can this be the new generation?!

advertisement
COURSE
PARADIGMS AND
PROBLEM
SHIFTS IN
SOCIOLOGY
UNIVERSITY OF
HAIFA
TENTH AND LAST
MEETING
JANUARY 20,
2013
WHERE IS
THE NEW
GENERATION OF
RESEARCH ON
SOCIAL
STRATIFICATION
AND MOBILITY?
CAN THIS BE
THE NEW
GENERATION?!
WOUT ULTEE
RADBOUD
UNIVERSITY
NIJMEGEN
THE NETHERLANDS
I HAVE BEEN ACTIVE IN RESEARCH COMMITTEE 28
ON STRATIFICATION AND MOBILITY OF THE
INTERNATIONAL SOCIOLOGICAL ASSOCIATION
WHEN THIS OLDER PERSON MET NEWCOMERS,
THE COMPLIMENT OF NEWCOMERS OFTEN WAS
THAT ONE RECEIVES GOOD COMMENTS IN RC28
I HAVE SOMETIMES REPLIED AS FOLLOWS:
I AND OTHER OLDER RC28 MEMBERS MAY TELL
YOU THAT YOUR RESEARCH QUESTION IS
WRONG:
LATER RC28 GENERATIONS ARGUED THAT EARLIER
GENERATIONS POSED ISSUES POORLY
AS SOMEONE WHO HAS BEEN
AROUND IN RC28 FOR A LONGER TIME
I CAME TO DISTINGUISH PAPERS
ACCORDING TO GENERATIONS
FROM TIME TO TIME THINGS
HAPPENED IN RC28 THAT MARKED A
BIG BREAK WITH THE PAST
AND THESE MAIN EVENTS MADE
RESEARCH OF THE OLDER
GENERATION NOT ONLY ANTIQUATED,
BUT FIT FOR THE DUSTBIN
MY
CRITICAL
LEARNING
PERIOD IN
RC28
SOCIOLOGY
INCLUDED
THE COVER
OF A NOW
DEFUNCT
BRITISH
WEEKLY
THE COVER STEMS FROM RIGHT
BEFORE THE 1980 PUBLICATION OF
GOLDTHORPE, SOCIAL MOBILITY
AND CLASS STRUCTURE IN MODERN
BRITAIN AND OF HALSEY, HEATH &
RIDGE, ORIGINS AND DESTINATIONS
THESE STUDIES MARKED A NEW
GENERATION OF RESEARCH IN
SOCIAL MOBILITY
NEW SOCIETY DID NOT NOTE THIS
SHIFT FROM THE GLASS
GENERATION TO THE GOLDTHORPE
GENERATION
THE EASY BUT SOMETIMES MISLEADING WAY
OF RECOGNIZING GENERATIONS IN PAPERS IS
LOOKING AT METHODS OF DATA ANALYSIS
THERE ARE A LOT OF METHODS,
AND SOMETIMES A PERVASIVE SHIFT OCCURS
FROM ONE METHOD TO ANOTHER:
FIRST GENERATION
PERCENTAGES
LIPSET 1956
SECOND GENERATION LINEAR REGRESSION
DUNCAN 1967
THIRD GENERATION
GOLDTHORPE 1978
ODDS RATIO’S
FOURTH GENERATION EVENT HISTORY MODELS
BLOSSFELD 1986
FIFTH GENERATION
???
???
FIFTH GENERATION
???
FROM STANDARD LINEAR REGRESSION MODELS
TO MULTI-LEVEL MODELS IS NO BIG THING
FROM LOGLINEAR MODELS
TO MULTI-LEVEL MULTI-NOMINAL LOGISTIC
REGRESSION MODELS IS NO BIG THING
HECKMAN CORRECTION FOR SELECTION BIAS
IS NO BIG THING
IMPUTATION OF MISSING VALUES
IS NO BIG THING
GENERATIONS ARE NOT
MARKED BY NEW METHODS
THE NEW GENERATION
IMPLIES AND SOMETIMES
ANNOUNCES LOUDLY
THAT RESEARCH QUESTIONS
OF THE OLD GENERATION
POSE THE ISSUE POORLY
THE OLD GENERATION USED
METHODS THAT INCORPORATE
HYPOTHESES THAT ARE
PRECISELY AT ISSUE
LE MONDE IN 2007 SHOWED A RISE IN THE PERCENT
OF DOWNWARDLY MOBILE FRENCH PERSONS
WHY IS DOWNWARD MOBILITY INCREASING?
THIS QUESTION POSES THE ISSUE POORLY
IT IS KNOWN THAT THE PERCENT OF PEOPLE WITH
THE HIGHEST POSSIBLE ORIGIN IS GROWING
EVEN WHEN THE CHANCES OF PERSONS FROM THE
HIGHEST LEVEL TO STAY AT THAT LEVEL RISE,
THE PERCENT DOWNWARD MOBILITY MAY RISE
DUNCAN DROPPED QUESTIONS ABOUT
‘DIFFERENCE SCORES’
THE SECOND GENERATION ASKS AFTER THE
STRENGTH OF THE ASSOCIATION BETWEEN ORIGIN
AND DESTINATION
OF COURSE, IN FRANCE THE
RELATION BETWEEN ORIGIN
AND DESTINATION
MEASURED BY A CORRELATION
OR BY ODDS RATIO’S
MAY HAVE BECOME WEAKER
TOO
BUT IS THIS SOMETHING TO
DEPLORE POLITICALLY?
