‘family ethnicity’ should we count? how should we count? why? robert didham (statistics nz) with...
Post on 25-Dec-2015
217 Views
Preview:
TRANSCRIPT
‘Family ethnicity’
Should we count?How should we count?
Why?
Robert Didham (Statistics NZ) with
Paul Callister (Victoria University of Wellington), Deb Potter (Statistics NZ )
Ethnicity (NZ Census 2006)
People of one or more (N=4million)….
European ethnicities: 68 percent
Maori ethnicity: 15 percent
Pacific ethnicities: 7 percent
Asian ethnicities: 9 percent
MELAA ethnicities: 1 percent
Other (mostly NZer): 11 percent
MELAA = Middle Eastern, Latin American and African
Multiple ethnicity
People with ethnicities in at least one other grouping ….
European ethnicities: 12 percent
Maori ethnicity: 47 percent
Pacific ethnicities: 30 percent
Asian ethnicities: 9 percent
MELAA ethnicities: 17 percent
Other (mostly NZer): 13 percent
Context for the analysis
• Ethnicity is becoming more complex and fluid – strong age component
• Ethnic intermarriage is common between all groups
• Family structures increasingly complex
Individual’s Ethnicities
Cultural identity
Ancestry
Birthplace
Social histories
Family ethnicity??
Social environment
The central issue
Aim of study – three themes
• What are the uses made of ethnic family/household measures?
• Are these common enough uses to warrant recommended measure(s) and, if so, what?
• How different are results are obtained using different methods of classifying the ethnicity of families?
Not an aim of the study
• To create an official standard for the measurement of family ethnicity
Nor to define the problematic term
“FAMILY”
Family ethnicity
Implicit assumption that family ethnicity exists:
Old proverb:
There is but one eye of the needlethrough which must passthe white,the black andthe red threads
Use of family ethnicity by policy makers, researchers and the wider
public
• terms such as “Maori family” or “Pacific household” are commonly used in public discourse
• policy makers sometimes use this language (e.g. in relation to child or family outcomes)
• researchers also sometimes use these descriptions
BUT• definitions generally not clearly stated
Currently Statistics New Zealand say “ethnicity is a personal attribute that could not be attributed to a group and should not be a standard output”
Possible uses of family ethnicity• Understanding
– Family needs/outcomes– Language retention– Inequalities– Settlement outcomes
• Targeting – Health services– Housing demand and supply
BUT• Often the full complexity of ethnicity within families precludes one single
measure
AND• If the measure is flawed, can the analysis be valid?
Some specific methodological challenges
• Ethnic data may not be collected for all (or any) family members
• Responses may be proxied (i.e. not self-defined)• Small sample sizes of most surveys• How to handle multiple responses for individuals• The “New Zealander” response, especially in 2006
census• Outcomes measures for families can also be complex• The legacy of prioritisation in some data collections
More challenges• What about:
– household members not in the family?– family members living outside the household?
• What to do about missing ethnic responses?
• Family ethnicity is not directly measured but is derived from individual responses.
• Derived family ethnicity may not accord with how family members may align themselves.
Methodologies tested
• Total counts (recommended for individuals)• Main single and multiple counts
(recommended for individuals)• Ethnic prioritisation (now not recommended)• Randomly allocate ethnic prioritisation• Fractional ethnicity (non-standard method)• Base the ethnicity of the family on
– the ethnicities of the child(ren)– the ethnicities on the parent(s)
Total counts
• Well suited to small surveys as it reports the size of each group and has fewer categories
• A significant number of families are counted more than once
• Some statistical analysis techniques are difficult
• Disguises diversity within the group
Single and multiple combination counts
• Can reflect the complexity of ethnicity within families • No double counting• But system creates a very large number of possible
combinations • No weighting is put on how often an ethnicity is
recorded within a family• It is possible to calculate corresponding total counts
from these data• But represents combinations of total counts –
combinations of combinations not workable.
Prioritisation
• Ethnic counts equal counts of the total population
• Order of prioritisation based on political not statistical logic
• Big losses to some groups, particularly Pacific people living in families
• Not recommended as output for individuals• False sense of clear ethnic boundaries
Random prioritisation
• Ethnic counts equal counts of the total population
• It is statistically neutral (i.e does not favour one group over another)
• Introduces unstable results at high levels of disaggregation and in multiway cross-tabulations
Fractional ethnicity
• Appears to indicate ‘strength’ of ethnicity
• But people, or families, are not ‘fractions’
• Only one group can be easily analysed at a time (for example ‘European families)
Ethnicity of only parents or only children
• Can estimate family ethnicity from one response (such as through parent benefit data)
• Do not know full complexity of family so will miss out some ethnic groups
• Is it the parents or children who influence family ethnicity?
• In couples does one focus on both partners or just one partner (and which one)?
• Same issues of whether to count the ethnicity by total counts or other methods
Two hypothetical families:both a couple with one child
• Family 1– Maori only mother – European only father – European and Maori child
• Family 2– Pacific and Maori mother – Pacific and European father– Pacific only child
(neither of these examples are unusual)
Output counts using main methodsTotal count Single and
combinationPrioritised Random
prioritisation
(on average, different each time)
Fractional ethnicity for 1st and 2nd Families
2 Maori families
1 Maori and European family
2 Maori families
0.67 Maori families
Fam 1 : 0.5
Fam 2 : 0.17
2 European families
1 Maori, Pacific and European family
0.67 European families
Fam 1: 0.5
Fam 2 : 0.17
1 Pacific family
0.67 Pacific families
Fam 1 : 0.0
Fam 2 : 0.67
Number of couples with children - 2001
All data (000s)
Total response
Prioritisation Main single and multiple
European 282 220 European 220Maori 63 63 Maori/ European 40Asian 33 30 Asian 23Pacific 30 22 Pacific 14
Maori 14Pacific/
European7
Asian / European 7Maori / Pacific/Eur 5Maori /Pacific 3
Average income for couples with children 2001
All data (000s)
Total response
Prioritisation Main single and multiple
European 65,241 67,942 European only 67,942Maori 51,290 51,254 European/Asian 61,420Asian 49,276 48,795 Asian/Maori/Eur 56,520Pacific 45,676 44,423 Pacific/European 56,464
Maori/European 54,191Maori/European/Pacific
51,166
Asian only 44,925Maori/Pacific 43,823Maori only 42,844Pacific only 36,807
Conclusions – back to basics
• SHOULD we develop a family ethnicity measure? – perhaps, for well defined specific tasks.
• HOW? – sensibly, relevantly and explicitly defined.
• WHY? – helpful, but may also be misleading.
Tentative conclusions
• Prioritised ethnicity is problematic for individuals and this is even more so in a family setting
• Fractional ethnicity looks superficially attractive for some types of analysis but is based on some very problematic assumptions
• Using the ethnicity of only a parent or child might be workable in some contexts
• Total counts and single/multiple ethnicity have their strengths and weaknesses – but by far the best options
Ultimately…
• In the end it is ‘horses for courses’ – depends on the end use
• BUT it is important that researchers and policy makers always clearly set out how they have defined family ethnicity. It does make a difference
Contact
• Robert Didham– Robert.Didham@stats.govt.nz
• Paul Callister– Paul.Callister@vuw.ac.nz
top related