Working with Sites Like Survey Monkey

1
Working with Sites Like Survey Monkey
2
Note

• Saying Survey Monkey or similar survey-building
  web sites is going to get old.
• Ill just say Survey Monkey sometimes because
  they seem to be a popular choice.
• There are many such sites and if you want to use
  one you should ask around and check online for
  recommendations.
• Legal-dont-sue-me disclaimer No endorsement of
  Survey Monkey is implied. Im not ignoring or
  dissing other such sites.

3
The Dilemma

• There are always two ways to proceed on a
  computer project
• Make it yourself
• Buy it from someone else
• This rule applies to everything hardware and
  software large and small projects personal and
  corporate systems.
• My solution Know how to make it yourself so you
  always have a choice.

4
Making Surveys Yourself

• This class has taught you how to make surveys
  yourself.
• This gives you
• Maximum flexibility to customize your surveys to
  your needs
• But it costs you
• Time and effort

5
Using Survey-Making Sites

• There are sites that have pre-built templates
  that allow you to make surveys quickly and
  easily.
• This gets you
• Your survey made quickly and easily
• It costs you
• The inability to do things not supported by the
  templates

6
One Issue

• One of the most common problems people who use
  survey-making sites report is the inability to
  assign participants to random conditions.
• Many of these sites make it difficult (or
  impossible, or charge too much) to allow you to
  gather preliminary (e.g., demographic)
  information and then send participants to a
  random condition.

7
The Solution

• One solution is
• Write a bunch of surveys, one for each condition,
  and host them on a site like Survey Monkey
• Write you own front end that will gather
  demographic information and
• your last page provides them with a link to
  their randomly assigned condition on Survey Monkey

8
The Solution
Your site
Survey Monkey
9
The Solution
Your site
Survey Monkey
10
The Solution
Your site
Survey Monkey
11
How Do We Do That?

• The pages on your site are developed in exactly
  the same way as the other pages weve done in
  this course.
• Your last page will either randomly or
  block-randomly give subjects a link to their
  condition.
• This is what we will learn how to do today

12
Random Assignment to Conditions

• If you have X number of conditions, you can
  assign people to them by generating a random
  number between 1 and X.
• This almost never actually evenly distributes
  everyone equally across the X conditions.
• Often it is close enough.
• Sometimes, it isnt. The larger the number of
  conditions and the smaller your subject size, the
  more likely there is to be a problem.
• Block randomization fixes this.

13
Block Randomization

• If you have X number of conditions, you randomly
  assign X people to the conditions, making sure
  each condition gets exactly one person.
• The next X people get assigned to the X
  conditions, making sure each condition gets
  exactly one person.
• Repeat for each X people.
• This gives you almost a perfectly even
  distribution (there is always some error).

14
Programming Block Randomization

• Make a table in Access called Conditions.
• There is the PrimaryKey field and one field for
  each condition.
• There are two records in that table.
• The PrimaryKey1 record counts the number of
  times each condition has been assigned.
• The PrimaryKey2 record holds the URLs for each
  condition.

15
Conditions Table

• The http//www.cnn.com/ condition has been
  assigned 6 times.
• When you are finished testing your pages and want
  to clear out your database before going live, be
  sure NOT TO ERASE the contents of this table!!!!
  Just set the counts to 0.

16
Programming Block Randomization

• To assign someone to a condition
• Figure how the maximum number of times a
  condition has been check out. 6 7 6 implies 7
• Select the conditions that have been assigned
  less than that number of times. (6 in this case)
• Randomly assign the participant to one of those
  conditions.
• Mark that the just-assigned condition has been
  assigned 7 times.
• Have a snippet of code to deal with all
  conditions having been checked out the same
  number of times.

17
A Typical Pattern
18
Finishing Up

• After a condition has been assigned to a
  participant, store the participants Subject ID
  and assigned condition in the AssignedConditions
  table.
• Give them a link or a button to go to their
  condition.
• You may or may not be able to send their Subject
  ID to the survey site. Check with them to find
  out if and how.

19
GoToCondition.asp

• There is a folder called Other_Survey_Sites in
  common files. There is a database with two
  tables and two ASP files.
• TestCondition.asp just has a button that calls
  the next page.
• GoToCondition.asp block-randomly assigns them to
  a condition and gives them a button to click to
  go to the next site.
• URLs are to regular sites. Replace them with the
  real URLs to your surveys once you know them.