THERE MAY BE TWO PROCESSES
BEHIND A LOWER ASSOCIATION
FIRST, THAT PEOPLE FROM LOW
ORIGIN WHO ATTAINED A DIPLOMA
NOWADAYS DO LESS WELL
SECOND, THAT PEOPLE FROM HIGH
ORIGIN WHO DID NOT ATTAIN A HIGH
DIPLOMA ONCE DID WELL, BUT
NOWADAYS WIND UP LOWER
MOBILITY QUESTIONS ARE ALWAYS QUESTIONS ABOUT
MOVEMENTS BETWEEN TWO STEPS ON A LADDER
BUT THERE ARE A LOT OF STEPS ON A LADDER
SO THE QUESTION OF THE CHANCES OF MOVING FROM
STEP 1 TO STEP 2 DIFFERS FROM THE QUESTION OF
THE CHANCES OF MOVING FROM STEP 2 TO STEP 3
WHEN COMPUTING LINEAR REGRESSION
COEFFICIENTS, IT IS ASSUMED THAT THE ASSOCIATION
BETWEEN ALL ORIGINS AND ALL DESTINATIONS IS
UNIFORM
YET THERE MAY BE HYPOTHESES STATING THAT IT IS
MORE DIFFICULT TO MOVE FROM STEP 1 TO STEP 2
THAN FROM STEP 2 TO STEP 3
THAT IS PART OF THE REASON WHY THE THIRD
GENERATION COMPUTED A SET OF ODDS RATIO’S
MOBILITY QUESTIONS ARE ALWAYS
QUESTIONS ABOUT MOVEMENTS OF PERSONS
BETWEEN TWO POINTS IN TIME
TO AVOID MISGUIDED QUESTIONS, THE
DISTANCE BETWEEN THESE TWO POINTS IN
TIME SHOULD BE EQUAL FOR ALL
INVESTIGATED PERSONS
HOWEVER, THE STANDARD RESEARCH DESIGN
OF OLDER GENERATIONS COLLECTS FOR A
RANDOM SAMPLE OF A COUNTRY’S
POPULATION
THEIR CLASS AT AGE 14 AND THEIR CLASS AT
THE TIME OF THE SURVEY
EVERY GENERATION RAISED THE QUESTION OF
WHETHER SOCIAL MOBILITY IS STABLE
THIS QUESTION CANNOT BE ANSWERED BY DATA
FOR THE ORIGIN AT SOME FIXED AGE AND THE
CURRENT JOB OF ALL THE MEMBERS OF ONE
SOCIETY AT ONE POINT IN TIME
HOWEVER, SOME SAID THAT AT AROUND AGE 35,
‘OCCUPATIONAL MATURITY’ SETS IN
IF THIS IS THE CASE, A COMPARISON OF
COHORTS ABOVE 35 DOES THE TRICK
BUT THE MATURITY THESIS BEGS THE QUESTION:
THE AGE OF MATURITY MAY VARY BETWEEN
COHORTS
TO ANSWER QUESTIONS ABOUT
INCREASING OR DECREASING MOBILITY
WITH DATA FROM ONE SURVEY
THE FOURTH GENERATION PROPOSED
THE COLLECTION OF JOB HISTORIES
FOR ALL MEMBERS OF A SOCIETY AT
THE TIME OF THE SURVEY
DATA FOR OCCUPATIONAL
TRAJECTORIES ALLOW FOR ANSWERS
TO QUESTIONS ABOUT AGE-, COHORT
AND PERIOD EFFECTS
THE FOURTH GENERATION DID NOT ADDRESS
QUESTIONS ABOUT OCCUPATIONAL CAREERS
IT DID NOT CAME AROUND TO THE QUESTION OF
WHETHER CAREERS ARE SELF-REINFORCING:
IF A PERSON HAPPENS TO START OUT AT A
HIGHER JOB AFTER LEAVING SCHOOL WITH
SPECIFIC CREDENTIALS, WILL THAT PERSON DO
BETTER LATER ON JUST BECAUSE OF THIS
HIGHER FIRST JOB?
THAT QUESTION WAS ANSWERED BY THE
ARROW FROM FIRST TO PRESENT JOB IN
DUNCAN’S SECOND-GENERATION PATH MODEL
OF THE SOCIO-ECONOMIC LIFE CYCLE
LIFE REMAINS MESSY IN RC28
SO, AT RC28 MEETINGS I
LOOK OUT FOR NEWCOMERS
WHO DECLARE THE
QUESTIONS OF OLDER
GENERATIONS WRONG
BUT IN THE COURSE OF TIME
I HAVE HEARD QUITE A FEW
MEMBERS OF OLDER
GENERATIONS SAY THAT
QUESTIONS OF NEWCOMERS
ARE WRONG
FROM THE EXEMPLAR OF
FIRST GENERATION TO THE
EXEMPLAR OF THE THIRD
GENERATION
RC28 RAISED QUESTIONS
ABOUT MEN ONLY
ARE THESE GAPS BEING
FILLED BY QUESTIONS ABOUT
WOMEN?
DO THESE QUESTIONS
ELIMINATE THE ERRORS OF
OLDER GENERATIONS?
