Prsentation PowerPoint

1
Surveys among difficult to reach populations
from theory to practice and reality Martine
Quaglia, Géraldine Vivier National Institute of
Demographic Studies Paris Survey Department
2
Homeless issue in France
Statisticians underline the limits of the
census(1950s)Increasing social demand (1954
Abbé Pierres call, 70 s economic crisis and
labor market changes)Creation of a special
working group by the National Council for
Statistical Information to focus on homelessness
(1993)
3
Surveys among homeless people, a methodology
adapted from

• American experiences surveys conducted from
  1984 to 1996 (Rand Corporation, Urban Institute,
  RTI, Census Bureau) European surveys from 1990
  2002 Including French surveys from 1995
  2002 1995 INED - Marpsat Firdion,
  Paris1996 Élan Retrouvé / INED - Kovess et
  Mangin-Lazarus, Paris1998 INED - Marpsat
  Firdion Young people aged 16 - 25, Paris
  suburbs2001 INSEE / INED - Brousse, Massé,
  Guiot de la Rochère National survey2002 INED -
  Marpsat Paris suburbs, Non-French
  speakers2002 INED - Marpsat Paris suburbs,
  rough sleepers met by outreach services.

4
A methodology also transferred and adapted to
Other populations such as drug users - 2002
InVS / Ined - J. Emmanuelli, M.
Jauffret-Roustide, feasibility study conducted in
Marseille (southern France) - 2004 InVS /
Ined J. Emmanuelli, M. Jauffret-Roustide,
extension to 5 French cities
5
A methodological challengeand its implications
(1)
No reliable sampling frame ? Necessary
mediation of services designed for the homeless
population shelters, soup kitchens, day centers
to design the sample and collect the data.
6
A methodological challengeand its implications
(2)
Multi stage sampling design stage 1 sampling
of cities (for a national survey) stage 2
sampling of centers and days of survey, stage
3 sampling of individuals through the service
they use (beds, meals, etc.). This last step of
the sampling strategy is done by the
interviewers.
7
A methodological challenge and its implications
(3)
? Construct a sampling frame by drawing up a
completelist of existing centers ? Collect
reliable data concerning the size of each
center,the service it provides, the way it is
run and its type ofpublic ? Adapt the
sampling of individuals to each case taking into
account the field realities ? Manage a rather
peculiar mode of contact between interviewers
interviewees through the centers.
8
From theoretical and methodological principles
To practical implications ? Evolutions and
field adaptations
9
1. Creating the sampling frame Field coverage
(1)
The sampling frame partly defines the field
targeted and covered, and its limits. Surveys
on homelessness Surveys among users of
services designed for the homeless population (in
a restricted sense)
10
Field coverage (2) an exhaustive list of services
Issues and difficulties for a full coverage of
the services - Which services to include ?
Conceptual definition / Practical
definition, feasibility - Frequent changes in
NGOs activities and environment - Specificity
of each service Applying explicit or implicit
policies and practices, the services attract and
receive different kinds of publics ??
Evolutions from 1995 to 2002
11
Surveys among users of services... Ined, Insee
(1995-2001)
Field
NGO s and public services shelters, hotel
rooms, apartments and
Doubled-up at relatives or friends
Street
Soup kitchen
Squats
Own dwellings
12
Complementary survey amongnon-users of
services... (Ined, 2002)
Field
Public space, street
Outreach services
Subway stations
Squats
Train stations
13
An exhaustive list of services
Lessons from the complementary survey among rough
sleepers met by outreach services (Ined, 2002)
? Feasibility of the survey in outreach
services, ? Possibility to include - new
services such as breakfasts provided by day
centers - and places which can be assimilated
to services
14
An exhaustive list of services ? Survey among
drug users (InVS-Ined 2004)Integration of both
service users and non-users
Day centers, Needle Exchange Programs, drug
treatment centers, fixed and mobile (buses,
outreach teams) Hospitals Specialized
shelters