20
GoToCondition.asp

• Lines 18-21
• Set the number of conditions. The conditions
  table in the database should have the same number
  of condX fields.
• This example uses 5 conditions and there are
  fields cond1, cond2, , cond5 in the conditions
  table.
• The size of AvailableConditions is one more than
  the number of conditions. Recall that arrays
  start counting at 0 so if we declare their length
  to be one larger than we need, we can ignore the
  0th slot.

21
GoToCondition.asp

• Lines 24-32
• We connect to the conditions table.
• Lines 35-48
• Figure out how many times the most-assigned
  condition has been assigned. Store that in
  MaxNumber.
• MaxNumber starts at 0. It then checks how many
  times each condition has been assigned. If a
  condition has been assigned more than MaxNumber
  of times, then MaxNumber gets set to that number.
  

22
GoToCondition.asp

• Lines 53-67
• Make a list of all the conditions that have been
  assigned fewer than MaxNumber of times. These
  are the conditions from which we want to pick.
• NumberOfAvailableConditions will store the number
  of conditions that have been assigned fewer than
  MaxNumber of times.
• AvailableConditions is an array that holds one
  condX for each condition we might assign (e.g.,
  cond1, cond4, cond5).

23
GoToCondition.asp

• Lines 70-86
• Lines 53-67 will fail if all the conditions have
  been checked out the same number of times.
• If that is the case, NumberOfAvailableConditions
  will be 0.
• Lines 70-86 copy ALL of the conditions (because
  they are all candidates for assignment) into the
  AvailableConditions array.

24
GoToCondition.asp

• Lines 87-93
• Pick a random number between 1 and the number of
  conditions we are considering for assignment.
• ChosenConditionNumber will hold the X part of
  condX.
• ChosenCondition will hold condX.
• Example. If we chose condition 3, then
  ChosenConditionNumber 3 and ChosenCondition
  cond3.

25
GoToCondition.asp

• Lines 96-101
• Line 99 increments by 1 the counter for the
  condition we just chose.
• Line 100 saves that to the database.

26
GoToCondition.asp

• Lines 103-110
• These lines open a connection to the
  AssignedConditions table, which will store which
  subjects got assigned to which conditions.
• Line 110 saves the Subject ID.
• There are two ways to save the condition.

27
GoToCondition.asp

• Lines 112-114
• These lines save something like cond1 or
  cond4 to the database.
• If you want just the 1 or 4 to go to the
  database, then comment out these lines and
  uncomment lines 121 and 127.

28
GoToCondition.asp

• Lines 116-130
• These lines extract just the number out of
  condX and save that X to the database.
• To use them, comment out lines 112-114 and
  uncomment 121 and 127.

29
GoToCondition.asp

• Lines 133 - 158
• The links to the other surveys must be stored
  somewhere. The two options are inside the web
  page and in the database.
• Lines 133-158 store the links inside the web
  page.
• For example, Lines 152-154 check to see if they
  were assigned to cond1. If so, the URL for
  that survey is chosen.
• To use these lines, uncomment 152-158 and copy
  and paste the number of IF blocks you need.

30
GoToCondition.asp

• Lines 161-174
• These lines assume you have the links to the
  surveys stored in the conditions table in a row
  with PrimaryKey2.
• These lines go to the database and fetch the
  correct URL.

31
GoToCondition.asp

• Lines 177-190
• This is the web page visible to the participant.
  
• It gives them a form with a button that sends
  them to the survey on the other site.
• You could also make this a link
• lta hrefltSurveyURLgtGo to next pagelt/agt

32
Conclusion

• The code in this page block-randomly assigns
  participants to conditions.
• It then gives them a button (or link) to a survey
  hosted on another site like Survey Monkey.
• This allows you to randomly assign conditions
  using your page but make the actual surveys using
  tools available on sites like Survey Monkey.

33
Conclusion

• Note that you can also use this code to assign
  participants to conditions without going to
  another site.
• For example, if your pages were in a fixed order
  and you assigned them to a condition on the
  demographics page, you could replace all that
  randomization code weve had on the
  demographics.asp page and replace it with this
  code.
• The demographics.asp ltformgt tag will point
  towards actionlt SurveyURL gt