2. 7 May 2002 User Surveys - Jarrett, Bachmann2
Outline
Introduction
When to do a user survey
10 steps to a survey
Feedback
3. 7 May 2002 User Surveys - Jarrett, Bachmann3
Benefits of user surveys
Define the characteristics of users
Get direct input from users
Can have a big sample size
4. 7 May 2002 User Surveys - Jarrett, Bachmann4
Challenges of user surveys
Deciding between interviews and
questionnaires
Getting a good design
Interpreting the responses
5. 7 May 2002 User Surveys - Jarrett, Bachmann5
Interview or questionnaire?
Interview
Talk to the user
Interviewer captures
the answers
Requires small
samples
Flexible
Questionnaire
Do not talk to the user
Receive written
answers from user
Supports large samples
Inflexible
6. 7 May 2002 User Surveys - Jarrett, Bachmann6
An example
Look at ‘before’ example
from a survey of project
managers
What does this tell you
about the project
managers?
7. 7 May 2002 User Surveys - Jarrett, Bachmann7
Our activity example
New company SuperIndex™
Plan to launch SuperIndex™ at STC 2003
Opportunity here to find out:
about users
when they’ll use the product
what they want to see in it
9. 7 May 2002 User Surveys - Jarrett, Bachmann9
When to do a user survey
To specify users
To establish user goals
To characterise user groups
Interviews: Find out about users
Questionnaires: Confirm numbers,
not discovery
10. 7 May 2002 User Surveys - Jarrett, Bachmann10
Pragmatic choice of method
No direct
contact with
users
Direct contact
with users
Little time to collect
information
Short
questionnaire
Telephone or
electronic interview
Plenty of time Detailed
questionnaire
Face-to-face
interview
11. 7 May 2002 User Surveys - Jarrett, Bachmann11
Ideal choice of method
Face-to-face
interview
Telephone or
electronic interview
Detailed
questionnaire
Short
questionnaire
12. 7 May 2002 User Surveys - Jarrett, Bachmann12
Let’s try it
Fill in ‘use of surveys’ questionnaire
How many?
Top tip?
13. 7 May 2002 User Surveys - Jarrett, Bachmann13
When NOT to do a user survey
Don’t yet have a business definition of
the users
Forced to use a method that doesn’t
match the time available
15. 7 May 2002 User Surveys - Jarrett, Bachmann15
10 steps
Preparation
1. Set goals
2. Decide on target group
3. Interview target users
Production
4. Create questions
5. Decide on delivery
6. Test
7. Revise for final survey
Delivery
8. Conduct
9. Compile and analyse
responses
10. Publish results
17. 7 May 2002 User Surveys - Jarrett, Bachmann17
1. Set goals for the survey
What question are you trying to
answer?
How do you need
to use the information?
When do you need
the information?
18. 7 May 2002 User Surveys - Jarrett, Bachmann18
2. Decide on the target group
Who are they?
Do they all own the same information?
Census or sample?
Do they want to tell you the information?
How will you find them?
19. 7 May 2002 User Surveys - Jarrett, Bachmann19
3. Interview some target users
Check that you have the right group
Establish correct language
Find out what they want to tell you
Validate and refine your goals
20. 7 May 2002 User Surveys - Jarrett, Bachmann20
Activity: Steps 1, 2, 3
Set goals: market for SuperIndex™
Decide on target group: STC attendees
Activity:
Create an interview plan
Try using it on your
neighbor
21. 7 May 2002 User Surveys - Jarrett, Bachmann21
Typical first attempt
Would you use SuperIndex?
When would you use it?
What benefits would you see from it?
When would you NOT use it?