THE ISSUE OF WOMEN’S
ECONOMIC DEPENDENCY
IF QUESTIONS ABOUT THE NUMBER OF STEPS A
MAN ROSE ARE MISGUIDED,
QUESTIONS ABOUT WIFE’S INCOME AS A
PERCENT OF HUSBAND’S INCOME ARE
MISGUIDED TOO
THE 50-50 COUPLES ARE A MIXED BAG,
DEPENDING UPON THE AVERAGE INCOME OF
THE COUPLE
ANNEMETTE SORENSEN & SARAH MCLANAHAN
AMERICAN JOURNAL OF SOCIOLOGY I987
IT IS WISE TO RAISE A STRING OF FOUR QUESTIONS
SO AS NOT TO JUMP ON THE TRAIN OF DEPENDENCY:
1. IS THE (HOURLY) INCOME OF A WIFE HIGHER, IF HER
HUSBAND HAS A HIGHER (HOURLY) INCOME?
2. DOES A WIFE’S INCOME NOT ONLY DEPEND UPON
HET OWN EDUCATION, BUT ALSO UPON THE
EDUCATION OF HER HUSBAND?
3. DOES A HUSBAND’S INCOME NOT ONLY DEPEND
UPON HIS OWN EDUCATION, BUT ALSO UPON THE
EDUCATION OF HIS WIFE?
4. IS THERE STILL A POSITIVE ASSOCIATION BETWEEN
WIFE’S AND HUSBAND’S INCOME AFTER TAKING INTO
ACCOUNT EDUCATIONAL HOMOGAMY AND PARTNER
EFFECTS?
THE ISSUE OF THE
DIVISION OF HOUSEHOLD LABOUR
QUESTIONS ABOUT THE TIME WOMEN SPEND ON HOUSEHOLD
CHORES RELATIVE TO THE TIME THEIR HUSBAND DOES SO
ARE MISGUIDED
THERE IS A 50-50 DIVISION OF LABOUR IF BOTH PARTNERS ARE
SLUTTISH AND IF THEY BOTH HAVE A POLLUTION PHOBIA
THE PROPER SEQUENCE OF QUESTIONS:
1. ARE THE HOUSEHOLD HOURS OF HUSBANDS AND WIVES
POSITIVELY RELATED?
2. TO WHAT EXTENT DOES OWN CLASS EXPLAIN THE
HOUSEHOLD HOURS OF HUSBANDS AND OF WIVES?
3. DOES THE ASSOCIATION BETWEEN THE HOURS OF
HUSBANDS AND WIVES PERSIST AFTER TAKING INTO
ACCOUNT CLASS HOMOGAMY AND PARTNER EFFECTS?
WHEN ESTIMATING MODELS ALSO STUDY INTERCEPTS!
THE FIRST GENERATION DID NOT SUCCEED IN APPLYING ITS
OCCUPATIONAL PRESTIGE SCALES WHEN ANSWERING
MOBILITY QUESTIONS
THE SECOND GENERATION DID SUCCEED IN THIS, LEADING
TO STATUS ATTAINMENT RESEARCH
THE THIRD GENERATION OBJECTED AGAINST PRESTIGE
GRADIENTS, SINCE INDUSTRIAL SOCIETIES WOULD CONSIST
OF A LIMITED NUMBER OD DISCRETE CLASSES
HOWEVER, UNLIKE MARXIST SOCIOLOGY, THE THIRD
GENERATION DID NOT ASSERT THAT THE MIDDLE CLASSES
WERE DISAPPEARING, NOR THAT THE RELATIVE SIZE OF THE
HIGHEST CLASSES WAS GETTING SMALLER
THE CLASS SCHEMA DOMINANT AT PRESENT IN THE
EUROPEAN WING OF RC28
CONSISTS OF TEN CLASSES
BUT THESE CLASSES RIGHT NOW ARE BEING DISQUALIFIED
AS MACRO-CLASSES AND THE SCHEMA IS BEING
CHALLENGED BY THE IDEA OF MICRO-CLASSES
THIS LOOKS LIKE A RETURN TO A STATUS GRADIENT
RIGHT NOW I AM INTERESTED IN RATHER
DIFFERENT QUESTIONS AND SQUARE TABLES
I TRY TO MAKE A SEQUENCE OF QUESTIONS
THAT DRILLS DEEPER AND DEEPER
A MORE AND MORE SPECIFIC SEQUENCE OF
QUESTIONS ABOUT WOMEN:
1. IS THE PERCENT OF EMPLOYED WOMEN
APPROACHING THAT OF EMPLOYED MEN?
2. ARE WOMEN WITH A JOB UNDERREPRESENTED AMONG THE HIGHEST CLASSES
AND THE OCCUPATIONS WITH THE HIGHEST
PRESTIGE (EVEN AFTER TAKING INTO ACCOUNT
THEIR EDUCATION AND LABOUR FORCE
EXPERIENCE)?
3. ARE HIGHLY EDUCATED WOMEN UNDERREPRESENTED AMONG THE SUPERVISORS?
AFTER EACH MORE DETAILED DESCRIPTIVE QUESTION,
RC28 SOCIOLOGISTS RAISED EXPLANATORY QUESTIONS
THE ANSWERS ALMOST ALWAYS AMOUNT TO
HYPOTHESES ABOUT DISCRIMINATION
- DISCRIMINATION BY MEN
IF MEN ARE DISCRIMINATING AGAINST WOMEN, THIS
SHOULD SHOW UP IN SITUATIONS WHERE MEN AND
WOMEN DEAL WITH EACH OTHER
SO MY DEEPER QUESTIONS ARE:
4. IF HIGHLY EDUCATED WOMEN ARE SUPERVISING, TO
WHAT EXTENT ARE THEY SUPERVISING WOMEN RATHER
THEN MEN?
