Selection strategy for the user
survey
User selection strategy – user survey
1
TECHNOLOGY & SOCIAL CHANGE GROUP
(TASCHA)
The Technology & Social Change Group (TASCHA)
at the University of Washington Information
School explores the design, use, and effects of
information and communication technologies in
communities facing social and economic
challenges. With experience in 50 countries,
TASCHA brings together a multidisciplinary
network of social scientists, engineers, and
development practitioners to conduct research,
advance knowledge, create public resources, and
improve policy and program design. Our purpose?
To spark innovation and opportunities for those
who need it most.
CONTACT
Technology & Social Change Group
University of Washington Information School
Box 354985
Seattle, WA 98195
Telephone: +1.206.616.9101
Email: tascha@uw.edu
Web: tascha.uw.edu
GLOBAL IMPACT STUDY
The Global Impact Study of Public Access to
Information & Communication Technologies is a
five-year project (2007-2012) to generate evidence
about the scale, character, and impacts of public
access to information and communication
technologies. Looking at libraries, telecenters, and
cybercafes, the study investigates impact in a
number of areas, including communication and
leisure, culture and language, education,
employment and income, governance, and health.
ABOUT THE AUTHORS
The Survey Working Group has contributed to
the design and implementation strategy of
the Global Impact Study surveys. The Survey
Working Group consists of:
George Sciadas, Statistics Canada (chair)
Mike Crandall, University of Washington
Beth Kolko, University of Washington
Carlos Manjarrez, Institute of Museum &
Library Services
Chris Rothschild, University of
Washington, Technology & Social
Change Group
Araba Sey, University of Washington,
Technology & Social Change Group
COPYRIGHT, LICENCING, DISCLAIMER
Copyright 2012, University of Washington.
This content is distributed under an
Attribution-Noncommercial-Share Alike
license. The views, opinions, and findings
expressed by the authors of this document do
not necessarily state or reflect those of
TASCHA, the University of Washington, or the
research sponsors.
Implemented by the University of
Washington’s Technology & Social Change
Group (TASCHA), the Global Impact Study is part
of Investigating the Social & Economic Impact of
Public Access to Information & Communication
Technologies — a broader CAD$7.9 million research
project supported by Canada’s International
Development Research Centre (IDRC) and a grant
to IDRC from the Bill & Melinda Gates Foundation.
Managed by IDRC, this project includes the Global
Impact Study of Public Access to Information &
Communication Technologies (this project) and The
Amy Mahan Research Fellowship Program, led by
Universitat Pompeu Fabra, which aims to deepen
the capacity of emerging scholars with the goal of
increasing the quality and quantity of research on
public access to ICT produced in developing
countries.
SPONSOR BLOCK
Sponsor info and logos.
User selection strategy – user survey
2
User Sampling Strategy Instructions
Number of users per venue and user stratification
Cybercafés:
Number of users per venue: 4
Sample stratification:
1. Gender: At each venue an even number of male and female users should be interviewed.
2. Time of day and day of the week: The total sample should be distributed across all the times of
day and days of the week that the venues provide public access to computing (based on the
contexts of the areas sampled). Note that this refers to the TOTAL sample and not the sample
drawn at each venue. For example, in an area where 5 venues are being sampled, a total of 20
respondents should be interviewed. If 40% of respondents (8 total) are to be sampled in the
evening, 4 respondents can be interviewed in the evening at two venues, equaling 8
respondents (40%). The 8 could also be distributed across all 5 venues. There is no requirement
that there should be an equal number of interviews for each time period in each venue. See
below for specific instructions on how to distribute the sample by time of day and day of week,
bearing in mind the above.
 Time of day: We expect that the interviews will be conducted during the three periods
of the day: morning, afternoon, evening/night. In areas where venues tend to be open
during only two time periods, we expect that the interviews will be conducted during
those periods. Aim to sample approximately 40% of users during the evening (because
we expect that venue traffic will be higher in the evenings).
 Day of week: We expect that the interviews will be conducted on all days of the week
the venues are generally open. Furthermore, we break the week into two distinct
periods: weekdays and weekends. Aim to sample approximately 40%of users on the
weekend (because we expect that venue traffic will be higher during the weekend).
3. The team must report to TASCHA on the method chosen to sample time of day and day of week
and why (i.e. what are the general operating hours and usage patterns of the venue in the
country)
4. Age: The sample will NOT be stratified by age. Selection should occur randomly, with age
variation accounted for in the randomness. However, users must be above 12 years old.
