Using survey data to research family relationships

1
Using survey data to research family relationships

• Angela Dale
• University of Manchester

2
Why would you?

• Are family relationships well captured by
  surveys?
• Yes
• and
• No

3
The yes bits

• Surveys are good at capturing family structure
  and composition
• Especially when the family is defined as being
  co-resident
• Asking for information on everyone living
  together and their relationships
• Many surveys use a household matrix to capture
  the relationship of each person to all others in
  the household

4
Making linkages
5
Relationships of other household members to
person 1 (hholds with gt1 person) How are the
people in your household related to each other? 1
Spouse 4,905 2 Cohabitee 892 3 Son
or daughter 6,319 4 Step-son/daughter
272 5 Foster child
11 6 Son/daughter-in-law
42 7 Parent/guardian 99 8 Step-parent
3 9 Foster-parent 0 10
Parent-in-law 23 11 Brother/sister
86 12 Step-brother/sister 2 13
Foster brother/sister 0 14
Brother/sister-in-law 13 15
Grandchild 144 16 Grand-parent
5 17 Other relative
32 18 Other non-relative 337 19 Civil
Partner 6 Notes
adopted children are included with natural
children. Source GHS 2006, R01 relationship
to person 1
6
Household type F (grouped) GHS, 2006 1 1 person
only 2,722 2 2 unrelated adults
549 3 married couple with dependent
children 7,267 4 married couple, independent
children 1,661 5 married couple, no
children 5,208 6 lone parent, dependent
children 1,937 7 lone parent, independent
children 579 8 2 families
460 9 same sex cohabitees
71 10 cohabiting couple, dependent children
1,265 11 cohabiting couple, independent
children 53 12 cohabiting
couple, no children 1,054
Valid cases 22924 Expands to 57 categories

7
More yes bits

• Understanding family processes is crucial to much
  social science
• Effects of parental characteristics on child
  outcomes, eg educational attainment, obesity
• Social mobility and the role of parental
  networks, informal support
• Informal caring
• Survey evidence provides one, but only one,
  important kind of knowledge on these topics

8
Some questions

• How many women aged 16-59 in Britain lived in
  three-generational households in 2001?
• a) 2 b) 10 c) 20
• In 2001 were lone parents with dependent children
  more likely to be
• Single, b) divorced, c) widowed,d) separated or
  e) married?

9
Answers

• How many women aged 16-59 in Britain lived in
  three-generational households in 2001?
• a) 2 b) 10 c) 20
• In 2001 were lone parents with dependent children
  more likely to be
• a) Single (39), b) divorced (35), c) widowed
  (5) d) separated (18), e)married (3)?

10
Answers - Bangladeshis

• How many women aged 16-59 in Britain lived in
  three-generational households in 2001?
• a) 2 b) 10 c) 20
• In 2001 were lone parents with dependent children
  more likely to be
• a) Single (5), b) divorced (12), c) widowed
  (39) or d) separated (23) married (30)?

11
Who are you interested in?

• Surveys allow you to explore
• Children - how do childrens educational outcomes
  relate to the number of siblings they have?
• Partnerships movement in and out of
  partnerships
• the elderly do they live alone or with other
  kin? Is this changing? How does it vary with
  socio-economic characteristics? With ethnicity?
• Women how is womens employment influenced by
  that of their partner?

12
The benefits of surveys

• You can use large good quality surveys someone
  else has collected
• You can get a nationally representative picture
  of the topic of interest
• You can make comparisons between groups based on
  the same questions
• You can make your own definitions of family but
  depends on questions asked and answers recorded

13
Family relationships the no bits

• Surveys set their own agenda and ask everyone the
  same questions
• Although good pilot work finds out salient issues
  from respondents you cannot pursue interesting
  answers you cannot ask why and how and what
  does it feel like
• Respondents do not have their own voice
• Some surveys provide space for write-ins

14
Family relationships the no bits

• The family is the site of conflict, intimacy,
  affection, rivalry ..
• Surveys cannot capture the depth of these
  relationships
• Surveys are not good at capturing ambivalence

15
Getting the best of both worlds?

• Can one combine the strengths of surveys with the
  strengths of qualitative methods?
• Yes, there are many good examples, but dont
  forget that different methods produce different
  knowledge

16
Surveys can provide a context for qualitative
research

• background information about the locality of a
  study
• Information on the demographic structure of a
  group, or their socio-economic characteristics
• Published data or simple on-line extraction -
  can often give you exactly what you need

17
Surveys can identify questions which need
qualitative research

• Surveys often throw up puzzles which need deeper,
  qualitative work to answer
• eg why do many Pakistani and Bangladeshi women
  have low levels of economic activity?
• Survey analysis can help to target a theoretical
  sample for deeper and richer analysis
• Eg a comparison between key groups of Pakistani
  and white women
• Analysis can be integrated, eg
• survey analysis-gt quali work -gt better model with
  survey data

18
Exploring differences

• Surveys and qualitative methods can give
  conflicting accounts
• Different methods generate different kinds of
  knowledge, so not surprising
• But deeper interrogation using both methods can
  be very valuable
• Can lead to greater understanding than either
  methods along
• Neither method is right or wrong but each can
  generate important questions for the other

19
How to have a go

• The Economic and Social Data Service provide
  access and support to many different surveys
• Cross-section and longitudinal
• Wide range of topics
• On-line quick and easy access

20
www.esds.ac.uk

• Introductory workshops
• Web-based resources
• Getting started Guides
• Guides to SPSS and STATA
• Methodological Guides
• Helpdesk