Check against step 3:
Interview some target users
23. 7 May 2002 User Surveys - Jarrett, Bachmann23
4. Create the questions
“similar to driving in freeway traffic while
drinking a cup of hot coffee and
answering an emergency call on a
cellphone”
“Many things are competing for
attention and failure to heed any of
them can spell disaster”
24. 7 May 2002 User Surveys - Jarrett, Bachmann24
4. Create the questions
Mix question types: choice and open
Use appropriate, unambiguous
language
Avoid leading questions
Present one question at a time
Keep questions concise
Keep positive
25. 7 May 2002 User Surveys - Jarrett, Bachmann25
4. Organize the questions
Keep to one topic at a time
Start with “easy” questions
Move from less invasive to more invasive
Minimise requests for personal information
26. 7 May 2002 User Surveys - Jarrett, Bachmann26
Activity: Create and organise
questions
Here is a list of questions
Rewrite them as good questions
Organise into topic order
27. 7 May 2002 User Surveys - Jarrett, Bachmann27
4. Write a preamble
Provide sufficient and clear instructions
Tell participants what the survey is for
Can they receive results if interested?
Are responses anonymous or confidential?
28. 7 May 2002 User Surveys - Jarrett, Bachmann28
Activity: Comment on preamble
You have been selected to take part in a survey
of Project Managers to identify which approaches
are being used to run projects. This
information will be used to re-develop
the current Life Cycle to meet the
needs of Group Technology.
Please answer the questions
below and return to ????? by
??/??/2002. It should only
take a few minutes
to complete.
29. 7 May 2002 User Surveys - Jarrett, Bachmann29
5. Decide on delivery format
Good for Problem
Mail anonymity slow
Website anonymity sampling
Email interested
respondents
privacy
spam
Telephone direct interaction time-consuming
Face-to-face direct interaction time-consuming,
costly
30. 7 May 2002 User Surveys - Jarrett, Bachmann30
6. Test the survey
Is the time involved to complete the
survey appropriate?
Are questions clear and unambiguous?
Are you targeting the correct group?
Is the survey interesting for the
respondents?
31. 7 May 2002 User Surveys - Jarrett, Bachmann31
Activity: Create Survey
Write out your questions in topic order
Swap surveys with someone at
your table
When complete, compare
your questions and
responses with
your partner
33. 7 May 2002 User Surveys - Jarrett, Bachmann33
7. Prepare the final survey
Use the test finding to refine the survey
Try analysing the results
Check the results against the survey
goals
Eliminate off-topic questions
Confirm delivery method for target
group
34. 7 May 2002 User Surveys - Jarrett, Bachmann34
8. Conduct the survey
Schedule the necessary time for delivery of
the survey
Plan to send a reminder message to target
users
Deliver the survey
Thank participants
35. 7 May 2002 User Surveys - Jarrett, Bachmann35
9. Compile responses
Responses: Compile using:
Up to 100 Spreadsheet
100 to 1000 Database, e.g. Access
Over 1000 Consider specialist
software e.g. SPSS
36. 7 May 2002 User Surveys - Jarrett, Bachmann36
9. Analyse responses
Calculate numerical responses
beware of false statistics
Group qualitative responses
don’t get too hung up on minor differences
Be prepared for surprises
some of your assumptions will be
challenged
37. 7 May 2002 User Surveys - Jarrett, Bachmann37
Activity: Analyze responses
Review the responses in the handout
Provide conclusions based
on the responses
Create a report about
the responses
Present your report
to the group
38. 7 May 2002 User Surveys - Jarrett, Bachmann38
“I’ve finished the report for you”
39. 7 May 2002 User Surveys - Jarrett, Bachmann39
10. Publish results
Don’t surprise people with bad news
Make sure publication is timely
Keep reports short
It’s OK to have some gaps
in the results, “more work
needed”
40. 7 May 2002 User Surveys - Jarrett, Bachmann40
Example questionnaire ‘after’
Distributed to 300 very busy project
managers
Obtained 100 responses
Showed that the two existing
methodologies are used very differently
41. 7 May 2002 User Surveys - Jarrett, Bachmann41
Questions
Caroline: caroline.jarrett@effortmark.co.uk
Karen: karenlynn@mindspring.com
Handouts:
http://karenlynn.home.mindspring.com/
http://www.effortmark.co.uk
42. 7 May 2002 User Surveys - Jarrett, Bachmann42
References
Don A. Dillman (2000)
Mail and Internet Surveys: The Tailored Design Method
John Wiley & Sons Inc, New York ISBN: 0-471-32354-3
A. N. Oppenheim (1992)
Questionnaire Design, Interviewing and Attitude
Measurement
Pinter Publishers Ltd, London ISBN: 1-85567-044-5
JoAnn T. Hackos and Janice C. Redish (1998)
User and Task Analysis for Interface Design
John Wiley and Sons, Inc. New York. ISBN: 0-471-17831-4
Notas do Editor
KB: Should questionnaire use an outline to mirror the interview plan of interviews?
