Measuring Behavioural Outcomes with Polling Booth Surveys

Measuring Behavioural Outcomes
with Polling Booth Surveys
B.M. Ramesh, PhD
Department of Community Health Sciences,
University of Manitoba, Winnipeg, Canada
&
Karnataka Health Promotion Trust, Bangalore, India
1
Background
• Focused HIV prevention program in 6 high
prevalence states in India
– KSAPS implementing the program in 9 districts of
Karnataka
– KHPT implementing the program in 16 districts in
Karnataka and 3 districts in Maharashtra.
• Main objective is to reduce the incidence of HIV
and STIs.
• Monitor changes in condom use and other
sexual behaviours
• Polling Booth Surveys (PBS) among FSWs
2
What is polling booth survey?
• Polling booth survey is:
– A group interview method
– Where the individuals give their responses
through a ballot box
– Where the individual responses are
anonymous and unlinked
3
Polling booth survey: Advantages
• More suitable to collect information on sensitive
and personal issues
• Reduces social desirability bias
• Respondent remains anonymous
• Easy to administer among the illiterate
• Increases sense of confidentiality among
participants
• Simple to analyze
• Not individualized face to face interview, but a
group approach using a moderator giving
4
instructions
Comparison of face-to-face interviews
with PBS
• Among FSWs
– The FTFI tends to overestimate the condom use
behaviour
– The PBS tends to minimize the exaggeration of
reported condom use
• Among the general population
– The FTFI tends to underestimate the risky sexual
behaviours
– The PBS tends to give a relatively higher reporting of
risky sexual behaviours
5
Comparison of FTFI and PBS for eliciting sensitive
behaviours among unmarried subjects in Mysore GPS
Unmarried women
Unmarried men
FTFI
(n=534)
PBS
(n=545)
FTFI
PBS
(n=882) (n=484)
Ever taken HIV test*
0.6%
2.5%
2.5%
4.5%
Vaginal (urethral) discharge last year
3.4%
20.9%
0.5%
7.9%
Genital ulcer last year
0.6%
7.7%
0.3%
7.6%
30.3%
47.9%
67.5%
76.0%
Ever had heterosexual sex
0.6%
2.4%
8.3%
20.9%
Ever been paid / paid for sex
0.0%
0.6%
1.8%
6.2%
Ever had anal sex with a man
0.0%
0.6%
0.8%
6.6%
Year – 2005/06
Ever seen a condom
*Among those who have heard of HIV/AIDS
6
Comparison of FTFI and PBS for eliciting sensitive
behaviours among married subjects in Mysore GPS
Married women
Year – 2005/06
Married men
FTFI
(n=2036)
PBS
(n=575)
Ever taken HIV test*
6.1%
14.6%
3.1%
6.0%
Vaginal (urethral) discharge last year
3.5%
27.5%
0.1%
10.5%
Genital ulcer last year
0.8%
17.6%
0.1%
6.1%
Pre-/extra-marital sex ever
0.8%
10.3%
18.7%
35.9%
Ever been paid / paid for sex
0.3%
3.1%
2.3%
7.6%
Husband (wife) ever had extra-marital sex
10.3%
29.9%
0.5%
2.7%
Husband ever had sex with FSW
1.9%
12.4%
NA
Ever had heterosexual anal sex
0.5%
4.4%
3.0%
5.6%
Ever had homosexual anal sex
NA
0.5%
5.4%
*Among those who have heard of HIV/AIDS
FTFI
PBS
(n=1201) (n=409)
7
Comparison of FTFI and PBS for eliciting sensitive
behaviours among unmarried subjects in Belgaum
Unmarried women
Unmarried men
FTFI
(n=464)
PBS
(n=631)
FTFI
PBS
(n=947) (n=750)
Ever taken HIV test*
3.4%
3.0%
2.6%
8.3%
Vaginal (urethral) discharge last year
0.7%
17.1%
1.9%
16.5%
Genital ulcer last year
0.0%
5.4%
1.2%
13.2%
Ever seen a condom
17.0%
19.0%
75.7%
79.7%
Ever had heterosexual sex
1.1%
2.7%
5.6%
19.9%
Ever been paid / paid for sex
0.0%
0.2%
0.8%
8.5%
Ever had anal sex with a man
0.0%
0.5%
0.6%
4.9%
Year - 2007
*Among those who have heard of HIV/AIDS
8
Comparison of FTFI and PBS for eliciting sensitive
behaviours among married subjects in the Belgaum
Married women
Year - 2007
Married men
FTFI
(n=2055)
PBS
(n=715)
FTFI
(n=1456)
PBS
(n=707)
Ever taken HIV test*
19.5%
20.9%
8.5%
15.9%
Vaginal (urethral) discharge last year
2.3%
19.9%
0.6%
17.7%
Genital ulcer last year
0.2%
14.5%
0.4%
11.0%
Pre-/extra-marital sex ever
0.2%
9.1%
11.7%
35.1%
Ever been paid / paid for sex
0.2%
3.6%
2.1%
12.9%
Husband (wife) ever had extra-marital sex
5.9%
15.5%
0.1%
3.7%
Husband ever had sex with FSW
0.9%
7.3%
NA
Ever had heterosexual anal sex
1.4%
7.8%
1.0%
5.4%
Ever had homosexual anal sex
NA
0.2%
5.9%
*:Among those who have heard of HIV/AIDS
9
PBS methods (1)
•
•
•
•
•
Members are separated from each
other in a polling booth
environment
Each participant is given 3 boxes
(Red, Green and White) and a
pack of cards
Cards are numbered
corresponding to the number of
questions asked
Cards are stacked in serial order
Responses
–
–
–
–
Card with the number corresponding to
the question asked is put into the Green
box if the response to the question is
YES
Card is put into the Red box if the
response is NO
Card is put into the White box if the
question is NOT APPLICABLE
Card is kept outside these boxes if the
participant does not want to respond to
that question
10
PBS Method (2)
• The moderator explains the
PBS method with an example
and a practice session
• Questions are read one by one
• The moderator needs to
– Make the exercise lively
– Ask each question clearly,
slowly and loudly so that
every participant hears the
question clearly, repeating
the question as required
– Use situations/stories while
asking the question
– Use local terms
– Give sufficient pause, and
11
not to hurry
PBS Method (3)
• At the end of administering the
questions:
– Collect the cards separately:
the ones that are in the Green
boxes, the ones that are in
Red boxes and the ones that
are in the White boxes
– Count the number of cards in
each box for each question
and record the tallies in the
reporting form
• Share and discuss the group
responses with the participants
• Document group discussion
points
12
Sampling in PBS
• The subjects for the PBS may be recruited using
any of the probability sampling methods
13
Polling booth questions
• Only the questions that have a definitive YES or
NO as answers are suitable to be canvassed in
PBS
– How often do you use condoms with your regular
clients? (Not suitable for PBS)
15
Challenges/Limitations
• Non-response could be higher as not all selected
individuals are able to participate in a group at the
designated time and place
• Can include only the questions that have a YES or NO
answers
• Can ask only a limited number of questions –
questionnaire cannot be as long as the one that could be
used in a face-to-face interview
• Can get only the group-aggregated data, and cannot be
analyzed with linkages to individual characteristics
16
For more information
Dr. B.M. Ramesh,
Director, Monitoring and Evaluation
Karnataka Health Promotion Trust
IT Park, 5th Floor, #1-4
Rajajinagar Industrial Area
Rajajinagar, Bangalore – 560 044, INDIA
Ph: +91-80-40400209
Email: [email protected]
17
`