Testing the modernization hypothesis and the socialist ...

University of Groningen

Testing the modernization hypothesis and the socialist ideology hypothesisSieben Inge Graaf Paul M de

Published inBritish Journal of Sociology

DOI10108000071310120071133

IMPORTANT NOTE You are advised to consult the publishers version (publishers PDF) if you wish to cite fromit Please check the document version below

Document VersionPublishers PDF also known as Version of record

Publication date2001

Link to publication in University of GroningenUMCG research database

Citation for published version (APA)Sieben I amp Graaf P M D (2001) Testing the modernization hypothesis and the socialist ideologyhypothesis a comparative sibling analysis of educational attainment and occupational status BritishJournal of Sociology 52(3) 441 httpsdoiorg10108000071310120071133

CopyrightOther than for strictly personal use it is not permitted to download or to forwarddistribute the text or part of it without the consent of theauthor(s) andor copyright holder(s) unless the work is under an open content license (like Creative Commons)

Take-down policyIf you believe that this document breaches copyright please contact us providing details and we will remove access to the work immediatelyand investigate your claim

Downloaded from the University of GroningenUMCG research database (Pure) httpwwwrugnlresearchportal For technical reasons thenumber of authors shown on this cover page is limited to 10 maximum

Download date 12-11-2019

brought to you by COREView metadata citation and similar papers at coreacuk

provided by University of Groningen

Inge Sieben and Paul M de Graaf

Testing the modernization hypothesis and thesocialist ideology hypothesis a comparativesibling analysis of educational attainment andoccupational status

ABSTRACT

In this study we present a comparative sibling analysis This enables us to test twomajor social mobility hypotheses ie the modernization hypothesis and thesocialist ideology hypothesis We employ survey data on brothers in EnglandHungary the Netherlands Scotland Spain and the USA covering a historicalperiod from 1916 till 1990 Results show that the effects of parental social classon educational attainment are smaller in technologically advanced societies andthat the effects of parental social class on occupational status are smaller in social-democratic and communist societies In addition the total family impact onoccupational status declines with modernization But overall we observe that thefamily of origin has not lost its importance for its sonsrsquo educational attainmentand occupational status yet

KEYWORDS Social strati cation status attainment sibling analysiscomparative research

INTRODUCTION AND THEORETICAL BACKGROUND

In this study we present a comparative sibling analysis of educationalattainment and occupational status We build on Blau and Duncanrsquos(1967) classic status attainment model that is a milestone in social mobilityresearch The status attainment model describes the process of strati -cation inequalities in family background lead to inequalities in edu-cational attainment which in turn lead to inequalities in occupationalstatus A comparative sibling analysis gives more insight into this strati -cation process since it combines two powerful perspectives First siblinganalysis enables us to estimate the total impact of the family of origin oneducational attainment and occupational status We will elaborate on this

British Journal of Sociology Vol No 52 Issue No 3 (September 2001) pp 441ndash467copy 2001 London School of Economics and Political Science ISSN 0007-1315 print1468-4446 onlinePublished by Routledge Journals Taylor amp Francis Ltd on behalf of the LSEDOI 10108000071310120071133

and other advantages of sibling analysis in the next section Second a com-parison between countries and historical periods makes it possible to testtwo major social mobility hypotheses ie the modernization hypothesisand the socialist ideology hypothesis

Modernization is the general term for the (mainly) economic processesdriven by technological change that took place during the last centuryRapid technological innovation and economic development led to a morecomplex and differentiated labour market in which ef cient selection pro-cesses were needed (Kerr Dunlop Harbison and Meyers 1960 Treiman1970) According to technological-functionalist theory (Collins 1971) thelabour market has changed into one with a growing need for highly quali- ed employees working in specialized jobs in the industrial and servicesector and the educational system has expanded to satisfy this growingneed Ef cient selection processes imply meritocratic principles it is notfamily background but individual qualities that determine school successand occupational achievement in modern societies A shift from lsquoascrip-tionrsquo towards lsquoachievementrsquo (Blau and Duncan 1967) has taken place inorder to get the right person in the right place The modernization hypoth-esis predicts that the impact of the family on educational attainment andoccupational status will be smaller and that the effect of educational attain-ment on occupational status will be larger in technologically advancedsocieties Not the family of origin but personal intellectual abilities willdetermine educational and occupational outcomes

There are some additional structural arguments why the family of originis losing its importance for status attainment in modern societies First thevery expansion of educational systems in technologically advancedsocieties makes that individual educational careers are longer which willnegatively affect the importance of the family in educational careers Therationale behind this is that the impact of the family of origin is smaller inlater stages of the educational career (Mare 1980) Second occupationalinheritance has lost ground in technologically advanced societies becauseof a decreasing size of occupations that are traditionally transmitted fromgeneration to generation (mainly through material possessions) likefarmers and the petty bourgeoisie These structural changes on the labourmarket will have contributed to the decreasing impact of the family onoccupational status

It is not only economic modernization but also cultural modernizationthat has given way to the weakening impact of the family of origin on edu-cational attainment and occupational status in modern societies Firsthand in hand with technological change there has been a shift from par-ticularistic to universalistic values (Parsons 1951) meritocratic principleshave become dominant Second many of the main functions of the familyhave weakened like the socialization of children (Allan 1985 Popenoe1988) In modern societies parents have less in uence on their childrenrsquosdecisions as children increasingly make their own choices

Next to economic and cultural modernization political circumstances

442 Inge Sieben and Paul M de Graaf

are important factors to explain differences in status attainment betweencountries and historical periods Parkin (1971) advanced the hypothesisthat educational equality and occupational mobility will be larger in coun-tries where left-wing parties have been in the government The main argu-ment underlying this socialist ideology hypothesis is that societies with asocial-democratic or communist government have implemented socialreforms to reduce social inequality (Heath 1981) The most importantreforms in this respect are those that meant to reduce educational inheri-tance lsquo( ) the educational system is a powerful mechanism for alteringthe balance of advantages between classes [ and it] is also particularlysuitable as an instrument of social change in so far as in most countries itis directly under the control of the statersquo (Parkin 1971 109) It was especi-ally in societies with a communist regime that educational reforms wereintroduced to achieve this goal like the quota systems that favouredchildren of working-class parents (Simkus and Andorka 1982) Educationalreforms have also been carried out in societies with a social-democraticgovernment although they may have been less severe An example is thelowering of schooling costs particularly for the nancially disadvantagedThese reforms will have resulted in a smaller impact of the family oforigin on educational attainment in social-democratic and communistsocieties

Other social reforms aimed to reduce occupational inheritance Againthese reforms were most visible in communist societies The legal andhereditary transmission of titles was abolished and the state expropriatedland and industrial possessions in most communist societies (Parkin 1971)The egalitarian ideology of communism made that allocation of occu-pational positions was in principle based on educational credentials onlyAlthough no direct measures to reduce occupational inheritance throughproperty have been introduced in societies with a social-democraticgovernment the egalitarian ideology is present in the occupational allo-cation process

Finally social-democratic and communist governments have tried toequalize income distributions through measures as taxation and the pro-vision of state welfare (Parkin 1971) Erikson and Goldthorpe (1992) arguethat egalitarian reward systems lead to a more equal basis for competitionbetween children from different class backgrounds Moreover these chil-dren are likely to have comparable levels of ambition because there is lesslsquonormative differentiationrsquo along class lines in these societies

The socialist ideology hypothesis thus predicts that social reforms willdiminish the effect of the family of origin on educational attainment andoccupational status in social-democratic and communist societies At thesame time the effect of educational attainment on occupational statuswill be larger in these societies All this will be reinforced by the familypolicies implemented in these societies These policies can be character-ized as being very lsquoliberalrsquo (Collins and Coltrane 1991) Important func-tions of the family like the socialization of children are taken over by state

Testing the modernization hypothesis and the socialist ideology hypothesis 443

institutions which makes that parents have less in uence on theirchildrenrsquos decisions

In short the modernization hypothesis predicts a smaller family impactin technologically advanced societies whereas the socialist ideologyhypothesis predicts a smaller family impact in social-democratic and com-munist societies In addition the hypotheses predict that the effect of edu-cational attainment on occupational status will be larger in technologicallyadvanced societies and in social-democratic and communist societiesThese hypotheses call for a test in a comparative approach Most compara-tive mobility research to date has used loglinear models (GanzeboomLuijkx and Treiman 1989 Wong 1994) comparative research on the statusattainment model itself has been limited to two or three countries (for anoverview see Treiman and Ganzeboom 1990) An exception is the study ofTreiman and Yip (1989) in which the status attainment model is estimatedfor 21 countries Their results corroborated the modernization hypothesisbut the socialist ideology hypothesis was not tested Rijken (1999)employed data on 20 countries and observed that ascription is less import-ant in the process of sonrsquos rst job status attainment in state-socialistsocieties She however did not distinguish social-democratic societies fromother western democratic societies In this study we test both themodernization hypothesis and the socialist ideology hypothesis employingsibling data from several countries and historical periods

ADVANTAGES OF SIBLING ANALYSIS

Following Blau and Duncan (1967) students of the status attainmentprocess have usually operationalized family background by fatherrsquos edu-cational attainment and occupational status Some have recognized thatthe original two indicators of family background offer a rather limited per-spective on the impact of the family of origin They have started to includeadditional variables in the status attainment model when studying a singlecountry Among these additional variables are motherrsquos educational attain-ment and occupational status (Hauser and Featherman 1977 Kalmijn1994) parentsrsquo income (Sewell and Hauser 1975) parental culturalresources (De Graaf 1986) parental aspirations and expectations (Sewelland Hauser 1980) intellectual ability ( Jencks et al 1972 Taubman 1976)and school quality (Coleman et al 1966) It is however dif cult to applythis strategy in a comparative perspective Moreover no matter how manyvariables are added to the model it is impossible to measure every relevantelement of family background The total impact of the family on edu-cational attainment and occupational status will be underestimated if onlyeffects of measured aspects are studied

This problem can be overcome by using information on the educationalattainments and occupational statuses of more than one sibling in a familyHauser and others (eg Hauser and Mossel 1985 Hauser and Wong 1989)


developed appropriate sibling models that estimate the total impact of thefamily by the degree of similarity between siblings in educational attain-ment and occupational status The underlying argument of these siblingmodels is that if the total effect of family background is stronger siblingswill be more alike in educational attainment and occupational status Thetotal family effect combines all aspects of the environment as shared by thesiblings This shared environment rst of all consists of parental resourceslike economic and cultural resources Genetic inheritance forms animportant part too because brothers and sisters have on average halftheir genes in common (Scarr and Grajek 1982) In addition siblings mayin uence each other by being role models teachers and facilitators (Beninand Johnson 1984) The major advantage of sibling analysis thus is that itcombines the effects of all factors related to the family of origin and thatit assesses the total effect of the family without measurement problems

Apart from this major advantage sibling analysis has three additionaladvantages First sibling models make it possible to determine the contri-bution of measured aspects of the family to the total family impactPrevious sibling research showed that traditional indicators of parentalsocial class like parentsrsquo educational and occupational status account forabout half of the total family impact (eg Hauser and Wong 1989) A largepart of the total family impact is thus not covered when only measuredaspects of the family are included The remaining unmeasured part of thetotal family impact consists of parental resources genetic inheritanceintersibling effects and other aspects of shared environment that are notrelated to the social class measures included A second additional advan-tage is that sibling analysis is statistically very powerful It produces morereliable family background effects because it deals with information onmore than one sibling per family Finally sibling analysis gives an unbiasedestimation of the effect of educational attainment on occupational statusThis effect tends to be overestimated in traditional research becausefamily background is inadequately controlled for (Bowles 1972) Assessingthe total impact of the family renders the lsquopurersquo effect of educationalattainment on occupational status

Previous sibling studies of the effects of family background on edu-cational attainment and occupational status have been limited to singlecountries eg the Federal Republic of Germany (De Graaf and Huinink1992) Hungary (Toka and Dronkers 1996) the Netherlands (Van Eijck1996) Norway (Sweetser 1975) Sweden (Erikson 1987) and the USA(Hauser and Mossel 1985) Comparisons between historical periods arescarce and limited to a few speci c countries only In this study we presenta comparative sibling analysis to test the modernization hypothesis and thesocialist ideology hypothesis In our view a test of these hypotheses for onlythe measured aspects of the family (ie the traditional indicators ofparental social class) does not tell the whole story It is equally importantto see whether the hypotheses also hold for the total impact of the familyon educational attainment and occupational status


DATA AND OPERATIONALIZATION

To test the hypotheses mentioned we collected survey data from severalcountries The data had to meet the following criteria in order to beincluded in our analyses First of all they had to be based on nationally rep-resentative samples Second they had to contain detailed information onage educational attainment and occupational status of the respondentand at least one of his siblings And third the datasets had to include infor-mation on fatherrsquos and motherrsquos educational attainment fatherrsquos occu-pational status and the number of siblings in a family Nine datasets fromsix countries (England Hungary The Netherlands Scotland Spain andthe USA) met these criteria An overview of the datasets can be found inAppendix A