EVEN MORE TO THE POINT:
5. HOW MANY MEN HAVE A WOMAN AS A SUPERVISOR?
THE NETHERLANDS IN 2009
GENDER OF SUPERVISOR
GENDER OF
MALE
FEMALE
MALE
88 %
12 %
FEMALE
56 %
44 %
EMPLOYED
PERSON
THE ODDS RATIO FOR THIS TABLE:
5.8
THE HIGHER CLASSES OF THE IN RC 28 DOMINANT CLASS
SCHEMA ARE CONTAINING AN EVER HIGHER PERCENT OF
A COUNTRY’S LABOUR FORCE
FOR SEVERAL COUNTRIES NOW RC28 MEMBERS
ESTABLISHED A TREND TOWARDS MORE RELATIVE
INTERGENERATIONAL CLASS MOBILITY
FOR MOST COUNTRIES IN THE PAST DECADES THE GINI
FOR INCOME INEQUALITY HAS BEEN RISING
(HOW) CAN THESE FINDINGS BE SQUARED?
DOES THIS JUXTAPOSITION INDICATE THAT RC28 SHOULD
MOVE BEYOND THE PRESENT CLASS SCHEMA?
RATHER THAN STATUS ATTAINMENT AND MACRO OR
MICRO CLASS MOBILITY,
SHOULD RC28 STUDY INCOME MOBILITY?
THE TRENDS IN RELATIVE INTERGENERATIONAL
CLASS MOBILITY ESTABLISHED BY RC28
SOCIOLOGISTS
REFER TO THE PERIOD AFTER WORLD WAR 2
AND SOMETIMES TO THE FULL 20TH CENTURY
DEPENDING UPON THE COUNTRY, THE RISE OF
THE GINI PERTAINS TO THE LAST THREE OR
FOUR THREE DECADES
AND WHEREAS MOBILITY PERTAINS TO
PERSONS TAKEN ON THEIR OWN
GINI’S PERTAIN TO INDIVIDUALS WITHIN
HOUSEHOLDS
TO WHAT EXTENT DID THE GINI SINCE 1975
RISE BECAUSE OF
1. A STRONGER EFFECT OF MACRO CLASS ON EARNINGS?
2. A STRONGER EFFECT OF LEVEL OF EDUCATION ON
EARNINGS?
3. MORE ONE PERSON HOUSEHOLDS AND MORE LONE
MOTHERS?
4. LOWER SOCIAL SECURITY BENEFITS?
5. MORE EDUCATIONAL HOMOGAMY?
6. A STRONGER RISE IN EMPLOYMENT FOR HIGHLY AS
COMPARED WITH LOWELY EDUCATED WIVES?
7. A STRONGER EFFECT OF FIELD OF EDUCATION ON
EARNINGS?
8. A STRONGER EFFECT OF MICRO CLASS ON EARNINGS?
BETTER CANDIDATES FOR FIFTH
GENERATION RESEARCH?
ROBERT MARE’S PLEA IN
DEMOGRAPHY OF 2011 FOR A SHIFT
FROM THE TWO-GENERATION MODELS
OF ALL PREVIOUS GENERATIONS TO
THREE GENERATION MODELS
I POSIT AS A WEAK ALTERNATIVE A
SHIFT FROM ONE-PARENT TO TWOPARENTS MODELS
I HAVE HEARD IT SAID THAT THREE
GENERATION QUESTIONS ARE SUPERFLUOUS:
HOW CAN THE THIRD GENERATION HAVE A
DIRECT INFLUENCE? ANY INFLUENCE MUST BE
INDIRECT!
I HAVE HEARD IT SAID THAT THREE
GENERATION QUESTIONS CAPITALIZE ON
MEASUREMENT ERROR:
IF FATHER’S JOB LEVEL IS MEASURED WITH
ERROR, WHICH IT ALWAYS IS, THE
MEASUREMENT ERROR SHOWS UP AS AN
EFFECT OF GRANDFATHER’S JOB LEVEL
I HAVE HEARD IT SAY THAT ONE OF MARE’S
ARGUMENTS AGAINST TWO-GENERATION
MODELS IS NOT RIGHT
MARE HOLDS THAT THE MARGINAL
DISTRIBUTION FOR FATHER’S JOB OR
EDUCATION IN A SQUARE MOBILITY TABLE
DOES NOT PERTAIN TO ALL MEN LIVING AT ONE
POINT IN TIME
AND THAT FATHERS ARE MORE LIKELY TO
HAVE BEEN SAMPLED IF THEIR NUMBER OF
CHILDREN WAS HIGHER
EARLIER GENERATIONS SURELY INTERPRETED
FATHERS’ MARGINAL IN THIS WRONG WAY
I WOULD SAY THAT IT IS EASY TO
DEAL WITH THE
OVERREPRESENTATION OF MEN
WITH MORE CHILDREN:
ENTER THE NUMBER OF CHILDREN
INTO THE REGRESSION MODEL
AND I WOULD SAY THAT IT IS A GOOD
THING THAT MEN WITHOUT
CHILDREN DROP OUT
DUNCAN ALREADY SAID THAT THERE
ARE WRONG INTERPRETATIONS OF
MARGINAL FOR FATHERS IN A
MOBILITY TABLE
BUT HE CAME UP WITH A GOOD ONE
AND GOLDTHORPE WITH AN EVEN
BETTER ONE
SONS COMPETE ON THE BASIS OF
RESOURCES FOR BETTER JOBS
AND ONE OF THEIR RESOURCES,
NEXT TO EDUCATION, IS THEIR
FATHER’S JOB LEVEL OR HIS
EDUCATION
IT IS GOOD THAT MEN WITHOUT
CHILDREN DROP FROM THE
SAMPLE,
PRECISELY BECAUSE THEY HAVE
NO CHILDREN
THEIR RESOURCES ARE WASTED
SINCE THEY DO NOT HAVE
CHILDREN WHO COMPETE FOR
BETTER JOBS
IT MAY BE EASY TO EXPLAIN A
BIVARIATE RELATION BETWEEN
GRANDFATHER’S AND SON’S JOB
LEVEL,
BUT THAT DOES NOT MEAN THIS
EXPLANATION FULLY WORKS
JUST AS IT WAS POSSIBLE TO
ACCOUNT FULLY FOR DOUBLE
(UN)EMPLOYMENT IN COUPLES BY
EDUCATIONAL HOMOGAMY AND THE
EFFECT OF EDUCATION ON
(UN)EMPLOYMENT
SO, LET US GIVE THREE GENERATION
STUDIES SOME CREDIT
AND LET US WAIT UNTIL WE HAVE
RESULTS
BEFORE WE HERALD IT AS THE NEXT
GENERATION OF RESEARCH
THE FIFTH GENERATION
(SEEMINGLY) DESCRIPTIVE
QUESTION:
IS SON’S JOB (OR EDUCATION?)