Make sure that the survey is the right tool for the job.
Understand the limitations of surveys. Users may misremember information or may misrepresent information to meet what they think are the goals of the survey or to make themselves look better (social desirability bias). Most will be biased in some way, especially in cases where all members of the target group are not equally motivated to respond. If you receive only negative feedback about a product, that may be because only unhappy customers were motivated to reply and slanted the results rather than an accurate reflection of where the product needs enhancement.
A user survey can be very effective when used to collect the right information at the right time and in the right format. The type of survey used—questionnaire vs. interview vs. test, lengthy vs. short, anonymous vs. confidential, self-administered vs. investigator-administered—depend on a number of factors.
A user survey can be very effective when used to collect the right information at the right time and in the right format. The type of survey used—questionnaire vs. interview vs. test, lengthy vs. short, anonymous vs. confidential, self-administered vs. investigator-administered—depend on a number of factors.
Before even deciding to conduct a survey, you must determine what information you need to capture. If the information is best captured using a user survey, then setting well-formed goals will keep the survey focused and increase the likelihood of collecting the right information.
Set well-defined and concrete goals. Instead of “Learn whether users understood the documentation” set a goal to “Learn whether human resources personnel were able to use the documentation to perform all basic application tasks.”
Rarely will you really need or want to poll all possible users, unless you have a very small target population. Understanding which users will provide the desired information will result in quality responses. It may also increase the response rate if the target group has a particular interest in the information being collected (for example, senior users being asked about on-screen legibility).
Instead of targeting all users of the product consider targeting specific groups, such as HR reps, women between the ages of 25 and 40, or users over the age of 55.
Once you know the target population is, increase your understanding of those users by interviewing sample members to help validate and refine your goals as well as give you a start on creating questions.
Using the information gathered, create a list of questions that capture the information. Organize the questions so that respondents are drawn into the survey, saving the more personal or difficult questions for later in the survey.
Note: Including 7 choices does not a Likert-type scale make.
Survey format
Structured: Every participant answers the same questions; good for remote delivery
Semi-structured: Every participant answers the same core questions, but the interviewer can expand areas as needed; good for telephone and direct interviews
Unstructured: The interviewer asks questions freely; best used for exploration not reaching concrete conclusions
Using the information gathered, create a list of questions that capture the information. Organize the questions so that respondents are drawn into the survey, saving the more personal or difficult questions for later in the survey.
Note: Including 7 choices does not a Likert-type scale make.
Survey format
Structured: Every participant answers the same questions; good for remote delivery
Semi-structured: Every participant answers the same core questions, but the interviewer can expand areas as needed; good for telephone and direct interviews
Unstructured: The interviewer asks questions freely; best used for exploration not reaching concrete conclusions
Example of last item: In the blood donation interview, all answers should be No except two questions near the end.
Anonymous is when the survey administer does not know the respondents’ identity. Confidential means that the survey administer knows who the respondents are, but will not convey that information—this requires more trust from participants and, therefore, a higher vested interest in the survey.
The survey should be delivered in a format that encourages accurate and complete information as well as high response rate. The higher the vested interested of the group, the more questions you can ask and the more likely you are to elicit direct interaction.
Verify that the survey is usable by the target users. Areas to check is whether the time involved to complete the survey is appropriate, whether the questions are clearly understood, and if the target group is the correct sample to survey.
Example for point 4: User surveys at the end of user guides and the response rate.
Using feedback from the test step, refine the survey and prepare the final version. Depending on the length and importance of the survey, you may want to iterate through steps 4-6 several times.
Schedule the necessary time for delivery of the survey, whether a mailed questionnaire or an on-site interview. For questionnaires, depending on the deadline for responses, plan to send a reminder message to target users.
Capture the information collected in the survey in a format that is unbiased and useful to stakeholders (management, development team members, and so forth).
Circulate the results and any pertinent analysis as soon as possible after the response deadline