Transcript Hungary

The experience of using Google
map based sampling
(Slovaks in Hungary – the ENRIEAST project)
Endre Sik (TÁRKI)
for the seminar organised by Eurofound (Dublin)
October 2010
Interplay of European, National and
Regional Identities
FP7-SSH, 2008-2011
Partner teams from ten countries (Austria (leading
partner IHS), Belarus, Germany, Hungary, Lithuania,
Poland, Russia, Slovakia, UK, Ukraine)
Geographical scope of the fieldwork
LA
LT
RU
BY
PL
UA
SK
HU
The twelve fieldworks
The location and rate of the Slovak national minority in
Hungary (% of the local population (Census 2001))
The list of the selected settlements and the main charateristics of the
sample
Total population
(N)
Slovak
population (N)
Density
(%)
Proportion in
the sample (%,
N=400))
Number
of
routes
Preferred
sampling
method
2 170
1 185
55
16
7
RRS
551
268
49
4
2
RRS
2 314
1 059
46
15
6
RRS
Kardos
793
345
43
5
2
RRS
Sámsonháza / Šamšon
308
122
40
2
1
RRS
Örménykút
526
191
36
3
1
RRS
Felsőpetény / Horné Pet'any
766
274
36
4
2
RRS
Erdőkürt / Kirt'
611
185
30
2
1
RRS
Lucfalva / Lucina
615
163
26
2
1
RRS
Kétsoprony
1 559
413
26
6
2
RRS
Bükkszentkereszt
1 274
259
20
3
2
BRRS
Bánk
678
134
20
2
1
BRRS
Galgaguta
738
145
20
2
1
BRRS
Pilisszántó / Santov
2 120
416
20
6
2
BRRS
Tótkomlós / Slovenský Komlóš
6 547
1 159
18
16
7
BRRS
Tardos
1 629
255
15
2
2
BRRS
Galgagyörk
1 062
158
15
2
1
BRRS
Sárisáp
2 918
405
14
6
2
BRRS
Vanyarc / Veňarec
1 384
147
11
2
1
BRRS
Settlement
Pilisszentkereszt / Mlynky
Répáshuta / Répášska Huta
Piliscsév / Čív
Number of successful and failed interview attempts and
the reasons of failour by sampling methods
random sampling
snowballing
random routes
boosted
Number of approached respondents (N=1231):
59
572
600
Number of valid questionnaires (N=407):
44
214
149
Number or refusals or failed interviews:
15
358
451
Reasons for
refusals
(N and %)
n
%
N
%
n
%
Person is not suitable for the sample
3
20
159
44
279
62
It is impossible to enter the building /to the flat
0
0
10
It is impossible to enter the building /to come to the
flat
0
0
18
No one opens the door
0
0
42
12
2
0
The respondent is inaccessible
0
0
19
5
74
16
The respondent can not be interviewed because of his
health
0
0
3
Incomplete/interrupted interview
0
0
0
0
0
0
1
7
71
20
43
10
The house is uninhabited /broken; an institution
11
73
64
18
19
4
Total
15
100
358
100
451
100
The respondent refuses interviewing
Response rate,
%
75%
3
5
1
37%
4
1
0
0
2
0
25%
Sampling tree for Slovaks in Hungary
(Designed by the IHS team, prepared by TÁRKI)
153
Failed addresses
(no screening)
600
Address is non-dwelling
64
Dwelling not accessible after 3 attempts89
RRS: routes
Failed interviews:
447
Successful
contacts (screening
done)
Respnd. not eligible 162
Refusal
71
Interrupted interviews 0
233
Successful interviews: 214
Accepted for DB 214
Excl. from DB
99
1231
Total NN of
individual RS
routes & SB
chains
Failed addresses
(no screening)
572
Boost.RRS:
routes
Address is non-dwelling
19
Dwelling not accessible after 3 attempts80
Failed interviews:
473
Successful
contacts (screening
done)
0
Respnd. not eligible: 281
Refusal
43
Interrupted interviews 0
324
Successful interviews: 149
Accepted for DB 149
Excl. from DB 0
59
Snowballing:
chains
Failed interviews:
11
48
Failed SB contacts
(no screening)
Successful
contacts (screening
done)
4
Respnd. not eligible: 3
Refusal: 1
Interrupted interviews: 0
Accepted for DB
Successful interviews: 44
4
4Excl. from DB 0
Total NN of
successful interviews
in database
407
The Google maps of the routes of Tótkomlós
The main charateristics of the pre- and post Google based
TÁRKI Omnibus sample selection procedure*
Before Google map
based sampling
After Google map
based sampling
Source of the selection of
random routes
Phone book or map
Google map
Task of the selection of
random routes (in general)
Local instructors (17 all
around the country)
TÁRKI instructors
Basis of selection
Fieldwork experience, local
knowledge
Structure of the settlement
(based on Google map)
Selection of multiple random
route starting points (in big
cities)
TÁRKI instructor (two starting
points per cities one in
downtown, one in the outskirt
randomly selested – change
in every third month)
TÁRKI instructor ((two
starting points per cities one
in downtown, one in the
outskirt selected from
different Google map cells –
change in every third month)
Central control over selection
None
Complete
Problems and solutions
Interviewer route selection
bias (e.g. better pavement,
more light, etc.) – no solution
More uncertainty in the
course of selection (e.g.
uninhabited houses, etc.) –
TÁRKI instructor provides
new starting point
* On the basis of interviews with Éva Fáklya and Valéria Németh (TÁRKI instructors).