NOT ONLY HIGHER IF HIS FATHER
HAD A HIGHER JOB (EDUCATION?)
TOO,
BUT ALSO IF, INDEPENDENT OF
THIS, HIS GRANDFATHER HAD A
HIGHER JOB (EDUCATION?) TOO?
THIS QUESTION BECOMES A
QUESTION LOADED WITH MORE
THEORETICAL IMPORT
WHEN VIEWED AGAINST THE
BACKGROUND OF THREE MAJOR
SOCIOLOGICAL PUBLICATIONS
FROM 1980
1980:
GOLDTHORPE
CLASS STRUCTURE AND
SOCIAL MOBILITY IN BRITAIN
1980:
COLLINS
THE CREDENTIAL SOCIETY
1979:
BOURDIEU
LA DISTINCTION
GOIN’ BACK QUESTION:
IS BRITAIN STILL A CLASS SOCIETY?
WHAT’S IT GONNA BE QUESTION:
WILL CREDENTIALS BECOME MORE AND MORE
IMPORTANT IN TECHNOLOGICALLY HIGHLY
DEVELOPED SOCIETIES LIKE THE USA?
SHIFTING MECHANISMS QUESTION:
LESS REPRODUCTION OF INEQUALITY BY THE
TRANSMISSION OF MATERIAL RESOURCES?
MORE REPRODUCTION OF INEQUALITY THROUGH
THE TRANSMISSION OF CULTURAL RESOURCES?
AND MORE RECONSTITUTION OF INEQUALITY
THROUGH AN INCREASE IN EDUCATIONAL
HOMOGAMY?
IF TECHNOLOGICALLY HIGHLY DEVELOPED
SOCIETIES,
AS INDICATED BY FATHER-SON
INTERGENERATIONAL MOBILITY,
ARE LESS AND LESS CLASS SOCIETIES
AND MORE AND MORE CREDENTIAL
SOCIETIES
THEN
THE QUESTION OF CREDENTIAL MOBILITY
GAINS THE UPPERHAND OVER THE
QUESTION OF CLASS MOBILITY OR
PRESTIGE MOBILITY
BREAKING DOWN THE QUESTION OF
INTERGENERATIONAL CLASS MOBILITY
INTO THE QUESTION OF THE EFFECT OF CLASS
ORIGINGS ON EDUCATION
AND THE QUESTION OF THE EFFECT OF EDUCATION
ON CLASS DESTINATION
DOES NOT GO FAR ENOUGH:
INTERGENERATIONAL CREDENTIAL MOBILITY IS
INTERESTING IN ITSELF
MULTIVARIATE MODELS HIGHLIGHT DIRECT (NET)
EFFECTS, WHEREAS TOTAL EFFECTS ARE MORE
IMPORTANT
AND SINCE MODELS OFTEN ARE NOT RECURSIVE,
INDIRECT EFFECTS REMAIN UNDERSTUDIED
THE TWO SMALLER QUESTIONS ARE QUITE DIFFERENT:
WHEREAS THE QUESTION OF SOCIAL ORIGIN AND
EDUCATIONAL ACHIEVEMENT
IS ABOUT FAMILIES AND STATES
THE QUESTION ABOUT EDUCATION AND DESTINATION
IS ABOUT THE SHORT-TERM EFFECTS OF CHANGES IN THE
BUSINESS CYCLE
AND ABOUT LONG-TERM EFFECTS OF TECHNOLOGICALLY
DRIVEN CHANGES IN OCCUPATIONAL STRUCTURES
BY DONATING AND BEQUESTING WEALTH TO
THEIR CHILDREN, PARENTS ONCE
REPRODUCED CLASS INEQUALITY
BY SPENDING TIME AND DOING THINGS WITH
THEIR CHILDREN, PARENTS NOWADAYS
REPRODUCE CREDENTIAL INEQUALITY
SINCE PARENTS NOWADAYS LIVE LONGER, THE
TRANSMISSION OF MATERIAL RESOURCES IS
DELAYED
WHEREAS MORE CULTURAL RESOURCES ARE
BEING TRANSMITTED
OUR OWN HYPOTHESIS, RENDERING
THREE-GENERATION QUESTIONS
INTERESTING:
THE INCREASE IN LONGEVITY IS SO STRONG
THAT NOT ONLY THE LIFE SPAN OF PARENTS AND
CHILDREN OVERLAP MORE
THE LIFE SPAN OF THESE CHILDREN AND THE LIFE SPAN
OF THEIR GRANDPARENTS ALSO OVERLAP MORE
IN CONTEMPORARY SOCIETIES THERE WILL BE A
STRONGER DIRECT EFFECT OF THE EDUCATION OF
GRANDPARENTS (FIRST GENERATION) ON EDUCATION IN
THE THIRD GENERATION (CHILDREN)
FIRST
GENERATION
1862 – 1950
1947
1880 - 1960
1955
SECOND GENERATION
1907 - 1979
1910 - 1983
1937
THIRD GENERATION
1946 - ??