From these nine datasets we selected all families with at least twobrothers of age 25ndash65 It is inherent in sibling analysis that families withonly one child are excluded from the analysis The reason why we did notinclude women1 is rather practical In three datasets there was only infor-mation available on men and their brothers The other datasets didcontain information on female siblings but many women were housewivesand thus not gainfully employed Furthermore research has shown thatthe occupational status attainment of women depends on gender speci cmechanisms like dual responsibilities (Treiman and Roos 1983) The agerestriction of 25 was based on the assumption that most individuals willhave completed their education at that age and have started their occu-pational careers Including respondents younger than 25 would bias thesample towards lower levels of education The upper limit of 65 was chosento improve the comparability of the surveys for this is the lowest age limitin the nine datasets Another advantage of including only those siblingsthat were 65 or younger is that one minimizes the possibility of selectionbias due to differential mortality (Ganzeboom and Treiman 1994)

After these selections we made pairs out of each possible combinationof brothers (ordered by age) Obviously more pairs of brothers could beformed in large families than in small families In order to prevent over-representation of brothers from large families the pairs were weighted bya factor lsquoone divided by the number of pairs formed in a familyrsquo Theseweighted sibling-pairs were our units of analysis Note that this weightingprocedure is a conservative one the real statistical power lies between thenumber of pairs formed and the number of families

Brothersrsquo educational attainments were made comparable acrossdatasets by following the approach of Ganzeboom and Treiman (1994)Their educational attainments were measured in lsquovirtual years of edu-cationrsquo that is the minimum number of years it takes to get to a certaineducational level For comprehensive educational systems as in Englandand the USA we used the years of education claimed by the respondent orthe years known to correspond to speci c levels of educational attainmentFor multidimensional tracked educational systems like in most continental


European countries we coded the years of education associated withimportant educational categories and interpolated the remaining cat-egories In this way the relative order of levels was maintained The edu-cational attainment of the parents was constructed out of the highesteducational attainment of the father and the mother and was alsomeasured in lsquovirtual years of educationrsquo

In order to get an international comparable occupational statusmeasure national occupation codes were rst matched with the Inter-national Standard Classi cation of Occupations 1988 (ILO 1990) TheseISCO88-codes were then scored according to the International Socio-Econ-omic Index (ISEI) scale which is based on objective characteristics of occu-pational categories namely education and average income This ISEI-scalewas developed by Ganzeboom De Graaf and Treiman (1992) and wasupdated by Ganzeboom and Treiman (1996) Fatherrsquos occupational statuswas measured in the same way and was based on his occupation when therespondent was 12ndash16 years of age Because some datasets had quite a fewmissing observations on this variable (up to 111 per cent for the Americandata) and because we wanted to preserve as many cases as possible wechose to impute the missing observations on fatherrsquos occupational statusper country cohort combination The imputations were based on estimatesobtained by regression analyses on the parentrsquos educational attainmentand ndash if known ndash fatherrsquos age It would have been more appropriate if weused external instrumental variables for this imputation but these vari-ables were not available Additional analyses however showed that resultswere about the same before and after the imputation We therefore con-clude that imputation is legitimate here

Family size was operationalized by the total number of brothers andsisters in a family If no direct information on the number of siblings wasavailable we counted the number of siblings included in the dataset

In this study we do not only compare between countries but alsobetween historical periods Therefore we constructed labour marketcohorts with a range of 15 years (1916ndash1930 1931ndash1945 1946ndash19601961ndash1975 and 1976ndash1990) These labour market cohorts are based onthe year in which the brothers made the transition into the labour marketbecause it is in those years that the impact of the family on occupationalstatus can be expected to be at its maximum The starting year in the occu-pational career is dened as the birth year added by the school leaving ageplus two When the two brothers in a pair entered the labour market indifferent historical periods we took the average of their starting years andallocated them to the cohort to which this average belongs

The combination of a country and a historical period (labour marketcohort) can be seen as a society In principle 30 societies could be formedsix countries times ve labour market cohorts Data was lacking on eight ofthese combinations which means that we have 22 societies at our disposalTable I displays the number of male sibling-pairs per society with full infor-mation on all variables mentioned The table shows that there is a large


variety in the number of cases This unequal distribution would bias theresults of our analysis as the outcomes would be governed by the largerdatasets We therefore reweighed the pairs of brothers in such a way thateach society has the same number of cases ie 100 cases This is an acceptedprocedure in comparative mobility research (Erikson and Goldthorpe1992) The 2200 pairs of brothers weighted are our units of analysis

Context variables were constructed for each society that is for eachcombination of country and labour market cohort Modernization can beoperationalized in many ways for example by the percentage of the labourforce working in agriculture the average educational attainment or thegross national product per capita Since these modernization indicatorsusually correlate highly and since the number of degrees of freedom inour analyses is limited (there are only 22 societies) we chose to includeone indicator of modernization ie the energy consumption per capita2

Modern societies are characterized by a high level of mechanization andindustrialization which implies lsquothe use of ( ) inanimated energy (fossilfuels and water power) to replace or augment human power in the extrac-tion processing and distribution of natural resources or products derivedtherefromrsquo (Davis 1955 255) We chose to operationalize socialist ideologyby only one indicator as well ie the percentage of seats in parliamenttaken by social-democratic or communist parties In societies with a com-munist regime this percentage is 100 per cent by de nition Both theenergy consumption per capita and the percentage of socialist seats inparliament were averaged over the ten years in each combination ofcountry and labour market cohort A more detailed description of thecontext variables can be found in Appendix B

A SIBLING MODEL

Hauser and his associates (eg Hauser and Mossel 1985 Hauser and Wong1989) developed linear-structural sibling models to estimate the impact ofthe family of origin on educational attainment and occupational status We


TABLE I Number of male sibling-pairs per societya

1916ndash1930 1931ndash1945 1946ndash1960 1961ndash1975 1976ndash1990 Total

England 1291 1881 1310 ndash ndash 4482Hungary 223 3445 4987 3238 120 12013The Netherlands ndash ndash 353 576 279 1208Scotland 395 834 907 139 ndash 2275Spain ndash 59 78 90 ndash 227USA ndash 66 146 200 83 495Total 1909 6285 7781 4243 482 20700

Notea a society is de ned as a country-labour market cohort combination

start our analyses with one of these sibling models Figure I shows this base-line model using the notation of LISREL ( Joumlreskog and Soumlrbom 1993)The top part of the model deals with the rst brother in a pair who is byde nition the older one The bottom part deals with the younger brotherin a sibling-pair There are three indicators of family background (x1 x2and x3) which are measured by parentsrsquo education (x1) fatherrsquos occu-pational status (x2) and the number of siblings in a family (x3) Becausethese three indicators are correlated we set the variance-covariance matrixof the indicators (F) to be free for each country A common family factorfor educational attainment (h5) is constructed out of the resemblance ineducational attainment between the two brothers The three measuredindicators of family background affect this common family factor for edu-cational attainment (g51 g52 and g53) There are no direct effects of themeasured indicators of family background on the individual educationalattainments of the older (h1) and of the younger brother (h2) which aremeasured by y1 and y2 respectively In the same way the common familyfactor for occupational status (h6) is represented by the resemblance inoccupational status between the two brothers There are only direct effectsof the three measured indicators of family background on this commonoccupational status (g61 g62 and g63) The individual occupational statusesof the older (h3) and younger brother (h4) are measured by y3 and y4

We constrained all coef cients in the measurement model to be equal toone This does not only simplify our model it also means that effects willbe the same for the older and the younger brother Earlier researchshowed that hardly any systematic differences exist in the impact of thefamily on educational attainment and occupational status between siblings(Hauser and Wong 1989 De Graaf and Huinink 1992 Van Eijck 1996)The individual effects of educational attainment on occupational status for


6

6

2

1 1

1

1

1

1

e at n

at erOtildes occ status

u er o sib in s

1

6

3

4

h 6

3

cc status sibling

Occ status sibling 1

42

65

31

h 1

5

2

ducation sibling 1

Education sibling 253

52

51

3

2

1

1

1

g

hb

b

b

g

g

g

g

g

h h

j

j

j

h

z

z

z

1

z

z

1

z

FIGURE I Baseline sibling model

the older (b31) and the younger brother (b42) were constrained to beequal Furthermore these lsquobetween family effectsrsquo were constrained to beequal to the lsquowithin family effectrsquo of educational attainment on occu-pational status (b65) Modelling this equality constraint we implied thatthere is no family bias in the effect of an individualrsquos educational attain-ment on his occupational status This is in line with the outcomes ofprevious research (Hauser and Mossel 1985 De Graaf and Huinink 1992)Finally the error variance of the older brotherrsquos educational attainment(z1) was set equal to the error variance of the younger brotherrsquos edu-cational attainment (z2) The same was done for the error variances of theolder (z3) and younger (z4) brotherrsquos occupational status The errorvariance of the common family factor for educational attainment (z5) rep-resented the unmeasured part of the family impact on educational attain-ment The smaller this unmeasured part the more the three indicators offamily background (that is parentsrsquo educational attainment fatherrsquos occu-pational status and the number of siblings in a family) covered the totalfamily impact on educational attainment The same holds for the errorvariance of the common family factor for occupational status (z6)

ESTIMATION OF THE OVERALL BASELINE MODEL

We started our analyses with an estimation of the described baseline modelfor all 22 societies simultaneously only the variance-covariance matrix (F)for the three measured indicators of family background varied acrosssocieties The results of this analysis can be found in Table II The lowerpart of the table refers to the t of the model The c2 of 1525707 issigni cant with 473 degrees of freedom which would imply that the modeldoes not t the data However the c2 should not be the sole basis for deter-mining model t for it is overly rigid in most cases (Bollen and Long1993) We therefore found our conclusions about the t of the model notonly on the c2 but also on two other t statistics The Goodness of FitIndex (GFI) of Joumlreskog and Soumlrbom (1993) ranges from zero to one TheGFI of our model is 0869 which is close to the lsquorule of thumbrsquo minimumvalue of 090 Raftery (1993) proposes another t statistic the BayesianInformation Coefcient (BIC)3 If the model ts the data well the BIC sta-tistic is negative and the more negative the better the model ts Ourmodel has a BIC of 2 3098177 which shows that the baseline model ts thedata well particularly since we do not see meaningful ways to improve the t4

The estimated parameters of the overall baseline model display theexpected effects Parentsrsquo education and fatherrsquos occupational status bothhave a positive effect on educational attainment the number of siblings ina family has a negative effect With respect to occupational status fatherrsquosoccupational status exerts a positive in uence the number of siblings hasa negative effect whereas the effect of parentsrsquo education is not statistically


signi cant The most important factor for occupational status is edu-cational attainment

But how large is the total family impact on educational attainment andoccupational status and to what extent does the measured part (ie thetraditional indicators of parental social class) represent this total impactTo answer these questions we decomposed the components of variance inthe overall baseline model The second and third columns in Table IIIdisplay this decomposition for educational attainment We split up the vari-ance in a between family and a within family component The betweenfamily component refers to the variance attributable to parental resourcesgenetic inheritance intersibling effects and all other aspects of sharedenvironment This between family component turns out to be 515 per centof the variation in individual educational attainment The within familycomponent refers to the proportion of variance attributable to individualcharacteristics and is 485 per cent Family characteristics are thus asimportant for educational attainment as individual characteristics areBoth the between and the within family component can be divided in anexplained and an unexplained part Table III shows that 451 per cent ofthe between family variance is explained by parentsrsquo educational attain-ment fatherrsquos occupational status and number of siblings in a family sothe measured part represented about half of the family factor for edu-cational attainment The same was observed in previous sibling researchboth with data on only one country (eg Hauser and Wong 1989 De Graaf


TABLE II Parameter estimates of the overall baseline model (n = 2200)

Estimate (se) Standardizedestimate

Effects on family factor for educational attainmentParental education (g51) 262 (018) 395Fatherrsquos occupational status (g52) 034 (004) 282Number of siblings in family (g53) 2 106 (019) 2 125

Effects on family factor for occupational statusParental education (g61) 002 (022) 001Fatherrsquos occupational status (g62) 173 (018) 307Number of siblings in a family (g63) 2 388 (086) 2 097Educational attainment (b31 = b42 = b65) 2778 (077) 434590a

Fit of the modelc2 (df = 473) 1525707GFI 0869BIC 2 3098177

Notesa the standardized solution displays different values for on the one hand b31 and b42 (434)and for b65 (590) because the standard deviations associated with these parameters differ insize signi cant (plt001)

and Huinink 1992) and with several datasets in a comparative siblingapproach (Sieben and De Graaf 2000) Since we did not include any indi-vidual characteristics explaining educational attainment the proportion ofwithin family variance explained is zero