Libraries
Number of users per venue: 4
Sample stratification:
1. Gender: At each venue an even number of male and female users should be interviewed.
2. Time of day and day of the week: Time of day and day of the week: The total sample should be
distributed across all the times of day and days of the week that the venues provide public
access to computing (based on the contexts of the areas sampled). Note that this refers to the
TOTAL sample and not the sample drawn at each venue. For example, in an area where 5 venues
are being sampled, a total of 20 respondents should be interviewed. If 40% of respondents (8
total) are to be sampled in the evening, 4 respondents can be interviewed in the evening at two
venues, equaling 8 respondents (40%). The 8 could also be distributed across all 5 venues. There
is no requirement that there should be an equal number of interviews for each time period in
each venue.
User selection strategy – user survey
3
Since libraries often have limited hours and days of operation, as well as times when computers
may not be open for public use, the implementation team must develop a sampling strategy
based on their knowledge of library hours of operation in the country, with the following
requirements:
 Time of day: We expect that the interviews will be spread across all times of the
day the venues are generally open. The time of day should be split into 3 periods:
morning, afternoon, evening/night (if applicable). There should be an even
distribution of users sampled during each period of the day that the library is
open. If there is more traffic during a certain time of the day, a higher proportion
of the sample should be from that time.
 Day of week: It is expected that the interviews are conducted on all days of the
week the venues are generally open. Furthermore, we break the week into two
distinct periods: weekdays and weekends. There should be an even distribution of
users sampled during the week and on the weekend. If there is proportionally
more traffic on the weekends, approximately 40% of the sampling should be
done on the weekend.
 The team must report to TASCHA on the method chosen to sample time of day and
day of week and why (i.e. what are the general operating hours and usage patterns
of the venue in the country)
3. Age: The sample will NOT be stratified by age. Selection should occur randomly, with age
variation accounted for in the randomness. However, users must be above 12 years old.
Telecenters:
Number of users per venue: 4
Sample stratification:
1. Gender: At each venue an even number of male and female users should be interviewed.
2. Time of day and day of the week: Time of day and day of the week: Time of day and day of the
week: The total sample should be distributed across all the times of day and days of the week
that the venues provide public access to computing (based on the contexts of the areas
sampled). Note that this refers to the TOTAL sample and not the sample drawn at each venue.
For example, in an area where 5 venues are being sampled, a total of 20 respondents should be
interviewed. If 40% of respondents (8 total) are to be sampled in the evening, 4 respondents
can be interviewed in the evening at two venues, equaling 8 respondents (40%). The 8 could
also be distributed across all 5 venues. There is no requirement that there should be an equal
number of interviews for each time period in each venue.
Since telecenters often have limited hours and days of operation, as well as times when
computers may not be open for public use, the implementation team must develop a sampling
strategy based on their knowledge of telecenter hours of operation in the country, with the
following requirement:
 Time of day: We expect that the interviews will be spread across all times of the
day the venues are generally open. The time of day should be split into 3 periods:
morning, afternoon, evening/night (if applicable). There should be an even
distribution of users sampled during each period of the day that the telecenter is
open. If there is more traffic during a certain time of the day, a higher proportion
of the sample should be from that time.
 Day of week: It is expected that the interviews are conducted on all days of the
week the venues are generally open. Furthermore, we break the week into two
User selection strategy – user survey
4
distinct periods: weekdays and weekends. There should be an even distribution of
users sampled during the week and on the weekend. If there is proportionally
more traffic on the weekends, approximately 40% of the sampling should be
done on the weekend.
 The team must report to TASCHA on the method chosen to sample time of day and
day of week and why (i.e. what are the general operating hours and usage patterns
of the venue in the country)
3. Age: The sample will NOT be stratified by age. Selection should occur randomly, with age
variation accounted for in the randomness. However, users must be above 12 years old.
Other venue types:
Number of users per venue: 4
Sample stratification:
1. Gender: At each venue an even number of male and female users should be interviewed.
2. Time of day and day of the week: Time of day and day of the week: Time of day and day of the
week: The total sample should be distributed across all the times of day and days of the week
that the venues provide public access to computing (based on the contexts of the areas
sampled). Note that this refers to the TOTAL sample and not the sample drawn at each venue.
For example, in an area where 5 venues are being sampled, a total of 20 respondents should be
interviewed. If 40% of respondents (8 total) are to be sampled in the evening, 4 respondents
can be interviewed in the evening at two venues, equaling 8 respondents (40%). The 8 could
also be distributed across all 5 venues. There is no requirement that there should be an equal
number of interviews for each time period in each venue.