1976
FOR CHILDREN TO BE BORN, A MOTHER
AND A FATHER ARE NECESSARY
THEREFORE INTERGENERATIONAL
QUESTIONS AS FATHER-SON QUESTIONS
MAY BE OFF THE MARK
CREDENTIALS IN THE THIRD GENERATION
DEPEND UPON CREDENTIALS OF MOTHERS
AND FATHERS IN THE SECOND GENERATION
THE HIGHER THE CREDENTIALS OF A
SECOND GENERATION HUSBAND, THE
HIGHER THE CREDENTIALS OF HIS WIFE
NET OF THIS, THE CREDENTIALS OF HIS
WIFE ALSO MAY BE HIGHER IF HIS FATHER
(FIRST GENERATION) HAS HIGHER
CREDENTIALS TOO
FIRST
SECOND
???
???
SPOUSE
SECOND
THIRD
SO, THE TOTAL EFFECT OF
CREDENTIALS IN THE FIRST
GENERATION ON THE CREDENTIALS
OF THE THIRD GENERATION MAY
CONSIST OF
1) A DIRECT EFFECT PLUS
2) AN INDIRECT EFFECT BY WAY OF
THE SECOND GENERATION PLUS
3) AN INDIRECT EFFECT BY WAY OF
THE SECOND GENERATION SPOUSE
IF NO DIRECT EFFECT OF THE FIRST
ON THE THIRD GENERATION IS
FOUND,
THIS DOES NOT MEAN THAT THERE
IS NO TOTAL EFFECT OF THE FIRST
ON THE THIRD GENERATION
THE SECOND ??? IS A !!!
FOR THE NETHERLANDS WE HAVE INDEED
FOUND THAT A MAN (WOMAN) HAS A
SPOUSE WITH A HIGHER LEVEL OF
EDUCATION
IF THAT PERSON HAS A HIGHER LEVEL OF
EDUCATION
AND, INDEPENDENT OF THIS, IF THE
FATHER OF THIS PERSON HAS A HIGHER
LEVEL OF EDUCATION
SEE DE GRAAF, SMEENK, ULTEE & TIMM, THE WHEN
AND WHOM OF FIRST MARRIAGE IN THE NETHERLANDS
IN (EDITORS) BLOSSFELD & TIMM, WHO MARRIES
WHOM?, 2003
THE FINDINGS WERE MADE BY COMBINING TWO
SURVEYS CONDUCTED IN THE NETHERLANDS, IN 1992
AND 1998
WARREN & HAUSER 1997 DID NOT FIND A DIRECT
EFFECT OF THE FIRST GENERATION ON THE THIRD
GENERATION FOR PERSONS LEAVING HIGH
SCHOOL IN 1957 IN WISCONSIN, USA
EROLA & MOISIO 2007 FOUND ALMOST NO DIRECT
OF GRANDPARENTS’ CLASS ON CHILDREN’S
CLASS FOR FINLAND 1950-2000
CHAN IN 2011 AT THE SPRING MEETING OF RC28 IN
ESSEX AND THE SPRING MEETING OF RC28 IN
HONG KONG
PRESENTED FINDINGS ON THE UNITED KINGDOM
SHOWING THAT
FIRST GENERATION CLASS HAS A DIRECT EFFECT
ON THIRD GENERATION CLASS, NET OF SECOND
GENERATION CLASS AND EDUCATION
(IN SAMPLES OF SOME 2,000, 6,000 AND 4,000
THIRD GENERATION PERSONS)
AT THE SUMMER 2011 MEETING OF RC28 IN IOWA CITY,
USA
MARE PLEADED IN FAVOR OF THREE-GENERATION
QUESTIONS
MARE ACCEPTED THAT IN TECHNOLOGICALLY HIGHLY
DEVELOPED SOCIETIES
THE THIRD GENERATION HAS NO DIRECT EFFECT IN
THE BROAD MIDDLE OF THE SOCIAL SCALE,
BUT DOES HAVE EFFECTS AT THE TOP AND BOTTOM
THESE EFFECTS ARE INTERACTIONS, THE FIRST AND
SECOND GENERATION DO MORE IN COMBINATION
THAN EACH GENERATION DOES ON ITS OWN
MARE ALSO WANTS TO STUDY THE SIZE OF THE
EFFECT OF THE FIRST GENERATION ON THE THIRD
GENERATION IN CHINA, BEFORE THE ADVENT OF
INDUSTRIALISM THE MOST DEVELOPED AGRARIAN
SOCIETY THE WORLD EVER KNEW
MY HYPOTHESIS HERE IS THAT IN THE
CONTEMPORARY NETHERLANDS,
BECAUSE OF THE GENERAL INCREASE
IN LONGEVITY,
THE TOTAL EFFECT OF FIRST
GENERATION EDUCATION ON THIRD
EDUCATION IS GROWING
CHILDREN NOWADAYS LIVE WITH THEIR
PARENTS, WITH THEIR GRANDPARENTS
LIVING APART
BUT CHILDREN VISIT THEIR
GRANDPARENTS
AND THE GRANDPARENTS DO THEIR
BEST TO MAKE THIS STAY ENJOYABLE
FOR THEIR GRANDCHILDREN