The same variance decomposition was made for occupational status Thebetween family variance indicating the in uence of shared environmentaccounts for 365 per cent of all variance in occupational status whereasthe within family component is 635 per cent Individual characteristicsthus are more important for occupational status than the family We alsoobserved that occupational status is further away from the parentalenvironment than educational attainment the percentage of varianceattributable to the family is lower for occupational attainment (365 percent) than for educational attainment (515 per cent) The traditional indi-cators of parental social class and the family factor of educational attain-ment explain 716 per cent of the between family variance in occupationalstatus The same factors apparently predict both sibling similarity in edu-cational attainment and sibling similarity in occupational status The withinfamily variance in occupational status can be explained by a less high pro-portion (189 per cent) which re ects the correlation between individualeducational and occupational attainment

DIFFERENCES BETWEEN SOCIETIES MODERNIZATION AND SOCIALISTIDEOLOGY

The next step in our analyses is to nd out which of the parameters of theoverall baseline model vary systematically with modernization and socialistideology across societies We start from the overall baseline model and letthe error variances z1 = z2 z3 = z4 z5 and z6 vary across societies5 Wehypothesized effects of parentsrsquo education fatherrsquos occupational statusand number of siblings in a family on educational attainment and occu-pational status to be smaller in technologically advanced societies(modernization hypothesis) and in social-democratic and communist


TABLE III Decomposition of components of variance in overall baseline model(n = 2200)

Educational attainment Occupational status

Between family variance 3884 (515) 86156 (365)Explained 1750 (451) 61668 (716)Unexplained 2134 (549) 24488 (284)

Within family variance 3663 (485) 149833 (635)Explained 0000 (00) 28272 (189)Unexplained 3663 (1000) 121561 (811)

Total variance 7547 (1000) 235989 (1000)

societies (socialist ideology hypothesis) At the same time the effect of edu-cational attainment on occupational status would be larger in thesesocieties Rindskopf (1984) showed how to model these macro-effects in aLISREL-approach Two latent variables without indicators attached tothem were included in the model in order to estimate the in uence ofmodernization and socialist ideology on the effects of family backgroundand educational attainment To make this clearer we depicted a part ofour sibling model in Figure II

Figure II displays a model to determine if modernization can explaindifferences between societies in the effect of parentsrsquo education on thefamily factor for educational attainment (g51) This is done by introducinga latent variable with no observed indicator (h7) into the model In Rind-skopf rsquos terminology this latent variable is called a phantom variable If one xes parameter b57 at the level of energy consumption for each society (egb57 = 3968 for England in 1916ndash1930) parameter g71 displays the effect ofmodernization on g51 The same procedure is followed for the effects ofmodernization on the parameters g52 g53 g61 g62 g63 and b31 = b42 = b65Analogously we estimate effects of socialist ideology on these parametersby introducing a second latent variable

Table IV gives the results of these analyses In the upper part of the tablethe effects of modernization are shown rst for Models AM to GM inwhich only one parameter at the time systematically varies with moderniz-ation then for Model HM in which the effects of modernization on all par-ameters are estimated together and nally for Model HMS in which theeffects of modernization and socialist ideology on all parameters are simul-taneously modelled The lower part of the table does the same with respectto socialist ideology In Models AS to GS the effects of socialist ideology onthe parameters are estimated one by one whereas in Model HS the effectsof socialism are modelled on all parameters simultaneously Model HMS isagain the model in which all parameters systematically vary with bothmodernization and socialist ideology The t statistics of these models canbe found in Appendix C

In Table IV we observe that modernization has a small signi cantlynegative in uence on the effect of parentsrsquo education on educational


157

ode ni ion

5

51arent ed ation

7

g

z

h

hb

g

j

FIGURE II Example of a phantom variable the effect of modernization on 51

attainment The impact of modernization however disappears when weallow all effects to vary with modernization The level of modernization ina society clearly affects the effect of fatherrsquos occupational status on edu-cational attainment negatively In addition the impact of modernizationon the effect of educational attainment on occupational status is signi -cantly positive These ndings all corroborate the modernization hypothe-sis In technologically more advanced societies the impact of traditionalindicators of parental social class on educational attainment is smaller andthe effect of onersquos educational attainment on occupational status is largerAt the same time however we see that modernization does not signi -cantly in uence the effect of the number of siblings in a family on edu-cational attainment and occupational status and the effects of parentsrsquoeducation and fatherrsquos occupational status on occupational status This isin contrast to the modernization hypothesis The effects of family back-ground on occupational status are not smaller in technologically advancedsocieties

The lower part of Table IV shows that the percentage of socialist seats inparliament does not affect the relationship between family backgroundand educational attainment This is not in line with the socialist ideologyhypothesis In social-democratic and communist societies the effects ofparental social class on educational attainment are not different than inother societies With respect to occupational attainment however weobserve the effects expected The percentage of socialist or communist


TABLE IV Sibling models with phantom variables effects of modernization andsocialist ideology (n = 2200)

Effects of modernization on Model AM2 GM Model HM Model HMS

g51 parental education ndash education (A) 2 007 (005) 2 004 (006) 2 004 (006)g52 occ status father ndash education (B) 2 002 (001) 2 002 (001) 2 002 (001)g53 number of siblings ndash education (C) 2 002 (006) 2 003 (007) 2 004 (007)g61 parental education ndash occ status (D) 022 (024) 2 047 (033) 2 018 (033)g62 occ status father ndash occ status (E) 003 (005) 2 004 (006) 2 006 (006)g63 number of siblings ndash occ status (F) 2 020 (030) 017 (032) 001 (032)b31 = b42 = b65 education ndash occ status (G) 181 (043) 271 (058) 196 (057)

Effects of socialist ideology on Model AS2 GS Model HS Model HMS

g51 parental education ndash education (A) 2 000 (000) 2 000 (000) 2 001 (001)g52 occ status father ndash education (B) 000 (000) 000 (000) 000 (000)g53 number of siblings ndash education (C) 000 (000) 000 (001) 000 (001)g61 parental education ndash occ status (D) 004 (002) 2 001 (002) 001 (003)g62 occ status father ndash occ status (E) 2 000 (000) 2 002 (001) 2 002 (001)g63 number of siblings ndash occ status (F) 2 006 (003) 2 002 (003) 2 003 (003)b31 = b42 = b65 education ndash occ status (G) 018 (004) 031 (006) 026 (006)

Notes signi cant (010ltplt005) signi cant (005ltplt001) signi cant (plt001)

seats in parliament has a signi cantly negative in uence on the effect offatherrsquos occupational status on occupational status Socialist ideology alsohas a negative impact on the effect of the number of siblings in a family onoccupational status but this impact disappears when allowing all par-ameters to systematically vary with socialism In general effects of familybackground on occupational status are smaller in social-democratic andcommunist societies Finally the effect of educational attainment on occu-pational status is positively in uenced by socialist ideology

So far we tested the modernization hypothesis and socialist ideologyhypothesis for the effects of measured aspects of family background oneducational attainment and occupational status It is however equallyimportant to test these hypotheses for the total impact of the family of origin

All aspects of the family are combined in this total family impact notonly traditional indicators of parental social class but also other parentalresources genetic inheritance intersibling effects and other aspects ofshared environment In Hauserrsquos sibling models the total family impact isde ned as the resemblance in educational attainment or occupationalstatus between the siblings (Hauser and Mossel 1985) This resemblance isbrought about by all measured and unmeasured aspects of the family oforigin We represented the total family impact on educational attainmentby the correlation between the educational attainment of the older brotherand the educational attainment of the younger brother in a sibling-pairAnalogously the total family impact on occupational status was rep-resented by the correlation between the occupational statuses of the broth-ers We used these correlation coef cients to determine the effects ofmodernization and socialist ideology on the total family impact

The relationship between the total family impact on educational attain-ment and modernization is displayed in Figure III The acronyms in this gure represent the total family impact of each society The rst threecharacters of an acronym refer to a speci c country the last two charactersto the rst year of a speci c labour market cohort Thus lsquospa46rsquo refers tothe country lsquoSpainrsquo in the labour market cohort lsquo1946ndash1960rsquo Figure IIIshows that the total family impact on educational attainment varies persociety The average correlation coefcient between brothersrsquo educationalattainments across all societies is 450 with a standard deviation of 110However there is no link between this total family impact on educationalattainment and modernization The regression line displayed in Figure IIIis practically horizontal and the correlation coef cient between the totalfamily impact and the level of energy consumption per capita is not sig-ni cant (r = 2 054) The modernization hypothesis thus cannot be cor-roborated in technologically advanced societies the family has not lost itsimportance in determining educational attainment

Figure IV shows that there also is no relationship between the total familyimpact on educational attainment and socialism The regression line isalmost at and the correlation coef cient between the total family impactand the percentage of socialist or communist seats in parliament is not



Energy consumption per capita

11010090807060504030201000

80

70

60

50

40

30

20

10

usa76

usa61usa46

usa31

spa61

spa46

spa31net76

net61

net46

net31

sco61sco46sco31

sco16hun76

hun61

hun46

hun31

hun16eng46

eng31

eng16

FIGURE III Relationship between the total family impact on educational attainmentand modernization (r = 2 054)

FIGURE IV Relationship between the total family impact on educational attainmentand socialist ideology (r = 2 120)

Percentage of socialist seats

1101009080706050403020100-10

80

70

60

50

40

30

20

10

usa76

usa61usa46

usa31

spa61

spa46

spa31net76

net61

net46

net31

sco61sco46sco31

sco16 hun76

hun61

hun46

hun31

hun16eng46

eng31

eng16

signi cant (r = 2 120) The socialist ideology hypothesis therefore has tobe rejected too Political circumstances do not have any effect on theimpact of the family on educational attainment

We now turn to the total family impact on occupational status Figure Vshows that this family impact varies across societies The average corre-lation coef cient between brothersrsquo occupational statuses is 362 with astandard deviation of 108 There is a clear relationship between the totalfamily impact on occupational status and modernization The regressionline in Figure V has a negative slope and the correlation coefcientbetween the total family impact and the level of energy consumption percapita is signi cantly negative (r = 2 439) The impact of the family onoccupational status is smaller in technologically advanced societies whichcorroborates the modernization hypothesis

Figure VI displays the relationship between the total family impact onoccupational status and socialism The regression line shows a positiveeffect of the percentage of socialist or communist seats in parliament andthe correlation coef cient between the total family impact and socialistideology is signi cantly positive (r = 436) These outcomes completelycontrast the socialist ideology hypothesis Socialist reforms did notweaken but strengthen the importance of the family for occupationalstatus


FIGURE V Relationship between the total family impact on occupational status andmodernization (r = 2 439)


11010090807060504030201000

80

70

60

50

40

30

20

10

usa76usa61

usa46

usa31

spa61spa46

spa31

net76

net61net46net31

sco61

sco46sco31

sco16

hun76

hun61

hun46hun31

hun16

eng46

eng31eng16

CONCLUSIONS

In this study we employed a comparative sibling analysis We observed that52 per cent of the variance in individual educational attainment and 36per cent of the variance in individual occupational status can be attributedto the family Traditional indicators of parental social class account forabout half (45 per cent) of this between family variance in educationalattainment and for 72 per cent of the between family variance in occu-pational status These ndings are in line with the outcomes of previoussibling research (eg Hauser and Sewell 1986) and they show that the totalimpact of the family on educational attainment and occupational status ismuch larger than can be measured by traditional indicators of parentalsocial class

We tested two major social mobility hypotheses The rst one themodernization hypothesis predicts that the impact of the family on edu-cational attainment and occupational status will be smaller in technologi-cally advanced societies because of modernization processes that havetaken place in these societies The analyses showed that this prediction isin keeping with the effects of traditional indicators of parental social classon educational attainment since the impact of parentsrsquo education andfatherrsquos occupational status are negatively affected by modernization



1101009080706050403020100-10

80

70

60

50

40

30

20

10

usa76usa61

usa46

usa31

spa61spa46

spa31

net76

net61net46net31

sco61

sco46sco31

sco16

hun76

hun61

hun46hun31

hun16

eng46

eng31eng16

FIGURE VI Relationship between the total family impact on occupational status andsocialist ideology (r = 436)

These results are in line with the results of Treiman and Yip (1989) On theother hand it seems that parents have successfully compensated for thisloss because the total family impact on educational attainment was notrelated to modernization Apparently parents have (consciously or uncon-sciously) used other strategies than social class related resources to in u-ence their sonsrsquo schooling Although it is not easy to determine what thesecompensating strategies are our outcomes show that in technologicallyadvanced societies the family has not lost its importance for educationalattainment An interpretation for the constant in uence of the familymight be in the very observation that nancial and cultural barriers havebecome less important in modern societies Parents can now decide whicheducational level is appropriate for their children If they use this newfreedom to aim at equal levels of education for all their sons the totalfamily impact would increase with modernization This tendency mightcompensate the decreasing effect of social class

Regarding the impact of the family on occupational status we observethat effects of parental social class are not related to modernizationHowever the modernization hypothesis is con rmed by the negative in u-ence of the level of energy consumption per capita on the total familyimpact on occupational status Apparently parents cannot in uence theoccupational careers of their sons as much as the educational careersParents have fewer strategies available to equalize the occupational oppor-tunities of their sons which is in line with the observation that parents canin uence their children only when they are relatively young and are livingat home In general however we conclude that the trend from lsquoascriptionrsquoto lsquoachievementrsquo that is said to have taken place in technologicallyadvanced societies is only partly observed

