2. C THE
1. Case study “Security perception”
o
n 2. Research under difficult security conditions
t
e 3. Methodological challenges
n
t 4. Research under poor infrastructure
5. Learning: an enriching experience
3. 1. Case study
“Security perception”
Small towns in “red zones” 2010 2011
- Team of 40 pollsters WAVE I WAVE II WAVE III WAVE IV WAVE I WAVE II
Universe Man and women older than 18, SEL1 to 4, “red zone” municipalities
Personal 82% 82% 89% 90% 90% 90%
Type of poll
Telephone 18% 18% 11% 10% 10% 10%
Duration 30 min
17 Oct 2010 – 16 Nov 2010 - 14 Dec 2010 - 6 Jan 2011 - 23 Sep 2011 - 22 Oct 2011 -
Field work dates 10 Nov 2010 13 Dec 2010 29 Dec 2010 26 Jan 2011 31 Oct 2011 28 Nov 2011
Personal: random neighborhood, block and home selection - block scan
Sampling
Telephone: random number selection from a list (phone directory) and telephone series
Sample 1952 1952 1952 1952 3276 3294
Global ME - %
2,20% 2,20% 2,20% 2,20% 1,71% 1,70%
(Confidence level - 95%)
Municipalities 34 30 32 30 32 33
Weighting variables Strategic area and department
4. 2. Research under
difficult security conditions
5. 1 Follow sense of survival
¡!
Attention to:
Environment abnormalities Latent threats
Working schedule or poll area limit
Strictly obey to stop the field work when felt unsecure
Pollsters Polled
¡! ¡! ¡! ¡!
Threats
6. 2 Alternative completion of missing polls
Via phone Sample area substitution
Pollsters Polled
In another municipality with a similar size and
belonging to the same geographical area
7. 3 End customer confidentiality
The name of the customer
Colombian
Military was omitted
Forces
Personnel trained to not to
reveal its identity
YanHaas S.A
Study by YanHaas as a
private enterprise independent
from government
8. 4 Communication logistics
With military units:
Pollsters had access to a toll free number to the armed forces:
Inform about any abnormal
situation
Inform the commanding officer of
the closest military unit
Inform the field coordinators in
Bogotá
9. 4 Communication logistics
With police
When arriving to a municipality:
Introduce
themselves at the
police station
to let them know of their
presence, stay period
and objective
Pollsters Police
11. 1 Research planning
Methodology: multi stage sampling
Standard sampling Small municipalities
sampling
First stage : Keep personnel together
Selection of Town screening way - sample dispersion
cartographic block
pollsters
Second stage :
selection of homes
pollsters
pollsters Municipality
Third stage: pollsters
selection of home
member
12. 1 Research planning
Sample design
2 010 :
Planned 68 municipalities in 4 consecutive waves
Executed 68 municipalities in two groups
wave 1 and 3 wave 2 and 4
Sampling point reduction: difficult task since normally contractual terms and
conditions are mandatory and non modifiable when working with government
institutions.
13. 2 Questionnaire design
2 010 :
Cultural level of target group
70% having ONLY basic elementary education
Simplifying the language used
14. 2 Questionnaire design
2 010 :
Difference between military force and police
For population is
them.
hard to understand the difference between
Military Police
Polled
Remember that MFs are the
who
Army, Navy or Air Force,
wear camouflaged clothes.
Police (not camouflaged).
Military Police Polled Pollsters
15. 2 Questionnaire design
Interviewees identity protection Control
Questionnaire document
- Registry of names and telephone
numbers in a separate document YanHaas S.A Name
Telephone
- Polled kept confidential and safe in
case questionnaires were stolen
(Puerto Asís) or lost
Special attention to use of adequate wording
- Interviewees should not notice through the questions that
there is a preconceived positioning of the guerrillas
- Use of “distractors” in question formulation
- Removal of questions which forced the interviewee to directly
take a stand for or against the MF
16. 2
2
2
Questionnaire design
011:
Questionnaire
Validity
Test to prove in what level the questions oriented to a
common objective were coherent Redundancy:
reduction of the #
Reliability: Cronbach’s Alpha of questions, giving
Logic inherent to the given answers for a group of space for a deeper
variables exploration on subjects of
higher
importance to the
97,2%First wave customer
17. 3 Field work
Workforce retention
Normal conditions
Direct contract
3 out of10 interviewers deserted
or refused to work on the
project
Solution Increase the sensation of security for the team
introducing themselves at police stations and worked in groups
Involve local subcontractors
Rethink distribution of teams - each subcontractor worked together with an
experienced interviewer from YanHaas
18. 3 Field work
Training challenges
Project’s Project’s
Director Methodology and Special security Director
questionnaires considerations
• Fill out of two separate documents– questionnaire and personal information document
• Instructions were given about acting cautious in front of any abnormal situation
• Close work with police and permanent communication strategy
19. 4 Quality control
Backcheking
Normal conditions “Security perception”
Pollsters Polled
Direct supervision Backchecking
Fear
10
No.
%
Solution
Fill out quality control
100 Backchecking 32,3
First
%
% 100% sample recontact wave
Backchecking 100 38,1
Second
%
35 %
%
wave
20. Making the report
5 Multivariate analysis
Variables included 253
first measurement
240
second measurement
Wide coverage
Interdisciplinary team:
sociologists
municipalities visited per
wave 70
? Indicators 1. Optimism
2. Security
3. State presence
YanHaas 4. Illegal group presence
5. MF image
21. 4. Research under poor
infrastructure conditions
22. Telephone infrastructure
1 Methodology controlled rates
2 010 : Mixed methodology
Pollsters Polled Pollsters Polled
Contractibility problem Solution
¡! 99 extra dialing resulting from each contactable phone
0% and 1% effectiveness number, changing the last two digits
in some populations For second wave telephone polls quota was reduced
from 13% to 10%
Impact
$ Expenses increased, decreasing the planned profitability of the project
23. 2 Urbanism
Urbanism
Dibulla- La Guajira
Low population level – small
municipalities
In second wave, houses that had been
polled previously were omitted
Canton de San Pablo-Chocó
Lack of nomenclature
Forced pollsters to capture all possible
reference information to locate the
house later if necessary
24. Road infrastructure
3 Weather conditions
Road landslides
Restricted passing
Road closings
Underwater towns and roads
Substitution of sampling points
Efficient communication