General practitioners
15
Field coverage (3) the survey among the centers
By telephone and/or visits? Several aims
1. Contact heads of centers and present the
project, 2. Identify all services provided 3.
Eliminate, if necessary, non eligible services
4. Collect information about each service and
especially an average number of
service users, on an
average day 5. Know the centers rules and
organization
16
The survey among the centers
Problems related to a full coverage of the
centers... - Social workers vs researchers
problematic - Discussions about the
objectives of the survey, its timeliness and
legitimacy, its methods... - Will to protect a
population considered vulnerable -
Sensitivity of the data collected number of
users linked to budgets and financial
stakes?Essential role of meetings with social
workers in order to explain, convince, and update
the sampling frame
17
Field coverage (4) service users
Other problems related to a full coverage of
services users Non French speakers are not
included ?A complementary survey among non
French speakers has been conducted by Ined in
2002.
18
2. Sampling of individuals through the service
they use the role of interviewers (1)

• Every service user must have the same
  probability to be interviewed, without any prior
  selection made by the interviewer or social
  workersInterviewers have to ? Respect the
  sampled dates of interviews, ? Select
  individuals randomly, according to the sampling
  method, chosen by advance during a first visit in
  the service, in regard of The importance of
  active lines (too many or not enough people)
  The layout of premises (one / several doors)
  The organization of services (in line queuing)
  The possible anticipation of expected users

19
The role of interviewers (2)

• Services users surveys (Ined Insee, 1995
  2001) - method with list (appointments, bed
  booking...) sampling worksheet - method
  without list

20
The role of interviewers (3)
? Outreach services surveys (Ined 2002, InVS
2004) According to the size of the active
lines - selection of the first person on
the right of the group met - or no selection
every person met is interviewed (excepted any
situation which might be a constraint in regard
to the person or the outreach team)
21
Sampling and weighting the counting of service
users (1)
In order to Apply the sampling method
according to the number of present users
Estimate the probability for interviewees to be
included ? Counting of every service eligible
user for the survey during the intervention in
the service Excluding non users of the
service, minors and non French speaking users
22
Sampling and weighting the counting of service
users (2)

• How to make the difference between users non
  users, under over 18, French speakers non
  speakers? - Team work sampler interviewer
  (Insee SD 2001, Ined outreach services 2002 ) or
  interviewer alone social workers collaboration
  (InVS 2004) - Handing out of 2 different lists
  of services according to the age group (under /
  over 18, Ined 1998)

23
3. Very special conditions of survey
Populations and places
? Go beyond common sense preconceived
notions about homelessness ? Tackle a
difficult sensitive questioning about difficult
life histories, illegal practices or propose a
self collected finger prick blood sample (drug
users survey)...
24
Weather, time and places
Interviewing conditions often -
precarious within the services but also in the
street, buses, subway stations... -
uncertain weather conditions, police
controlsUnusual working hours - evening
and night ? Refusal from some professional
interviewers
25
Very special conditions of survey consequences

• Adapted selection of interviewers? Importance
  of random sampling ? Importance of interviewer
  training (including a presentation of services
  and public done by a social worker) ? Presence
  of research team on field, regular contacts with
  interviewers by phone or through regulation
  meetings ? Unusual in the French context
  compensation offered to interviewees (phone
  cards, radios, etc.)

26
4. Unusual survey requirementsthe essential
collaboration between research teams and services
(1)
Double constraint1. Adapt to each service in
accordance with - its rules organization,
public, collective life(Ex outreach services
observation, meetings with directors, following
of teams ) - the relationships between social
workers and service users (rough sleepers, drug
users)
27
The essential collaboration between research
teams and services (2)
Double constraint2. Respect survey methodology
and ethical rules - get in touch with the
person drawn (scattered accommodation, mobile
teams) - maintain a distance with social
workers (services wanting to interfere) -
protect interviewee anonymity and confidentiality
of responses (position towards the service)
28
Unusual survey requirementsImpacts on surveys
and adaptations
? ? Shortening questionnaire (Ined outreach
services) ? Giving up blood sample collection
(InVS survey) ? Not doing interviews in
services wanting to interfere ? Replacing
services within the same category
29
Thank you !