The second social mobility hypothesis socialist ideology hypothesisstates that reforms in social-democratic and communist societies willweaken the impact of the family on educational attainment and occu-pational status in these societies Our results showed however thatparentsrsquo education fatherrsquos occupational status and the number of sib-lings in a family have the same impact on educational attainment in social-democratic and communist societies as in other societies In addition thetotal impact of the family on educational attainment is not affected by thenumber of seats in parliament taken by socialist or communist parties Thismeans that social reforms implemented to reduce educational inequalityhave not been very effective in social-democratic and communist societiesIn contradiction to the socialist ideology hypothesis the family of originremains very important for educational attainment in these societies

In line with the socialist ideology hypothesis the effect of fatherrsquos occu-pational status on occupational status is smaller in social-democratic andcommunist societies The same holds for the effect of the number of sib-lings in a family on occupational status The total family impact on occu-pational status however is not smaller in social-democratic and communistsocieties On the contrary the impact of the family is positively related to


the percentage of socialist or communist seats in parliament This might becaused by the following mechanism Siblings resemble each other in edu-cational attainment (in all societies) Since the effect of educational attain-ment on occupational status was so much higher in social-democratic andcommunist societies this has led to a larger resemblance in occupationalstatus between siblings In short our results imply that the social reformsin social-democratic and communist societies were effective in reducingthe direct inheritance of status positions but they were not able to dimin-ish the importance of the family for occupational status

With this study we hoped to prove that sibling analysis is of value forcomparative social mobility research If we had incorporated onlymeasured indicators of parental social class in our models and had notestimated the total impact of the family on educational attainment andoccupational status our conclusions with respect to the modernizationhypothesis and the socialist ideology hypothesis would have been differentComparative sibling analysis sheds new light on the total impact of thefamily on educational attainment and occupational status The major con-clusion is that the family has not lost its importance yet It is however dif -cult to determine the exact content of this total family impact Next toparental social class other parental resources may play a role as well asgenetic inheritance intersibling effects and all other aspects of sharedenvironment In our opinion the main attraction of future comparativesibling research does not lie in opening this black box Research shows thatadditional measures of parental resources do not increase the explainedproportion of variance very much (Van Eijck 1996)6 Moreover siblingmodels cannot improve our insight into the effects of genetic inheritancesince these effects are found in both the impact of traditional indicators ofparental social class and in the remaining part of the total family impactOther designs like adoptiontwin designs are needed in order to estimatethe in uence of inherited genes but these designs are also not unprob-lematic7 We think that comparative sibling research should focus on whatit is good at namely estimating the total impact of the family In thisrespect a closer look at the mechanisms behind modernization and social-ist ideology would improve our understanding of the impact of the familyin the process of strati cation

(Date accepted January 2001) Inge Siebenand

Paul M De GraafDepartment of SociologyUniversity of Nijmegen



APPENDIX A Overview of data used in this paper

Dataset Country Time-Period Maximum number of sibingsin dataset

Halsey A H Goldthorpe J H Payne C England 1916ndash1960 Max 1 siblingand Heath A F 1978 Oxford Social Mobility random out of all brothers Inquiry 1972 Oxford Oxford Social older than 16Mobility Group

Harcsa I and Kulsar R 1983 Social Hungary 1916ndash1975 Max 7 siblingsMobility Study Hungary 1983 BudapestCentral Statistical Ofce

Central Statistical Of ce 1992 Hungarian Hungary 1931ndash1990 Max 6 siblingsSocial Mobility and Life History Survey 1992Budapest Central Statistical Of ce

Ultee W C and Ganzeboom H B G Netherlands 1946ndash1990 Max 9 siblings1993 Family Survey Dutch Population 1992 if more than 9 the ones that1993 Nijmegen Department of Sociology come closest to respondent in Nijmegen University age

De Graaf N D De Graaf P M Netherlands 1946ndash1990 Max 3 siblingsKraaykamp G and Ultee W C 1999 random out of all siblingsFamily Survey Dutch Population 1998Nijmegen Department of SociologyNijmegen University

Moore R and Payne G 1975 Scottish Scotland 1916ndash1975 Max 1 siblingMobility Study 1974ndash1975 Aberdeen random out of all brothersUniversity of Aberdeen older than 16

Centro de Investigaciones Sobre la Spain 1931ndash1975 Max 2 siblingsRealidad Social 1991 Cires Monthly Survey eldest and youngest older May 1991 Education and Social Mobility than 18Madrid ASEP Inc

Davis J Coleman J Nie N Riley J and USA 1916ndash1960 Max 1 siblingJencks C 1973 NORC Amalgam Survey eldestDecember 1973 Chicago National OpinionResearch Center

Davis J A and Smith T W 1996 General USA 1931ndash1990 Max 1 siblingSocial Survey 1994 Chicago National random out of all siblingsOpinion Research Center

APPENDIX B Context variables for each society (country-cohort combination)

Society Energy consumption Socialist seats inper capitaa parliamentb

England1916ndash1930 3968c 2001931ndash1945 3932c 2501946ndash1960 4554 502

Hungary1916ndash1930 0481c 471931ndash1945 0597c 721946ndash1960 1498 8741961ndash1975 2896 10001976ndash1990 3745 1000

Continued


APPENDIX B Continued


The Netherlands1931ndash1945 1668c 1761946ndash1960 2143 3641961ndash1975 4328 3571976ndash1990 6343 380

Scotland1916ndash1930 3968c 2001931ndash1945 3932c 2501946ndash1960 4554 5021961ndash1975 5175 484

Spain1931ndash1945 0370c 651946ndash1960 0673 001961ndash1975 1389 00

USA1916ndash1930 5884 031931ndash1945 5312 061946ndash1960 7432 001961ndash1975 10103 001976ndash1990 10249 00

Notesa The energy consumption in tonnes of coal equivalent per capita was based on the average

energy consumption of all years in a particular cohort and refers to the apparent con-sumption of coal lignite petroleum products natural gas and hydro and nuclear electric-ity Coke manufactured gas and electricity internationally traded were considered to havebeen consumed by the importing country Figures were based on Banks (1983) and UnitedNations (several years)

b The percentage of seats won by social-democratic or communist parties in parliament wasbased on the average percentage of all years in that cohort To compute this average per-centages were assigned to all years Years between two elections were assigned the percent-age of seats won in the previous election If an election was held in the rst half of a yearthen the percentage of seats won in that election counted for that year If an election washowever held in the second half of a year the percentage of seats won in that electioncounted not for the following year Figures are based on Mackie and Rose (1991) andNohlen (1969) More information about parties and elections in these societies is availableon request from the rst author

c No information is available for the period of the Second World War (1940ndash1945) Follow-ing Banks (1983) we estimated the consumption of energy by using a linear imputationbased on the scores of before (1939) and after (1946) the war For the missing observationsduring the First World War (1916ndash1918) we had to adopt a different strategy because thereis only information available on the consumption of energy after this war Therefore weestimated a linear regression equation for data from the period 1919ndash1925 with energyconsumption as the dependent variable and year (in two digits) as the independent vari-able For England the equation is lsquo3319 + 00287 yearrsquo for the USA lsquo284071 + 01342 yearrsquo and for Hungary lsquo 2 0725 + 00528 yearrsquo We used these equations to compute themissing levels of energy consumption for the period 1916ndash1919


APP

EN

DIX

C

Fit s

tatis

tics

of m

odel

s in

whi

ch p

aram

eter

s of

the

over

all b

aseli

ne m

odel

var

y ac

ross

soc

ietie

s

c2df

pG

FIB

ICC

ontr

ast

d (c2

)d (

df)

p

Mod

el O

ove

rall

base

line

mod

el +

var

ying

z1

= z 2

z3

= z 4

z5

z6

434

7238

90

060

922

3368

01

Mod

el A

M g

51va

ryin

g w

ith

mod

ern

izat

ion

432

9538

80

060

922

3660

00

O-A

M1

771

018

Mod

el A

S g

51va

ryin

g w

ith

soc

ialis

m43

406

388

005

092

232

582

9O

-AS

006

10

82

Mod

el B

M g

52va

ryin

g w

ith

mod

ern

izat

ion

430

6638

80

070

922

3362

30

O-B

M4

061

004

Mod

el B

S g

52va

ryin

g w

ith

soc

ialis

m43

431

388

005

092

233

586

5O

-BS

041

10

52

Mod

el C

M g

53va

ryin

g w

ith

mod

ern

izat

ion

434

6038

80

050

922

3358

35

O-C

M0

121

073

Mod

el C

S g

53va

ryin

g w

ith

soc

ialis

m43

469

388

005

092

233

582

6O

-CS

003

10

87

Mod

el D

M g

61va

ryin

g w

ith

mod

ern

izat

ion

433

9238

80

050

922

3259

04

O-D

M0

801

037

Mod

el D

S g

61va

ryin

g w

ith

soc

ialis

m43

135

388

006

092

233

616

1O

-DS

337

10

07

Mod

el E

M g

62va

ryin

g w

ith

mod

ern

izat

ion

434

4238

80

050

922

3358

53

O-E

M0

301

059

Mod

el E

S g

62va

ryin

g w

ith

soc

ialis

m43

439

388

005

092

233

585

7O

-ES

033

10

57

Mod

el F

M g

63va

ryin

g w

ith

mod

ern

izat

ion

434

2838

80

050

922

3358

68

O-F

M0

441

002

Mod

el F

S g

63va

ryin

g w

ith

soc

ialis

m42

964

388

007

092

233

633

2O

-FS

508

10

00

Mod

el G

M b

31=

b 42

= b 6

5va

rin

g w

ith

mod

ern

izat

ion

418

3438

80

140

932

3374

62

O-G

M16

38

10

00M

odel

GS

b31

= b 4

2=

b 65

vary

ing

wit

h s

ocia

lism

406

3338

80

250

912

3386

62

O-G

S28

38

10

00

Mod

el H

M a

ll va

ryin

g w

ith

mod

ern

izat

ion

408

6938

20

170

922

3325

61

O-H

M26

03

70

00M

odel

HS

all

vary

ing

wit

h s

ocia

lism

385

2238

20

440

912

3349

08

O-H

S49

50

70

00M

odel

HM

S a

ll va

ryin

g w

ith

mod

ern

izat

ion

an

d s

ocia

lism

368

0037

50

590

932

3297

87

O-H

MS

667

114

000

1 In a companion study in which wepresent a comparative sibling analysis ofthe effects of family background on edu-cational attainment only we included bothmen and women in our analysis (Siebenand De Graaf 2000)

2 We also estimated models withanother indicator of modernization iethe percentage of the labour forceworking in agriculture The correlationbetween this indicator and the energyconsumption per capita was 2 71 Theresults of these additional analyses gaveway to the same conclusions as the resultspresented here

3 BIC = c2 2 df elog (nk) withc2 = Chi-squared statistic of the modeldf = number of degrees of freedomn = sample sizek = number of observed variables

The Bayesian Information Criteriumhas been criticized by Weakliem (1999) asan instrument for model selection Raftery(1999) and Xie (1999) however arguethat the BIC remains a very useful instru-ment In this paper we base our choice forthe baseline sibling model estimated rstand foremost on theoretical argumentsMoreover we do not rely on one measureof model t exclusively but discuss threemeasures ie c2 GFI and BIC

4 Relaxing the constraints imposeddoes not signi cantly improve the t of thebaseline sibling model with the exceptionof allowing effects of family background oneducational attainment to vary betweenthe older and the younger brother(l25 sup1 1) However a clear pattern in theseeffects cannot be discerned Most societieshardly show any differences between theolder and the younger brother in somesocieties effects of family background oneducational attainment are larger for theolder brother and in some societies theseeffects are larger for the younger brotherMoreover our focus in this paper is not ondifferences in effects of family backgroundwithin families but on differences betweenfamilies We therefore chose a model withthe constraints mentioned

5 This model has a c2 of 434718 with389 degrees of freedom Compared to theoverall baseline model this is a signi cant

improvement Furthermore the GFIincreases to 0917 and the BIC statistic isstronger negative (2 3368012) than in theoverall baseline model (See Model O inAppendix C)

6 The proportion of explained vari-ance increases only slightly if additionalmeasures of parental resources areincluded because the resources and thetraditional indicators of parental socialclass are relatively highly correlated

7 To distinguish between effects due toenvironment and effects due to genes acombination of adoption and twin studiescould be used A study of LichtensteinPedersen and McClearn (1992) onSwedish twins reared apart and togethershows that environmental effects oneducational and occupational attainmentare smaller than genetic effects This holdsparticularly for men and for the youngerage groups However adoptiontwindesigns also have disadvantages First of allthese designs are based on additionalassumptions that are lsquoboth implausibleand restrictiversquo (Hauser and Wong 1989151) A second disadvantage of adoptiontwin designs is that adoption agenciesselect only speci c families for adoptionnamely those that can provide a stablehome environment They prefer marriedcouples with high educational attainmentsand stable occupations This means thattwins reared apart often live in quitesimilar circumstances and that the vari-ation in environment is not very large Athird disadvantage of adoptiontwindesigns is that they are not very suitable forcomparative research because data onadopted twins are not easy to nd andhardly ever are nationally representative

