilityand Fertility in Urban Latin America. Unpublished Ph.D. dissertation, Duke University.
Bresard, M. 1950. Mobile Sociale et Dimensionde la Familie. Population 3:533-566.
Cohen, Jacob. 1968. Multiple Regression as aGeneral Data-Analytic System. Psychological Bulletin 70:426-443.
Duncan, Otis Dudley. 1966. Methodological Issues in the Analysis of Social Mobility. InNeil J. Smelser and Seymour Martin Lipset(eds.), Social Structure and Mobility in Economic Development. Chicago: Aldine Publishing Co.
Featherman, David L. 1970. Marital Fertilityand the Process of Socio-Economic Attainment: An Examination of the Mobility Hypothesis. Pp. 104-131 in Larry L. Bumpassand Charles F. Westoff, The Later Yearsof Childbearing. Princeton: Princeton University Press.
Germani, G. 1963. Problems of EstablishingValid Social Research in the UnderdevelopedAreas. In Berthold F. Hoselitz and WilbertE. Moore (eds.), Industrialization and Society. Paris: UNESCO.
--. 1966. Social and Political Consequencesof Mobility. In Neil J. Smelser and SeymourMartin Lipset (eds.), Social Structure andMobility in Economic Development. Chicago: Aldine Publishing Co.
Girard, A. 1951. Mobilite Sociale et Dimensions de la Familie. Population 1 :103-124.
Glass, David V., and J. R. Hall. 1954. SocialMobility in Britain: A Study of Inter-Generation Changes in Status. Pp, 177-259 inDavid V. Glass (ed.), Social Mobility inBritain. London: Routledge and Paul.
Goldberg, David. 1959. The Fertility of TwoGeneration Urbanites. Population Studies12 :214-222.
Hall, John, and D. Caradog Jones. 1950. SocialGrading of Occupations, British Journal ofSociology 1:31-55.
Hodge, Robert W., and Donald J. Treiman.1966. Occupational Mobility and AttitudesToward Negroes. American Sociological Review 31 :93-102.
Hope, Keith. 1971. Social Mobility and Fertility. American Sociological Review 36: 10191032.
Hutchinson, Bertram. 1957. Social Grading ofOccupations in Brazil. British Journal of Sociology 8:176-189.
--. 1961. Fertility, Social Mobility and Urban Migration in Brazil. Population Studies14:182-189.
Iutaka, Sugiyama. 1965. Social StratificationResearch in Latin America. Latin AmericanResearch Review 1:7-34.
Kantner, John F., and Clyde V. Kiser. 1954.The Interrelation of Fertility, Fertility Planning and Intergenerational Social Mobility.Milbank Memorial Fund Quarterly 32 :69103.
Kessin, Kenneth. 1971. Social and PsychologicalConsequences of Intergenerational Occupational Mobility. American Journal of Sociology 77:1-18.
Laslett, Barbara. 1971. Mobility and Work Satisfaction: A Discussion of the Use and Interpretation of Mobility Models. AmericanJournal of Sociology 77: 19-35.
Melicher, Emanuel. 1965. Least-Squares Analysis of Economic Survey Data. Pp. 373-385in American Statistical Association, Proceedings of the Business and Economic StatisticsSection. Washington, D. C.: American Statistical Association.
Miller, S. M. 1971. The Future of Social Mobility Studies. American Journal of Sociology77:62-65.
Miro, Carmen A. 1966. Some MisconceptionsDisproved: A Program of Comparative Fertility Surveys in Latin America. Pp. 615-634in Bernard Berelson (ed.), Family Planningand Population Programs ; a Review ofWorld Developments. Chicago : Universityof Chicago Press.
Moore, Wilbert E. 1965. The Impact of Industry. Englewood Cliffs: Prentice Hall.
Perrucci, Carolyn Cummings. 1967. Social Origins, Mobility Patterns and Fertility. American Sociological Review 32 :615-625.
Rapoport, Anatol. 1959. Uses and Limitationsof Mathematical Models in Social Science.Pp, 348-372 in Llewellyn Gross (ed.), Symposium on Sociological Theory, New York:Harper and Row.
Riemer, Ruth, and Clyde V. Kiser. 1954. Economic Tension and Social Mobility in Relation to Fertility Planning and Size ofPlanned Family. Milbank Memorial FundQuarterly 32: 167-231.
Scott, Wolf. 1958. Fertility and Social MobilityAmong Teachers. Population Studies 11:251261.
Suits, Daniel B. 1957. Use of Dummy Variablesin Regression Equations, Journal of American Statistical Association 52 :548-551.
Tien, H. Yuan. 1961. The Social MobilityjFertility Hypothesis Reconsidered: An Ernpirical Study. American Sociological Review26:247-257.
--. 1965. Social Mobility and ControlIed
Occupational Mobility
Fertility; Family Origins and Structure ofthe Australian Academic Elite. New Haven:College and University Press.
Treiman, Donald J. 1970. Industrialization andSocial Stratification. In Edward O. Laumann(ed.), Social Stratification: Research andTheory for the 1970's. Indianapolis: BobbsMerrill Co.
--. 1965. Status Discrepancy and Prejudice.American Journal of Sociology 71:651-664.
Vekemans, Rev. R., and J. L. Segundo. 1963.Essay of a Socio-Economic Typology ofLatin American Countries. In Egbert DeVries and Jose Medina Echavarria (eds.),Soeial Aspects of Economic Development inLatin America. Paris: UNESCO.
Vorwaller, Darrel J. 1970. Social Mobility andMembership in Voluntary Associations.·
17
American Journal of Sociology 75:481-495.Westoff, Charles F. 1953. The Changing Focus
of Differential Fertility Research: The Social Mobility Hypothesis. Milbank MemorialFund Quarterly 31:24-38.
--, Marvin Bressler, and Philip C. Sagi.1960. The Concept of Soeial Mobility: AnEmpirical Inquiry. American SociologicalReview 25:375-385.
--, Robert G. Potter, and Philip C. Sagi.1963. The Third Child: A Study in the Prediction of Fertility. Princeton: PrincetonUniversity Press.
--, Robert G. Potter, Philip C. Sagi, andElliot G. Mishler. 1961. Family Growth inMetropolitan America. Princeton: PrincetonUniversity Press.