Since other venue types often have limited hours and days of operation, as well as times when
computers may not be open for public use, the implementation team must develop a sampling
strategy based on their knowledge of other venue type hours of operation in the country, with
the following requirement:
 Time of day: We expect that the interviews will be spread across all times of the
day the venues are generally open. The time of day should be split into 3 periods:
morning, afternoon, evening/night (if applicable). There should be an even
distribution of users sampled during each period of the day that the venue is
open. If there is more traffic during a certain time of the day, a higher proportion
of the sample should be from that time.
 Day of week: It is expected that the interviews are conducted on all days of the
week the venues are generally open. Furthermore, we break the week into two
distinct periods: weekdays and weekends. There should be an even distribution of
users sampled during the week and on the weekend. If there is proportionally
more traffic on the weekends, approximately 40% of the sampling should be
done on the weekend.
 The team must report to TASCHA on the method chosen to sample time of day and
day of week and why (i.e. what are the general operating hours and usage patterns
of the venue in the country)
3. Age: The sample will NOT be stratified by age. Selection should occur randomly, with age
variation accounted for in the randomness. However, users must be above 12 years old.
User selection strategy – user survey
5
How to select individual users once in the venue – Applicable to all venues





Selecting individual users to interview should be done using a consistent yet random method.
The actual methods by which users are selected should be decided at the country level and
implemented uniformly by all teams within a country. It is recommended that two methods are
chosen so that interviewers can have options to select from based on conditions at the venue.
To maintain consistency, however, no more than two methods should be used. Potential
strategies include (but are not limited to):
o Select every nth male, and every nth female. For example, the interviewer could go into
the room and select a “starting” point to count the users. The interviewer could choose
every 3rd male every the 3rd female from the starting point to interview. First the 3rd
male from the starting point would be selected and interviewed. After completing the
first interview, the interviewer returns to the starting point and counts to the 3rd female
for the next interview. Following that interview, the 6th male from the starting point
would be selected for the next interview, followed by the 6th female from the starting
point. For larger venues a larger interval can be used.
o As soon as the interviewer is ready to conduct a user survey, select whichever male
finishes using the computer first. Once that interview is finished, select whichever
female then finishes using the computer.
Teams should contact the Survey Working Group (via Chris Rothschild, chriskr@uw.edu) to get
feedback on potential selection methods.
Once in the field, adjustments can be made to the selection method if needed. For example, if
the method was to choose every 5th male and female but there are only 4 people total in the
venue, the method can be adjusted to select every other male and female.
Teams will need to provide a discussion of the methods employed and justification for their
selection in their survey report.
Other instructions - Applicable to all venues:
About the venue
1. Venues must have public access to computing. If when arriving at the venue the interviewer
realizes that the venue is not open to the public ever, a new venue MUST be selected. If the
interviewer learns that there are times of the day when the venue is open to the public, the
interviewer should come back during one of those times.
2. If the interviewer arrives at the venue and it is closed (for the day) or there are no users, he/she
should try to visit that venue at least one more time. This is where contacting the venue ahead
of time becomes important to increase field time efficiency. If, however, pre-contact is not
possible (e.g. there is no phone number for the venue) or the venue is closed despite calling
ahead, the team is free to use its own discretion on how many times to return. In some cases,
the team may not be planning on staying in the area overnight so returning the next day would
not be possible.
Selecting users
1. Note: it is the sole responsibility of the interviewer to select users to include in the sample. Do
NOT allow venue operators to decide. If the operator tries to suggest respondents to
interview, the interviewer may thank him/her for the suggestion and say something to the
effect of, “thank you but the design of the survey is such that we do not want to bias the survey
User selection strategy – user survey
6
by interviewing the “best” user.” The interviewer should use his/her discretion to determine the
most appropriate language.
2. Users must be or have been seen personally using a computer
3. Users must be above 12 years old
4. Do not select multiple respondents at a time to interview. Select one, do the interview, then
return and select the next.
5. It doesn’t matter if the venue operator is interviewed first, in-between interviewing users, or
after all users at the venue have been interviewed. Generally speaking, however, permission
should be sought from the operator to do the interviews before beginning.
6. Interviews can be held inside the venue or just outside, whichever is easiest given the situation
and cultural context.
Items to document during field work
1. The number of venues and users sampled at different times and days. (This will need to be
included in the bi-weekly reports to TASCHA. If it appears to TASCHA that the sampling is
becoming skewed to particular times of day or days of the week, adjustments will need to be
made to balance things out as appropriate)
2. Venues that were scheduled to be sampled but eventually not interviewed, with notation on
venues that
 are permanently closed
 are temporarily closed
 are not open to the public
 have no computers
3. Methods employed for user selection (will need to be included in bi-weekly reports)
4. Sampling strategy for time of day and day of the week (will need to be included in bi-weekly
reports)
User selection strategy – user survey
7