BIBLIOGRAPHY

Allan G 1985 Family Life Domestic Roles andSocial Organization Oxford BlackwellBanks A S 1983 Cross-National Time SeriesData 1815ndash1983 [machine readable datale]Binghamton NY Center for ComparativePolitical Research State University of NewYorkBenin M H and Johnson D R 1984


NOTES

lsquoSibling Similarities in Educational Attain-ment a Comparison of Like-sex and Cross-sex Sibling Pairsrsquo Sociology of Education57(1) 11ndash21Blau P M and Duncan O D 1967 TheAmerican Occupational Structure New YorkWileyBollen K A and Long J S 1993 lsquoIntro-ductionrsquo in K A Bollen and J S Long(eds) Testing Structural Equation ModelsNewbury Park CA Sage PublicationsBowles S 1972 lsquoSchooling and Inequalityfrom generation to generationrsquo Journal ofPolitical Economy 80(3) S219ndash51Coleman J S Campbell E Q HobsonC F McPartland J Mood A MWeifeld F D and York R L 1966 Equal-ity of Educational Opportunity WashingtonDC US Government Printing Of ceCollins R 1971 lsquoFunctional and Con ictTheories of Educational StratificationrsquoAmerican Sociological Review 36(6) 1002ndash19Collins R and Coltrane S 1991 Sociologyof Marriage and the Family Gender Love andProperty Chicago Nelson-Hall PublishersDavis K 1955 lsquoSocial and DemographicAspects of Economic Development inIndiarsquo in S Kuznets W E Moore and J JSpengler (eds) Economic Growth BrazilIndia Japan Durham NC Duke UniversityPressDe Graaf P M 1986 lsquoThe Impact of Finan-cial and Cultural Resources on Edu-cational Attainment in the NetherlandsrsquoSociology of Education 59(3) 237ndash46De Graaf P M and Huinink J J 1992lsquoTrends in Measured and UnmeasuredEffects of Family Background on Edu-cational Attainment and OccupationalStatus in the Federal Republic ofGermanyrsquo Social Science Research 21(1)84ndash112Erikson R 1987 lsquoThe Long Arm of Originthe effects of family background on occu-pational achievementrsquo in U Bergryd andC G Jansen (eds) Sociological MiscellanyEssays in Honour of Gunnar Boalt Stock-holm University of Stockholm Depart-ment of SociologyErikson R and Goldthorpe J H 1992The Constant Flux A Study of Class Mobility inIndustrial Societies Oxford ClarendonPressGanzeboom H B G De Graaf P M andTreiman D J with an appendix by De

Leeuw J 1992 lsquoA Standard InternationalSocio-Economic Index of OccupationalStatusrsquo Social Science Research 21(1) 1ndash56Ganzeboom H B G Luijkx R andTreiman D J 1989 lsquoIntergenerationalClass Mobility in Comparative Perspec-tiversquo Research in Social Stratification andMobility 8 3ndash84Ganzeboom H B G and Treiman D J1994 lsquoPreliminary results on educationalexpansion and educational achievementin comparative perspectiversquo in H ABecker and P L J Hermkens (eds) Soli-darity of Generations Demographic Economicand Social Change and its Consequences Pro-ceedings of a Symposium held on April 7 and 8at Utrecht University the Netherlands Amster-dam Thesis Publishersmdashmdashmdash 1996 lsquoInternationally ComparableMeasures of Occupational Status for the1988 International Standard Classi cationof Occupationsrsquo Social Science Research25(3) 201ndash39Hauser R M and Featherman D L 1977The Process of Stratication Trends and Analy-ses New York Academic PressHauser R M and Mossel P A 1985 lsquoFra-ternal Resemblance in Educational Attain-ment and Occupational Statusrsquo AmericanJournal of Sociology 91(3) 650ndash73Hauser R M and Sewell W H 1986lsquoFamily Effects in Simple Models of Edu-cation Occupational Status and EarningsFindings from the Wisconsin and Kalama-zoo Studiesrsquo Journal of Labor Economics4(3) S83-S120Hauser R M and Wong R S-K 1989lsquoSibling Resemblance and IntersiblingEffects in Educational Attainmentrsquo Soci-ology of Education 62(3) 149ndash71Heath A 1981 Social Mobility GlasgowFontanaILO (International Labour Of ce) 1990International Standard Classication of Occu-pations ISCO-88 Geneva InternationalLabour Of ceJencks C Smith M Acland H Bane MJ Cohen D Gintis H Heyns B andMichelson S 1972 Inequality a Reassess-ment of the Effect of Family and Schooling inAmerica New York Basic BooksJoumlreskog K G and Soumlrbom D 1993LISREL 8 Users Reference Guide ChicagoScienti c Software InternationalKalmijn M 1994 lsquoMotherrsquos Occupational


Status and Childrenrsquos Schoolingrsquo AmericanSociological Review 59(2) 257ndash75Kerr C Dunlop J T Harbison F Hand Meyers C A 1960 Industrialism andIndustrial Man Cambridge MA HarvardUniversity PressLichtenstein P Pedersen N L andMcClearn G E 1992 lsquoThe Origins of Indi-vidual Differences in Occupational Statusand Educational Level a Study of TwinsReared Apart and Togetherrsquo Acta Sociolog-ica 35(1) 13ndash31Mackie T T and Rose R 1991 The Inter-national Almanac of Electoral Histor y FullyRevised Third Edition Basingstoke Mac-MillanMare R D 1980 lsquoSocial Background andSchool Continuation Decisionsrsquo Journal ofthe American Statistical Association 75(370)295ndash305Nohlen D 1969 Die Wahl der Parlamenteund andere Staatsorgane Ein HandbuchBerlin Walter de GruyterParkin F 1971 Class Inequality and PoliticalOrder Social Stratication in Capitalist andCommunist Societies London MacGibbonand KeeParsons T 1951 The Social System NewYork Free PressPopenoe D 1988 Disturbing the Nest FamilyChange and Decline in Modern Societies NewYork Aldine de GruyterRaftery A E 1993 lsquoBayesian Model Selec-tion in Structural Equation Modelsrsquo inK A Bollen and J S Long (eds) TestingStructural Equation Models Newbury ParkCA Sage Publicationsmdashmdash 1999 lsquoBayes Factors and BICrsquo Socio-logical Methods amp Research 27(3) 411ndash27Rijken S 1999 Educationa l Expansion andStatus Attainment a Cross-national andOver-time Comparison Amsterdam ThelaThesisRindskopf D 1984 lsquoUsing Phantom andImaginary Variables to Parameterize Con-straints in Linear Structural Modelsrsquo Psy-chometrika 49(1) 37ndash47Scarr S and Grajek S 1982 lsquoSimilaritiesand Differences Among Siblingsrsquo in M ELamb and B Sutton-Smith (eds) SiblingRelationships Their Nature and SignicanceAcross the Lifespan Hillsdale NJ LawrenceErlbaum AssociatesSewell W H and Hauser R M 1975 Edu-cation Occupation and Earnings Achievement

in the Early Career New York AcademicPressmdashmdash 1980 lsquoThe Wisconsin LongitudinalStudy of Social and Psychological Factorsin Aspirations and Achievementsrsquo Researchin Sociology of Education and Socialization1(1) 59ndash99Sieben I and De Graaf P M 2000 lsquoTheImpact of the Family on EducationalAttainment a Comparative Sibling Analy-sisrsquo paper presented at the 95th annualmeeting of the American SociologicalAssociation Washington DCSimkus A and Andorka R 1982 lsquoInequal-ities in Educational Attainment inHungary 1923ndash1972rsquo American SociologicalReview 47(6) 267ndash81Sweetser D A 1975 lsquoEducation and Privi-lege An Analysis of Sibling OccupationalMobilityrsquo Acta Sociologica 18(1) 23ndash35Taubman P 1976 lsquoThe Determinants ofEarnings Genetics Family and OtherEnvironments a Study of White MaleTwinsrsquo American Economic Review 66(5)858ndash70Toka G and Dronkers J 1996 lsquoSiblingResemblance in Educational AttainmentOccupational Prestige and Wealth inHungary during the Communist RegimersquoEuropean Sociological Review 12(3) 251ndash69Treiman D J 1970 lsquoIndustrialization andSocial Strati cationrsquo in E O Laumann(ed) Social Stratification Research andTheor y for the 1970s Indianapolis Bobbs-MerrillTreiman D J and Ganzeboom H B G1990 lsquoCross-National Comparative Status-Attainment Researchrsquo Research in SocialStratication and Mobility 9 105ndash27Treiman D J and Roos P A 1983 lsquoSexand Earnings in Industrial Society A Nine-Nation Comparisonrsquo American Journal ofSociology 89(3) 612ndash50Treiman D J and Yip K-B 1989 lsquoEdu-cational and Occupational Attainment in21 Countriesrsquo in M L Kohn (ed) Cross-National Research and the Analysis of Edu-cational Systems Newbury Park CA SageUnited Nations [several years] StatisticalYearbook New York United NationsVan Eijck K 1996 Family and Opportunity aSibling Analysis of the Impact of Family Back-ground on Education Occupation and Con-sumption Tilburg Tilburg University PressWeakliem D L 1999 lsquoA Critique of the


Bayesian Information Criterion for ModelSelectionrsquo Sociological Methods amp Research27(3) 359ndash97Wong R S-K 1994 lsquoPostwar MobilityTrends in Advanced Industrial Societiesrsquo

Research in Social Stratication and Mobility13 121ndash44Xie Y 1999 lsquoThe Tension Between Gener-ality and Accuracyrsquo Sociological Methods ampResearch 27(3) 428ndash35


Inge Sieben and Paul M de Graaf

Testing the modernization hypothesis and thesocialist ideology hypothesis a comparativesibling analysis of educational attainment andoccupational status

ABSTRACT

In this study we present a comparative sibling analysis This enables us to test twomajor social mobility hypotheses ie the modernization hypothesis and thesocialist ideology hypothesis We employ survey data on brothers in EnglandHungary the Netherlands Scotland Spain and the USA covering a historicalperiod from 1916 till 1990 Results show that the effects of parental social classon educational attainment are smaller in technologically advanced societies andthat the effects of parental social class on occupational status are smaller in social-democratic and communist societies In addition the total family impact onoccupational status declines with modernization But overall we observe that thefamily of origin has not lost its importance for its sonsrsquo educational attainmentand occupational status yet

KEYWORDS Social strati cation status attainment sibling analysiscomparative research

INTRODUCTION AND THEORETICAL BACKGROUND

In this study we present a comparative sibling analysis of educationalattainment and occupational status We build on Blau and Duncanrsquos(1967) classic status attainment model that is a milestone in social mobilityresearch The status attainment model describes the process of strati -cation inequalities in family background lead to inequalities in edu-cational attainment which in turn lead to inequalities in occupationalstatus A comparative sibling analysis gives more insight into this strati -cation process since it combines two powerful perspectives First siblinganalysis enables us to estimate the total impact of the family of origin oneducational attainment and occupational status We will elaborate on this

British Journal of Sociology Vol No 52 Issue No 3 (September 2001) pp 441ndash467copy 2001 London School of Economics and Political Science ISSN 0007-1315 print1468-4446 onlinePublished by Routledge Journals Taylor amp Francis Ltd on behalf of the LSEDOI 10108000071310120071133





































A SIBLING MODEL










6

6

2

1 1

1

1

1

1

e at n


u er o sib in s

1

6

3

4

h 6

3

cc status sibling


42

65

31

h 1

5

2

ducation sibling 1


52

51

3

2

1

1

1

g

hb

b

b

g

g

g

g

g

h h

j

j

j

h

z

z

z

1

z

z

1

z

























Total variance 7547 (1000) 235989 (1000)