FOR THE CHILDREN THIS STAY IS
QUALITY TIME
OUR DATA PERTAIN TO THE
NETHERLANDS
AND WERE COLLECTED IN FIVE
SIMILAR SURVEYS CALLED THE
FAMILY SURVEY DUTCH
POPULATION
ABBREVIATED IN DUTCH FNB
IN SOUTH AFRICA THIS IS A BANK
YEARS OF SURVEY:
1992
1998
2000
2003
2009
IN EACH SURVEY YEAR ABOUT
1,000 PRIMARY RESPONDENTS
WERE INTERVIEWED
IF THEY WERE OFFICIALLY
MARRIED OR LIVING IN A
COMMON-LAW MARRIAGE,
THEIR PARTNERS WERE
INTERVIEWED TOO
IF A PARENT WAS STILL ALIVE,
THAT PARENT RECEIVED A MAIL
QUESTIONNAIRE
IF A CHILD WAS 25 YEARS OR
OLDER, THAT CHILD RECEIVED
A MAIL QUESTIONNAIRE
ALL PRIMARY RESPONDENTS (SECOND
GENERATION) WERE ASKED AFTER THE
EDUCATION OF THEIR FATHER AND MOTHER (FIRST
GENERATION)
THE PRESENT SPOUSE OF EACH PRIMARY
RESPONDENT WAS ASKED AFTER THE EDUCATION
OF HER OR HIS FATHER AND MOTHER
ALL PRIMARY RESPONDENTS WERE ASKED AFTER
THE EDUCATION OF ALL THE CHILDREN (THIRD
GENERATION) THEY HAVE WITH THEIR PRESENT
SPOUSE
IN THE MAIL QUESTIONNAIRE CHILDREN (THIRD
GENERATION) AND PARENTS (FIRST GENERATION)
WERE ASKED AFTER THEIR EDUCATION
SO, WE HAVE TWO MEASURES FOR THIRD
GENERATION AND FIRST GENERATION EDUCATION
WE WILL ASSUME IN LATER RESEARCH THAT
PROXY MEASURES ARE LESS ACCURATE THAN
FIRSTHAND MEASURES
WE WILL NOT ESTIMATE FIRSTHAND
STATISTICAL MODELS HERE
AT LEAST IN OUR FIRST SURVEY YEAR WE UNDERSTANDABLY - DID NOT YET HAVE ENOUGH
SECOND GENERATION CASES
WITH DOUBLE MEASURES FOR THE FIRST
GENERATION AND FOR THE THIRD GENERATION
FROM OUR FIVE SURVEYS, WITH SOME 5,000
PRIMARY RESPONDENTS
WE EXTRACTED 2,400 CHILDREN 25 YEARS OR
OLDER
SOMETIMES MORE THAN ONE CHILD PER PARENT,
(WE APPLIED A CLUSTER CORRECTION TO THE
STANDARD ERRORS)
FOR THESE PERSONS WE HAVE VALID SCORES FOR
OWN EDUCATION,
FATHER’S EDUCATION,
MOTHER’S EDUCATION,
EDUCATION OF FATHER’S FATHER,
EDUCATION OF FATHER’S MOTHER,
EDUCATION OF MOTHER’S FATHER, AND
EDUCATION OF MOTHER’S MOTHER
RESULTS OF A TABULAR ANALYSIS
THERE IS A SIGNIFICANT TOTAL EFFECT
OF GRANDFATHER’S EDUCATION ON
CHILD’S EDUCATION
WHEN HOLDING CONSTANT FOR
FATHER’S EDUCATION, FOR THOSE WITH A
HIGHLY EDUCATED FATHER, THERE IS NO
EFFECT OF GRANDFATHER’S EDUCATION
ON CHILD’S EDUCATION
FOR THOSE WITH A LOWLY EDUCATED
FATHER, THERE IS A POSITIVE EFFECT OF
GRANDFATHER’S EDUCATION ON CHILD’S
EDUCATION
SO, THERE IS AN INTERACTION EFFECT,
BUT NOT THE ONE PREDICTED BY MARE
Table 3. Relationship between education of grandparents and child by education parents (row %)
Father
Father's father
No tertiary
Tertiary
Child
Child
No tertiary
Tertiary
No tertiary
Tertiary
No tertiary
70
30
31
69
Tertiary
53
47
28
72
Odds ratio's: 2.08 and 1.12
Mother
Mother's father
No tertiary
Tertiary
Child
Child
No tertiary
Tertiary
No tertiary
Tertiary
No tertiary
65
35
29
71
Tertiary
54
46
28
72
Odds ratio's: 1.61 and 1.02
Source: Family Survey Dutch Population, 1992, 1998, 2000, 2003 and 2009
THEN WE ESTIMATED LINEAR REGRESSION MODELS
WE RECODED LEVELS OF EDUCATION ACCORDING THE
NUMBER OF YEARS PEOPLE MINIMALLY STUDY FOR A
CREDENTIAL
WE PLAYED AROUND BY ESTIMATING TWELVE MODELS