157

ode ni ion

5

51arent ed ation

7

g

z

h

hb

g

j



















11010090807060504030201000

80

70

60

50

40

30

20

10

usa76

usa61usa46

usa31

spa61

spa46

spa31net76

net61

net46

net31

sco61sco46sco31

sco16hun76

hun61

hun46

hun31

hun16eng46

eng31

eng16




1101009080706050403020100-10

80

70

60

50

40

30

20

10

usa76

usa61usa46

usa31

spa61

spa46

spa31net76

net61

net46

net31

sco61sco46sco31

sco16 hun76

hun61

hun46

hun31

hun16eng46

eng31

eng16







11010090807060504030201000

80

70

60

50

40

30

20

10

usa76usa61

usa46

usa31

spa61spa46

spa31

net76

net61net46net31

sco61

sco46sco31

sco16

hun76

hun61

hun46hun31

hun16

eng46

eng31eng16

CONCLUSIONS





1101009080706050403020100-10

80

70

60

50

40

30

20

10

usa76usa61

usa46

usa31

spa61spa46

spa31

net76

net61net46net31

sco61

sco46sco31

sco16

hun76

hun61

hun46hun31

hun16

eng46

eng31eng16




























Continued













APP

EN

DIX

C

Fit s

tatis

tics

of m

odel

s in

whi

ch p

aram

eter

s of

the

over

all b

aseli

ne m

odel

var

y ac

ross

soc

ietie

s

c2df

pG

FIB

ICC

ontr

ast

d (c2

)d (

df)

p

Mod

el O

ove

rall

base

line

mod

el +

var

ying

z1

= z 2

z3

= z 4

z5

z6

434

7238

90

060

922

3368

01

Mod

el A

M g

51va

ryin

g w

ith

mod

ern

izat

ion

432

9538

80

060

922

3660

00

O-A

M1

771

018

Mod

el A

S g

51va

ryin

g w

ith

soc

ialis

m43

406

388

005

092

232

582

9O

-AS

006

10

82

Mod

el B

M g

52va

ryin

g w

ith

mod

ern

izat

ion

430

6638

80

070

922

3362

30

O-B

M4

061

004

Mod

el B

S g

52va

ryin

g w

ith

soc

ialis

m43

431

388

005

092

233

586

5O

-BS

041

10

52

Mod

el C

M g

53va

ryin

g w

ith

mod

ern

izat

ion

434

6038

80

050

922

3358

35

O-C

M0

121

073

Mod

el C

S g

53va

ryin

g w

ith

soc

ialis

m43

469

388

005

092

233

582

6O

-CS

003

10

87

Mod

el D

M g

61va

ryin

g w

ith

mod

ern

izat

ion

433

9238

80

050

922

3259

04

O-D

M0

801

037

Mod

el D

S g

61va

ryin

g w

ith

soc

ialis

m43

135

388

006

092

233

616

1O

-DS

337

10

07

Mod

el E

M g

62va

ryin

g w

ith

mod

ern

izat

ion

434

4238

80

050

922

3358

53

O-E

M0

301

059

Mod

el E

S g

62va

ryin

g w

ith

soc

ialis

m43

439

388

005

092

233

585

7O

-ES

033

10

57

Mod

el F

M g

63va

ryin

g w

ith

mod

ern

izat

ion

434

2838

80

050

922

3358

68

O-F

M0

441

002

Mod

el F

S g

63va

ryin

g w

ith

soc

ialis

m42

964

388

007

092

233

633

2O

-FS

508

10

00

Mod

el G

M b

31=

b 42

= b 6

5va

rin

g w

ith

mod

ern

izat

ion

418

3438

80

140

932

3374

62

O-G

M16

38

10

00M

odel

GS

b31

= b 4

2=

b 65

vary

ing

wit

h s

ocia

lism

406

3338

80

250

912

3386

62

O-G

S28

38

10

00

Mod

el H

M a

ll va

ryin

g w

ith

mod

ern

izat

ion

408

6938

20

170

922

3325

61

O-H

M26

03

70

00M

odel

HS

all

vary

ing

wit

h s

ocia

lism

385

2238

20

440

912

3349

08

O-H

S49

50

70

00M

odel

HM

S a

ll va

ryin

g w

ith

mod

ern

izat

ion

an

d s

ocia

lism

368

0037

50

590

932

3297

87

O-H

MS

667

114

000










BIBLIOGRAPHY



NOTES














































A SIBLING MODEL










6

6

2

1 1

1

1

1

1

e at n


u er o sib in s

1

6

3

4

h 6

3

cc status sibling


42

65

31

h 1

5

2

ducation sibling 1


52

51

3

2

1

1

1

g

hb

b

b

g

g

g

g

g

h h

j

j

j

h

z

z

z

1

z

z

1

z

























Total variance 7547 (1000) 235989 (1000)






157

ode ni ion

5

51arent ed ation

7

g

z

h

hb

g

j



















11010090807060504030201000

80

70

60

50

40

30

20

10

usa76

usa61usa46

usa31

spa61

spa46

spa31net76

net61

net46

net31

sco61sco46sco31

sco16hun76

hun61

hun46

hun31

hun16eng46

eng31

eng16




1101009080706050403020100-10

80

70

60

50

40

30

20

10

usa76

usa61usa46

usa31

spa61

spa46

spa31net76

net61

net46

net31

sco61sco46sco31

sco16 hun76

hun61

hun46

hun31

hun16eng46

eng31

eng16







11010090807060504030201000

80

70

60

50

40

30

20

10

usa76usa61

usa46

usa31

spa61spa46

spa31

net76

net61net46net31

sco61

sco46sco31

sco16

hun76

hun61

hun46hun31

hun16

eng46

eng31eng16

CONCLUSIONS





1101009080706050403020100-10

80

70

60

50

40

30

20

10

usa76usa61

usa46

usa31

spa61spa46

spa31

net76

net61net46net31

sco61

sco46sco31

sco16

hun76

hun61

hun46hun31

hun16

eng46

eng31eng16




























Continued













APP

EN

DIX

C

Fit s

tatis

tics

of m

odel

s in

whi

ch p

aram

eter

s of

the

over

all b

aseli

ne m

odel

var

y ac

ross

soc

ietie

s

c2df

pG

FIB

ICC

ontr

ast

d (c2

)d (

df)

p

Mod

el O

ove

rall

base

line

mod

el +

var

ying

z1

= z 2

z3

= z 4

z5

z6

434

7238

90

060

922

3368

01

Mod

el A

M g

51va

ryin

g w

ith

mod

ern

izat

ion

432

9538

80

060

922

3660

00

O-A

M1

771

018

Mod

el A

S g

51va

ryin

g w

ith

soc

ialis

m43

406

388

005

092

232

582

9O

-AS

006

10

82

Mod

el B

M g

52va

ryin

g w

ith

mod

ern

izat

ion

430

6638

80

070

922

3362

30

O-B

M4

061

004

Mod

el B

S g

52va

ryin

g w

ith

soc

ialis

m43

431

388

005

092

233

586

5O

-BS

041

10

52

Mod

el C

M g

53va

ryin

g w

ith

mod

ern

izat

ion

434

6038

80

050

922

3358

35

O-C

M0

121

073

Mod

el C

S g

53va

ryin

g w

ith

soc

ialis

m43

469

388

005

092

233

582

6O

-CS

003

10

87

Mod

el D

M g

61va

ryin

g w

ith

mod

ern

izat

ion

433

9238

80

050

922

3259

04

O-D

M0

801

037

Mod

el D

S g

61va

ryin

g w

ith

soc

ialis

m43

135

388

006

092

233

616

1O

-DS

337

10

07

Mod

el E

M g

62va

ryin

g w

ith

mod

ern

izat

ion

434

4238

80

050

922

3358

53

O-E

M0

301

059

Mod

el E

S g

62va

ryin

g w

ith

soc

ialis

m43

439

388

005

092

233

585

7O

-ES

033

10

57

Mod

el F

M g

63va

ryin

g w

ith

mod

ern

izat

ion

434

2838

80

050

922

3358

68

O-F

M0

441

002

Mod

el F

S g

63va

ryin

g w

ith

soc

ialis

m42

964

388

007

092

233

633

2O

-FS

508

10

00

Mod

el G

M b

31=

b 42

= b 6

5va

rin

g w

ith

mod

ern

izat

ion

418

3438

80

140

932

3374

62

O-G

M16

38

10

00M

odel

GS

b31

= b 4

2=

b 65

vary

ing

wit

h s

ocia

lism

406

3338

80

250

912

3386

62

O-G

S28

38

10

00

Mod

el H

M a

ll va

ryin

g w

ith

mod

ern

izat

ion

408

6938

20

170

922

3325

61

O-H

M26

03

70

00M

odel

HS

all

vary

ing

wit

h s

ocia

lism

385

2238

20

440

912

3349

08

O-H

S49

50

70

00M

odel

HM

S a

ll va

ryin

g w

ith

mod

ern

izat

ion

an

d s

ocia

lism

368

0037

50

590

932

3297

87

O-H

MS

667

114

000










BIBLIOGRAPHY



NOTES













6

6

2

1 1

1

1

1

1

e at n


u er o sib in s

1

6

3

4

h 6

3

cc status sibling


42

65

31

h 1

5

2

ducation sibling 1


52

51

3

2

1

1

1

g

hb

b

b

g

g

g

g

g

h h

j

j

j

h

z

z

z

1

z

z

1

z

























Total variance 7547 (1000) 235989 (1000)






157

ode ni ion

5

51arent ed ation

7

g

z

h

hb

g

j



















11010090807060504030201000

80

70

60

50

40

30

20

10

usa76

usa61usa46

usa31

spa61

spa46

spa31net76

net61

net46

net31

sco61sco46sco31

sco16hun76

hun61

hun46

hun31

hun16eng46

eng31

eng16




1101009080706050403020100-10

80

70

60

50

40

30

20

10

usa76

usa61usa46

usa31

spa61

spa46

spa31net76

net61

net46

net31

sco61sco46sco31

sco16 hun76

hun61

hun46

hun31

hun16eng46

eng31

eng16







11010090807060504030201000

80

70

60

50

40

30

20

10

usa76usa61

usa46

usa31

spa61spa46

spa31

net76

net61net46net31

sco61

sco46sco31

sco16

hun76

hun61

hun46hun31

hun16

eng46

eng31eng16

CONCLUSIONS





1101009080706050403020100-10

80

70

60

50

40

30

20

10

usa76usa61

usa46

usa31

spa61spa46

spa31

net76

net61net46net31

sco61

sco46sco31

sco16

hun76

hun61

hun46hun31

hun16

eng46

eng31eng16




























Continued













APP

EN

DIX

C

Fit s

tatis

tics

of m

odel

s in

whi

ch p

aram

eter

s of

the

over

all b

aseli

ne m

odel

var

y ac

ross

soc

ietie

s

c2df

pG

FIB

ICC

ontr

ast

d (c2

)d (

df)

p

Mod

el O

ove

rall

base

line

mod

el +

var

ying

z1

= z 2

z3

= z 4

z5

z6

434

7238

90

060

922

3368

01

Mod

el A

M g

51va

ryin

g w

ith

mod

ern

izat

ion

432

9538

80

060

922

3660

00

O-A

M1

771

018

Mod

el A

S g

51va

ryin

g w

ith

soc

ialis

m43

406

388

005

092

232

582

9O

-AS

006

10

82

Mod

el B

M g

52va

ryin

g w

ith

mod

ern

izat

ion

430

6638

80

070

922

3362

30

O-B

M4

061

004

Mod

el B

S g

52va

ryin

g w

ith

soc

ialis

m43

431

388

005

092

233

586

5O

-BS

041

10

52

Mod

el C

M g

53va

ryin

g w

ith

mod

ern

izat

ion

434

6038

80

050

922

3358

35

O-C

M0

121

073

Mod

el C

S g

53va

ryin

g w

ith

soc

ialis

m43

469

388

005

092

233

582

6O

-CS

003

10

87

Mod

el D

M g

61va

ryin

g w

ith

mod

ern

izat

ion

433

9238

80

050

922

3259

04

O-D

M0

801

037

Mod

el D

S g

61va

ryin

g w

ith

soc

ialis

m43

135

388

006

092

233

616

1O

-DS

337

10

07

Mod

el E

M g

62va

ryin

g w

ith

mod

ern

izat

ion

434

4238

80

050

922

3358

53

O-E

M0

301

059

Mod

el E

S g

62va

ryin

g w

ith

soc

ialis

m43

439

388

005

092

233

585

7O

-ES

033

10

57

Mod

el F

M g

63va

ryin

g w

ith

mod

ern

izat

ion

434

2838

80

050

922

3358

68

O-F

M0

441

002

Mod

el F

S g

63va

ryin

g w

ith

soc

ialis

m42

964

388

007

092

233

633

2O

-FS

508

10

00

Mod

el G

M b

31=

b 42

= b 6

5va

rin

g w

ith

mod

ern

izat

ion

418

3438

80

140

932

3374

62

O-G

M16

38

10

00M

odel

GS

b31

= b 4

2=

b 65

vary

ing

wit

h s

ocia

lism

406

3338

80

250

912

3386

62

O-G

S28

38

10

00

Mod

el H

M a

ll va

ryin

g w

ith

mod

ern

izat

ion

408

6938

20

170

922

3325

61

O-H

M26

03

70

00M

odel

HS

all

vary

ing

wit

h s

ocia

lism

385

2238

20

440

912

3349

08

O-H

S49

50

70

00M

odel

HM

S a

ll va

ryin

g w

ith

mod

ern

izat

ion

an

d s

ocia

lism

368

0037

50

590

932

3297

87

O-H

MS

667

114

000










BIBLIOGRAPHY



NOTES

































Total variance 7547 (1000) 235989 (1000)






157

ode ni ion

5

51arent ed ation

7

g

z

h

hb

g

j



















11010090807060504030201000

80

70

60

50

40

30

20

10

usa76

usa61usa46

usa31

spa61

spa46

spa31net76

net61

net46

net31

sco61sco46sco31

sco16hun76

hun61

hun46

hun31

hun16eng46

eng31

eng16




1101009080706050403020100-10

80

70

60

50

40

30

20

10

usa76

usa61usa46

usa31

spa61

spa46

spa31net76

net61

net46

net31

sco61sco46sco31

sco16 hun76

hun61

hun46

hun31

hun16eng46

eng31

eng16







11010090807060504030201000

80

70

60

50

40

30

20

10

usa76usa61

usa46

usa31

spa61spa46

spa31

net76

net61net46net31

sco61

sco46sco31

sco16

hun76

hun61

hun46hun31

hun16

eng46

eng31eng16

CONCLUSIONS





1101009080706050403020100-10

80

70

60

50

40

30

20

10

usa76usa61

usa46

usa31

spa61spa46

spa31

net76

net61net46net31

sco61

sco46sco31

sco16

hun76

hun61

hun46hun31

hun16

eng46

eng31eng16




























Continued













APP

EN

DIX

C

Fit s

tatis

tics

of m

odel

s in

whi

ch p

aram

eter

s of

the

over

all b

aseli

ne m

odel

var

y ac

ross

soc

ietie

s

c2df

pG

FIB

ICC

ontr

ast

d (c2

)d (

df)

p

Mod

el O

ove

rall

base

line

mod

el +

var

ying

z1

= z 2

z3

= z 4

z5

z6

434

7238

90

060

922

3368

01

Mod

el A

M g

51va

ryin

g w

ith

mod

ern

izat

ion

432

9538

80

060

922

3660

00

O-A

M1

771

018

Mod

el A

S g

51va

ryin

g w

ith

soc

ialis

m43

406

388

005

092

232

582

9O

-AS

006

10

82

Mod

el B

M g

52va

ryin

g w

ith

mod

ern

izat

ion

430

6638

80

070

922

3362

30

O-B

M4

061

004

Mod

el B

S g

52va

ryin

g w

ith

soc

ialis

m43

431

388

005

092

233

586

5O

-BS

041

10

52

Mod

el C

M g

53va

ryin

g w

ith

mod

ern

izat

ion

434

6038

80

050

922

3358

35

O-C

M0

121

073

Mod

el C

S g

53va

ryin

g w

ith

soc

ialis

m43

469

388

005

092

233

582

6O

-CS

003

10

87

Mod

el D

M g

61va

ryin

g w

ith

mod

ern

izat

ion

433

9238

80

050

922

3259

04

O-D

M0

801

037

Mod

el D

S g

61va

ryin

g w

ith

soc

ialis

m43

135

388

006

092

233

616

1O

-DS

337

10

07

Mod

el E

M g

62va

ryin

g w

ith

mod

ern

izat

ion

434

4238

80

050

922

3358

53

O-E

M0

301

059

Mod

el E

S g

62va

ryin

g w

ith

soc

ialis

m43

439

388

005

092

233

585

7O

-ES

033

10

57

Mod

el F

M g

63va

ryin

g w

ith

mod

ern

izat

ion

434

2838

80

050

922

3358

68

O-F

M0

441

002

Mod

el F

S g

63va

ryin

g w

ith

soc

ialis

m42

964

388

007

092

233

633

2O

-FS

508

10

00

Mod

el G

M b

31=

b 42

= b 6

5va

rin

g w

ith

mod

ern

izat

ion

418

3438

80

140

932

3374

62

O-G

M16

38

10

00M

odel

GS

b31

= b 4

2=

b 65

vary

ing

wit

h s

ocia

lism

406

3338

80

250

912

3386

62

O-G

S28

38

10

00

Mod

el H

M a

ll va

ryin

g w

ith

mod

ern

izat

ion

408

6938

20

170

922

3325

61

O-H

M26

03

70

00M

odel

HS

all

vary

ing

wit

h s

ocia

lism

385

2238

20

440

912

3349

08

O-H

S49

50

70

00M

odel

HM

S a

ll va

ryin

g w

ith

mod

ern

izat

ion

an

d s

ocia

lism

368

0037

50

590

932

3297

87

O-H

MS

667

114

000










BIBLIOGRAPHY



NOTES















157

ode ni ion

5

51arent ed ation

7

g

z

h

hb

g

j



















11010090807060504030201000

80

70

60

50

40

30

20

10

usa76

usa61usa46

usa31

spa61

spa46

spa31net76

net61

net46

net31

sco61sco46sco31

sco16hun76

hun61

hun46

hun31

hun16eng46

eng31

eng16




1101009080706050403020100-10

80

70

60

50

40

30

20

10

usa76

usa61usa46

usa31

spa61

spa46

spa31net76

net61

net46

net31

sco61sco46sco31

sco16 hun76

hun61

hun46

hun31

hun16eng46

eng31

eng16







11010090807060504030201000

80

70

60

50

40

30

20

10

usa76usa61

usa46

usa31

spa61spa46

spa31

net76

net61net46net31

sco61

sco46sco31

sco16

hun76

hun61

hun46hun31

hun16

eng46

eng31eng16

CONCLUSIONS





1101009080706050403020100-10

80

70

60

50

40

30

20

10

usa76usa61

usa46

usa31

spa61spa46

spa31

net76

net61net46net31

sco61

sco46sco31

sco16

hun76

hun61

hun46hun31

hun16

eng46

eng31eng16




























Continued













APP

EN

DIX

C

Fit s

tatis

tics

of m

odel

s in

whi

ch p

aram

eter

s of

the

over

all b

aseli

ne m

odel

var

y ac

ross

soc

ietie

s

c2df

pG

FIB

ICC

ontr

ast

d (c2

)d (

df)

p

Mod

el O

ove

rall

base

line

mod

el +

var

ying

z1

= z 2

z3

= z 4

z5

z6

434

7238

90

060

922

3368

01

Mod

el A

M g

51va

ryin

g w

ith

mod

ern

izat

ion

432

9538

80

060

922

3660

00

O-A

M1

771

018

Mod

el A

S g

51va

ryin

g w

ith

soc

ialis

m43

406

388

005

092

232

582

9O

-AS

006

10

82

Mod

el B

M g

52va

ryin

g w

ith

mod

ern

izat

ion

430

6638

80

070

922

3362

30

O-B

M4

061

004

Mod

el B

S g

52va

ryin

g w

ith

soc

ialis

m43

431

388

005

092

233

586

5O

-BS

041

10

52

Mod

el C

M g

53va

ryin

g w

ith

mod

ern

izat

ion

434

6038

80

050

922

3358

35

O-C

M0

121

073

Mod

el C

S g

53va

ryin

g w

ith

soc

ialis

m43

469

388

005

092

233

582

6O

-CS

003

10

87

Mod

el D

M g

61va

ryin

g w

ith

mod

ern

izat

ion

433

9238

80

050

922

3259

04

O-D

M0

801

037

Mod

el D

S g

61va

ryin

g w

ith

soc

ialis

m43

135

388

006

092

233

616

1O

-DS

337

10

07

Mod

el E

M g

62va

ryin

g w

ith

mod

ern

izat

ion

434

4238

80

050

922

3358

53

O-E

M0

301

059

Mod

el E

S g

62va

ryin

g w

ith

soc

ialis

m43

439

388

005

092

233

585

7O

-ES

033

10

57

Mod

el F

M g

63va

ryin

g w

ith

mod

ern

izat

ion

434

2838

80

050

922

3358

68

O-F

M0

441

002

Mod

el F

S g

63va

ryin

g w

ith

soc

ialis

m42

964

388

007

092

233

633

2O

-FS

508

10

00

Mod

el G

M b

31=

b 42

= b 6

5va

rin

g w

ith

mod

ern

izat

ion

418

3438

80

140

932

3374

62

O-G

M16

38

10

00M

odel

GS

b31

= b 4

2=

b 65

vary

ing

wit

h s

ocia

lism

406

3338

80

250

912

3386

62

O-G

S28

38

10

00

Mod

el H

M a

ll va

ryin

g w

ith

mod

ern

izat

ion

408

6938

20

170

922

3325

61

O-H

M26

03

70

00M

odel

HS

all

vary

ing

wit

h s

ocia

lism

385

2238

20

440

912

3349

08

O-H

S49

50

70

00M

odel

HM

S a

ll va

ryin

g w

ith

mod

ern

izat

ion

an

d s

ocia

lism

368

0037

50

590

932

3297

87

O-H

MS

667

114

000










BIBLIOGRAPHY



NOTES



























11010090807060504030201000

80

70

60

50

40

30

20

10

usa76

usa61usa46

usa31

spa61

spa46

spa31net76

net61

net46

net31

sco61sco46sco31

sco16hun76

hun61

hun46

hun31

hun16eng46

eng31

eng16




1101009080706050403020100-10

80

70

60

50

40

30

20

10

usa76

usa61usa46

usa31

spa61

spa46

spa31net76

net61

net46

net31

sco61sco46sco31

sco16 hun76

hun61

hun46

hun31

hun16eng46

eng31

eng16







11010090807060504030201000

80

70

60

50

40

30

20

10

usa76usa61

usa46

usa31

spa61spa46

spa31

net76

net61net46net31

sco61

sco46sco31

sco16

hun76

hun61

hun46hun31

hun16

eng46

eng31eng16

CONCLUSIONS





1101009080706050403020100-10

80

70

60

50

40

30

20

10

usa76usa61

usa46

usa31

spa61spa46

spa31

net76

net61net46net31

sco61

sco46sco31

sco16

hun76

hun61

hun46hun31

hun16

eng46

eng31eng16




























Continued













APP

EN

DIX

C

Fit s

tatis

tics

of m

odel

s in

whi

ch p

aram

eter

s of

the

over

all b

aseli

ne m

odel

var

y ac

ross

soc

ietie

s

c2df

pG

FIB

ICC

ontr

ast

d (c2

)d (

df)

p

Mod

el O

ove

rall

base

line

mod

el +

var

ying

z1

= z 2

z3

= z 4

z5

z6

434

7238

90

060

922

3368

01

Mod

el A

M g

51va

ryin

g w

ith

mod

ern

izat

ion

432

9538

80

060

922

3660

00

O-A

M1

771

018

Mod

el A

S g

51va

ryin

g w

ith

soc

ialis

m43

406

388

005

092

232

582

9O

-AS

006

10

82

Mod

el B

M g

52va

ryin

g w

ith

mod

ern

izat

ion

430

6638

80

070

922

3362

30

O-B

M4

061

004

Mod

el B

S g

52va

ryin

g w

ith

soc

ialis

m43

431

388

005

092

233

586

5O

-BS

041

10

52

Mod

el C

M g

53va

ryin

g w

ith

mod

ern

izat

ion

434

6038

80

050

922

3358

35

O-C

M0

121

073

Mod

el C

S g

53va

ryin

g w

ith

soc

ialis

m43

469

388

005

092

233

582

6O

-CS

003

10

87

Mod

el D

M g

61va

ryin

g w

ith

mod

ern

izat

ion

433

9238

80

050

922

3259

04

O-D

M0

801

037

Mod

el D

S g

61va

ryin

g w

ith

soc

ialis

m43

135

388

006

092

233

616

1O

-DS

337

10

07

Mod

el E

M g

62va

ryin

g w

ith

mod

ern

izat

ion

434

4238

80

050

922

3358

53

O-E

M0

301

059

Mod

el E

S g

62va

ryin

g w

ith

soc

ialis

m43

439

388

005

092

233

585

7O

-ES

033

10

57

Mod

el F

M g

63va

ryin

g w

ith

mod

ern

izat

ion

434

2838

80

050

922

3358

68

O-F

M0

441

002

Mod

el F

S g

63va

ryin

g w

ith

soc

ialis

m42

964

388

007

092

233

633

2O

-FS

508

10

00

Mod

el G

M b

31=

b 42

= b 6

5va

rin

g w

ith

mod

ern

izat

ion

418

3438

80

140

932

3374

62

O-G

M16

38

10

00M

odel

GS

b31

= b 4

2=

b 65

vary

ing

wit

h s

ocia

lism

406

3338

80

250

912

3386

62

O-G

S28

38

10

00

Mod

el H

M a

ll va

ryin

g w

ith

mod

ern

izat

ion

408

6938

20

170

922

3325

61

O-H

M26

03

70

00M

odel

HS

all

vary

ing

wit

h s

ocia

lism

385

2238

20

440

912

3349

08

O-H

S49

50

70

00M

odel

HM

S a

ll va

ryin

g w

ith

mod

ern

izat

ion

an

d s

ocia

lism

368

0037

50

590

932

3297

87

O-H

MS

667

114

000










BIBLIOGRAPHY



NOTES
















11010090807060504030201000

80

70

60

50

40

30

20

10

usa76usa61

usa46

usa31

spa61spa46

spa31

net76

net61net46net31

sco61

sco46sco31

sco16

hun76

hun61

hun46hun31

hun16

eng46

eng31eng16

CONCLUSIONS





1101009080706050403020100-10

80

70

60

50

40

30

20

10

usa76usa61

usa46

usa31

spa61spa46

spa31

net76

net61net46net31

sco61

sco46sco31

sco16

hun76

hun61

hun46hun31

hun16

eng46

eng31eng16




























Continued













APP

EN

DIX

C

Fit s

tatis

tics

of m

odel

s in

whi

ch p

aram

eter

s of

the

over

all b

aseli

ne m

odel

var

y ac

ross

soc

ietie

s

c2df

pG

FIB

ICC

ontr

ast

d (c2

)d (

df)