WHEN CONSIDERING ALL CHILDREN, THERE IS A EFFECT
OF FATHER’S FATHER’S EDUCATION, NET OF FATHER’S
EDUCATION
BUT IT BECOMES INDIRECT UPON THE INTRODUCTION
OF MOTHER’S EDUCATION
THIS FINDING REPEATS ITSELF FOR WOMEN,
BUT NOT FOR MEN
IN THE SEPERATE COHORTS NO SIGNIFICANT EFFECTS
ARE FOUND OF FATHER’S FATHER’S EDUCATION NET OF
FATHER’S EDUCATION
FINALLY WE ESTIMATED LOGISTIC
REGRESSION MODELS WITH THE ODDS
OF TERTIARY CREDENTIALS VERSUS
LOWER CREDENTIALS AS THE
PHENOMENON TO BE PREDICTED
WE PLAYED ONCE MORE AROUND
WITH MODELS
FOR ALL THIRD GENERATION
PERSONS, THERE IS NO DIRECT
EFFECT OF FATHER’S FATHER’S
EDUCATION ON CHILD’S EDUCATION,
THAT DIRECT EFFECT IS NOT THERE
EITHER FOR MALE CHILDREN OR
FEMALE CHILDREN SEPERATELY
WHEN CONSIDERING THE EARLIER COHORT,
THERE IS AN INSIGNIFICANT TOTAL EFFECT OF
FATHER’S FATHER’S EDUCATION
IN THE LATER COHORT THERE IS A SIGNIFICANT
TOTAL EFFECT OF FATHER’S FATHER’S
EDUCATION
AND A DIRECT EFFECT OF FATHER’S FATHER’S
EDUCATION, NET OF FATHER’S EDUCATION
WHEN BRINGING IN MOTHER’S EDUCATION, THE
EFFECT OF FATHER’S FATHER’S EDUCATION
BECOMES ALMOST FULLY INDIRECT
THIS TELLS ABOUT INDIRECT EFFECTS OF FIRST
GENERATION EDUCATION BY WAY OF THE
EDUCATION OF THE SECOND GENERATION
SPOUSE
THAT IS WHY I SAY THAT IT IS BETTER TO SHIFT
FROM ONE-PARENT TO TWO-PARENTS MODELS
THAN FROM TWO-GENERATIONS TO THREEGENERATIONS MODELS
THERE MAY BE NO DIRECT THIRD
GENERATION EFFECT
IF THE TWO GENERATION MODEL IS
NOT A ONE-PARENT MODEL
BUT A TWO-PARENTS MODELS
THIS PRESENTATION OBVIOUSLY IS A REPORT
ON WORK IN PROGRESS
WE STILL WANT TO ESTIMATE RECURSIVE
MODELS
WE HAVE TO DEVISE A BETTER TEST OF
CHANGES IN THE COURSE OF TIME BY
RECODING YEAR OF BIRTH
NOT INTO TWO CATEGORIES,
BUT BY TAKING IT IS A METRIC (PERHAPS
ACCORDING TO SOME MEASURE FOR LIFE
EXPECTANCY)
WE STILL HAVE TO APPLY CORRECTIONS BY
REPLACING PROXY MEASURES FOR THE FIRST
AND THIRD GENERATION BY FIRSTHAND
MEASURES (BUT WE DO NOT HAVE THEM FOR
EVERYONE IN OUR SAMPLE)
IF THERE IS MEASUREMENT
ERROR IN FATHER’S EDUCATION
WILL IT GO INTO FATHER’S
FATHER’S EDUCATION
IF FATHER’S FATHER’S
EDUCATION IS MEASURED WITH
MORE ERROR THAN FATHER’S
EDUCATION?
Table 6. Measurement error in education, multiple measurements
Corre- SimilaAnswer
Answer
Answer
lation
rity in
according to
according to
according to between answers
parent
child
grandparent answers
(%) N
Years Tertiary Years Tertiary Years Tertiary
Child
Father's
father
Father's
mother
Mother's
father
Mother's
mother
Years Tertiary
14.1
41
13.8
36
-
-
0.83
92 133
10.9
16
-
-
10.9
15
0.74
87 126
8.6
1
-
-
8.4
2
0.72
99 206
11.0
20
-
-
10.2
13
0.76
91 113
8.6
3
-
-
8.7
2
0.68
98 267
Source: Family Survey Dutch Population, 1998
blue if difference between answers is significant at p < 0.01; red if difference is
significant at p < 0.05
AS YET, I DO NOT FEEL
HOPELESS ABOUT MY
BOURDIEU HYPOTHESES
ALL MY POWERPOINTS FROM HAIFA WILL BE
POSTED ON MY WEBSITE
TYPE IN GOOGLE
WOUT ULTEE
CLICK THE FIRST HIT,
GO ON THE LEFT TO PRESENTATIONS, THEN TO
FOREIGN PRESENTATIONS
THERE YOU WILL FIND THE POWERPOINTS
Download