p

Mod

el O

ove

rall

base

line

mod

el +

var

ying

z1

= z 2

z3

= z 4

z5

z6

434

7238

90

060

922

3368

01

Mod

el A

M g

51va

ryin

g w

ith

mod

ern

izat

ion

432

9538

80

060

922

3660

00

O-A

M1

771

018

Mod

el A

S g

51va

ryin

g w

ith

soc

ialis

m43

406

388

005

092

232

582

9O

-AS

006

10

82

Mod

el B

M g

52va

ryin

g w

ith

mod

ern

izat

ion

430

6638

80

070

922

3362

30

O-B

M4

061

004

Mod

el B

S g

52va

ryin

g w

ith

soc

ialis

m43

431

388

005

092

233

586

5O

-BS

041

10

52

Mod

el C

M g

53va

ryin

g w

ith

mod

ern

izat

ion

434

6038

80

050

922

3358

35

O-C

M0

121

073

Mod

el C

S g

53va

ryin

g w

ith

soc

ialis

m43

469

388

005

092

233

582

6O

-CS

003

10

87

Mod

el D

M g

61va

ryin

g w

ith

mod

ern

izat

ion

433

9238

80

050

922

3259

04

O-D

M0

801

037

Mod

el D

S g

61va

ryin

g w

ith

soc

ialis

m43

135

388

006

092

233

616

1O

-DS

337

10

07

Mod

el E

M g

62va

ryin

g w

ith

mod

ern

izat

ion

434

4238

80

050

922

3358

53

O-E

M0

301

059

Mod

el E

S g

62va

ryin

g w

ith

soc

ialis

m43

439

388

005

092

233

585

7O

-ES

033

10

57

Mod

el F

M g

63va

ryin

g w

ith

mod

ern

izat

ion

434

2838

80

050

922

3358

68

O-F

M0

441

002

Mod

el F

S g

63va

ryin

g w

ith

soc

ialis

m42

964

388

007

092

233

633

2O

-FS

508

10

00

Mod

el G

M b

31=

b 42

= b 6

5va

rin

g w

ith

mod

ern

izat

ion

418

3438

80

140

932

3374

62

O-G

M16

38

10

00M

odel

GS

b31

= b 4

2=

b 65

vary

ing

wit

h s

ocia

lism

406

3338

80

250

912

3386

62

O-G

S28

38

10

00

Mod

el H

M a

ll va

ryin

g w

ith

mod

ern

izat

ion

408

6938

20

170

922

3325

61

O-H

M26

03

70

00M

odel

HS

all

vary

ing

wit

h s

ocia

lism

385

2238

20

440

912

3349

08

O-H

S49

50

70

00M

odel

HM

S a

ll va

ryin

g w

ith

mod

ern

izat

ion

an

d s

ocia

lism

368

0037

50

590

932

3297

87

O-H

MS

667

114

000










BIBLIOGRAPHY



NOTES










CONCLUSIONS





1101009080706050403020100-10

80

70

60

50

40

30

20

10

usa76usa61

usa46

usa31

spa61spa46

spa31

net76

net61net46net31

sco61

sco46sco31

sco16

hun76

hun61

hun46hun31

hun16

eng46

eng31eng16




























Continued













APP

EN

DIX

C

Fit s

tatis

tics

of m

odel

s in

whi

ch p

aram

eter

s of

the

over

all b

aseli

ne m

odel

var

y ac

ross

soc

ietie

s

c2df

pG

FIB

ICC

ontr

ast

d (c2

)d (

df)

p

Mod

el O

ove

rall

base

line

mod

el +

var

ying

z1

= z 2

z3

= z 4

z5

z6

434

7238

90

060

922

3368

01

Mod

el A

M g

51va

ryin

g w

ith

mod

ern

izat

ion

432

9538

80

060

922

3660

00

O-A

M1

771

018

Mod

el A

S g

51va

ryin

g w

ith

soc

ialis

m43

406

388

005

092

232

582

9O

-AS

006

10

82

Mod

el B

M g

52va

ryin

g w

ith

mod

ern

izat

ion

430

6638

80

070

922

3362

30

O-B

M4

061

004

Mod

el B

S g

52va

ryin

g w

ith

soc

ialis

m43

431

388

005

092

233

586

5O

-BS

041

10

52

Mod

el C

M g

53va

ryin

g w

ith

mod

ern

izat

ion

434

6038

80

050

922

3358

35

O-C

M0

121

073

Mod

el C

S g

53va

ryin

g w

ith

soc

ialis

m43

469

388

005

092

233

582

6O

-CS

003

10

87

Mod

el D

M g

61va

ryin

g w

ith

mod

ern

izat

ion

433

9238

80

050

922

3259

04

O-D

M0

801

037

Mod

el D

S g

61va

ryin

g w

ith

soc

ialis

m43

135

388

006

092

233

616

1O

-DS

337

10

07

Mod

el E

M g

62va

ryin

g w

ith

mod

ern

izat

ion

434

4238

80

050

922

3358

53

O-E

M0

301

059

Mod

el E

S g

62va

ryin

g w

ith

soc

ialis

m43

439

388

005

092

233

585

7O

-ES

033

10

57

Mod

el F

M g

63va

ryin

g w

ith

mod

ern

izat

ion

434

2838

80

050

922

3358

68

O-F

M0

441

002

Mod

el F

S g

63va

ryin

g w

ith

soc

ialis

m42

964

388

007

092

233

633

2O

-FS

508

10

00

Mod

el G

M b

31=

b 42

= b 6

5va

rin

g w

ith

mod

ern

izat

ion

418

3438

80

140

932

3374

62

O-G

M16

38

10

00M

odel

GS

b31

= b 4

2=

b 65

vary

ing

wit

h s

ocia

lism

406

3338

80

250

912

3386

62

O-G

S28

38

10

00

Mod

el H

M a

ll va

ryin

g w

ith

mod

ern

izat

ion

408

6938

20

170

922

3325

61

O-H

M26

03

70

00M

odel

HS

all

vary

ing

wit

h s

ocia

lism

385

2238

20

440

912

3349

08

O-H

S49

50

70

00M

odel

HM

S a

ll va

ryin

g w

ith

mod

ern

izat

ion

an

d s

ocia

lism

368

0037

50

590

932

3297

87

O-H

MS

667

114

000










BIBLIOGRAPHY



NOTES




































Continued













APP

EN

DIX

C

Fit s

tatis

tics

of m

odel

s in

whi

ch p

aram

eter

s of

the

over

all b

aseli

ne m

odel

var

y ac

ross

soc

ietie

s

c2df

pG

FIB

ICC

ontr

ast

d (c2

)d (

df)

p

Mod

el O

ove

rall

base

line

mod

el +

var

ying

z1

= z 2

z3

= z 4

z5

z6

434

7238

90

060

922

3368

01

Mod

el A

M g

51va

ryin

g w

ith

mod

ern

izat

ion

432

9538

80

060

922

3660

00

O-A

M1

771

018

Mod

el A

S g

51va

ryin

g w

ith

soc

ialis

m43

406

388

005

092

232

582

9O

-AS

006

10

82

Mod

el B

M g

52va

ryin

g w

ith

mod

ern

izat

ion

430

6638

80

070

922

3362

30

O-B

M4

061

004

Mod

el B

S g

52va

ryin

g w

ith

soc

ialis

m43

431

388

005

092

233

586

5O

-BS

041

10

52

Mod

el C

M g

53va

ryin

g w

ith

mod

ern

izat

ion

434

6038

80

050

922

3358

35

O-C

M0

121

073

Mod

el C

S g

53va

ryin

g w

ith

soc

ialis

m43

469

388

005

092

233

582

6O

-CS

003

10

87

Mod

el D

M g

61va

ryin

g w

ith

mod

ern

izat

ion

433

9238

80

050

922

3259

04

O-D

M0

801

037

Mod

el D

S g

61va

ryin

g w

ith

soc

ialis

m43

135

388

006

092

233

616

1O

-DS

337

10

07

Mod

el E

M g

62va

ryin

g w

ith

mod

ern

izat

ion

434

4238

80

050

922

3358

53

O-E

M0

301

059

Mod

el E

S g

62va

ryin

g w

ith

soc

ialis

m43

439

388

005

092

233

585

7O

-ES

033

10

57

Mod

el F

M g

63va

ryin

g w

ith

mod

ern

izat

ion

434

2838

80

050

922

3358

68

O-F

M0

441

002

Mod

el F

S g

63va

ryin

g w

ith

soc

ialis

m42

964

388

007

092

233

633

2O

-FS

508

10

00

Mod

el G

M b

31=

b 42

= b 6

5va

rin

g w

ith

mod

ern

izat

ion

418

3438

80

140

932

3374

62

O-G

M16

38

10

00M

odel

GS

b31

= b 4

2=

b 65

vary

ing

wit

h s

ocia

lism

406

3338

80

250

912

3386

62

O-G

S28

38

10

00

Mod

el H

M a

ll va

ryin

g w

ith

mod

ern

izat

ion

408

6938

20

170

922

3325

61

O-H

M26

03

70

00M

odel

HS

all

vary

ing

wit

h s

ocia

lism

385

2238

20

440

912

3349

08

O-H

S49

50

70

00M

odel

HM

S a

ll va

ryin

g w

ith

mod

ern

izat

ion

an

d s

ocia

lism

368

0037

50

590

932

3297

87

O-H

MS

667

114

000










BIBLIOGRAPHY



NOTES






















APP

EN

DIX

C

Fit s

tatis

tics

of m

odel

s in

whi

ch p

aram

eter

s of

the

over

all b

aseli

ne m

odel

var

y ac

ross

soc

ietie

s

c2df

pG

FIB

ICC

ontr

ast

d (c2

)d (

df)

p

Mod

el O

ove

rall

base

line

mod

el +

var

ying

z1

= z 2

z3

= z 4

z5

z6

434

7238

90

060

922

3368

01

Mod

el A

M g

51va

ryin

g w

ith

mod

ern

izat

ion

432

9538

80

060

922

3660

00

O-A

M1

771

018

Mod

el A

S g

51va

ryin

g w

ith

soc

ialis

m43

406

388

005

092

232

582

9O

-AS

006

10

82

Mod

el B

M g

52va

ryin

g w

ith

mod

ern

izat

ion

430

6638

80

070

922

3362

30

O-B

M4

061

004

Mod

el B

S g

52va

ryin

g w

ith

soc

ialis

m43

431

388

005

092

233

586

5O

-BS

041

10

52

Mod

el C

M g

53va

ryin

g w

ith

mod

ern

izat

ion

434

6038

80

050

922

3358

35

O-C

M0

121

073

Mod

el C

S g

53va

ryin

g w

ith

soc

ialis

m43

469

388

005

092

233

582

6O

-CS

003

10

87

Mod

el D

M g

61va

ryin

g w

ith

mod

ern

izat

ion

433

9238

80

050

922

3259

04

O-D

M0

801

037

Mod

el D

S g

61va

ryin

g w

ith

soc

ialis

m43

135

388

006

092

233

616

1O

-DS

337

10

07

Mod

el E

M g

62va

ryin

g w

ith

mod

ern

izat

ion

434

4238

80

050

922

3358

53

O-E

M0

301

059

Mod

el E

S g

62va

ryin

g w

ith

soc

ialis

m43

439

388

005

092

233

585

7O

-ES

033

10

57

Mod

el F

M g

63va

ryin

g w

ith

mod

ern

izat

ion

434

2838

80

050

922

3358

68

O-F

M0

441

002

Mod

el F

S g

63va

ryin

g w

ith

soc

ialis

m42

964

388

007

092

233

633

2O

-FS

508

10

00

Mod

el G

M b

31=

b 42

= b 6

5va

rin

g w

ith

mod

ern

izat

ion

418

3438

80

140

932

3374

62

O-G

M16

38

10

00M

odel

GS

b31

= b 4

2=

b 65

vary

ing

wit

h s

ocia

lism

406

3338

80

250

912

3386

62

O-G

S28

38

10

00

Mod

el H

M a

ll va

ryin

g w

ith

mod

ern

izat

ion

408

6938

20

170

922

3325

61

O-H

M26

03

70

00M

odel

HS

all

vary

ing

wit

h s

ocia

lism

385

2238

20

440

912

3349

08

O-H

S49

50

70

00M

odel

HM

S a

ll va

ryin

g w

ith

mod

ern

izat

ion

an

d s

ocia

lism

368

0037

50

590

932

3297

87

O-H

MS

667

114

000










BIBLIOGRAPHY



NOTES



















BIBLIOGRAPHY



NOTES










Testing the modernization hypothesis and the socialist ...

Documents

Transcript of Testing the modernization hypothesis and the socialist ...