Randomization is possible by pre-populating the survey with data during launch

Using Key Survey to Conduct Experiments (Randomization)

Key Survey has a feature that allows you to conduct experiments in an online survey. This feature enables you to present different versions of a question or different blocks of questions to particular subgroups of respondents. These subgroups can be selected randomly or on some other basis, such as sex, race or age. This section explains how to use this feature.

Your “sample” is the group of people you are inviting to participate in your survey. In order for Key Survey to send email invitations to the people in your sample, you need to upload a list of email addresses before you launch your survey. But in addition to email address you can also include other information about each member of your sample. That additional information can be their name, a random number or a string of text. The random number can be used to set skip patterns in your questionnaire. The text string can be used to present a different version of your question to each respondent in your sample. The names can be included if you want to personalize email invitations and reminder messages. However names are not required when uploading sample. This feature makes it possible to use Key Survey for experiments.

The whole process will be explained in 3 sections:

Create and save the sample file with required data
Write instructions in the survey questionnaire
Pre-populate data from the sample into the questionnaire.

1. CREATE AND SAVE THE SAMPLE FILE WITH REQUIRED DATA

Below is an example of a sample member file that you will upload to Key Survey in order to launch your email invitations along with other information. Remember if you are not doing any experiment or don’t need any extra information your sample file will be sufficient with only the first field..

Randomization/experiment is possible by pre-populating the survey with data during launch. This data resides in the same file as the sample, which is a .csv file. In this file random numbers and strings are assigned to the respondents.

Your .csv data file should look as follows (instead of commas it may contain semicolons or tabulations as delimiters).

Fig: 1 (Sample File)

nrahma@princeton.edu	Naila	Rahman	1	I	white and black
efreela@princeton.edu	Ed	Freeland	2	D	blue and green
grace@princeton.edu	Grace	Cooper	3	R	red and orange
graceco@verizon.net	Grace	Cooper	1	I	white and black
rahmann@yahoo.com	Naila	Rahman	2	D	blue and green
arnaila11@yahoo.com	Naila	Rahman	3	R	red and orange
efreela1@verizon.net	Ed	Freeland	1	I	white and black

nrahman@princeton.edu,Naila,Rahman,1,I,white and black

efreelan@princeton.edu,Ed,Freeland,2,D,blue and green

grace@princeton.edu,Grace,Cooper,3,R,red and orange

graceco@verizon.net,Grace ,Cooper,1,I,white and black

rahmann@yahoo.com,Naila ,Rahman,2,D,blue and green

arnaila11@yahoo.com,Naila,Rahman,3,R,red and orange

efreela1@verizon.net,Ed,Freeland,1,I,white and black

The process of pre-populating data with their email addresses is called "Autofill". This advanced feature allows you to fill in responses automatically. To use this option follow the steps below:

Create .csv (comma separated values) Data File in UTF-8 encoding containing information you want to import. To ensure that correct encoding is used,

please open the sample file (which you’ve created in .csv) in Notepad and
click Save As, provide the file name that ends in .csv,
in the middle field choose ‘All Files’ and ‘UTF-8’ in the bottom encoding field.

.csv file opened in Notepad

The 3 fields to be filled in.

Your sample file is now ready. (For details you can visit Export E-mail Addresses From Data Sources.)

2. WRITE INSTRUCTIONS IN THE SURVEY QUESTIONNAIRE

In your survey questionnaire you must have a question which will be used to pre-populate data from the sample, preferably the first or second question.

Your survey question which will be pre-populated may look like this (Note that the order of questions is the same as the order of Data File fields – that is your sample file). This question should be hidden from the respondent.

Data

Email
FirstName
LastName
RandomNumber
Candidate
Color

Information from your sample file will be imported and stored in this question. We will then use the “piping” process within the survey to do experiments (i.e. assign different versions of questions to different respondents, different text to different respondents etc.)

Below is an example of a survey questionnaire with 16 questions. Where :

· Question # 2 is the hidden question and all information from the data source (sample) will be stored here.

· Respondents will see Question # 3 but with different text for groups of respondents.

· Questions 5 through 7 will be sent to one group of respondents

· Questions 8 through 10 to another and

· Questions 11 through 13 to another group of respondents.

The rest of the questions are for all respondents.

1. This is a test survey to show how to do randomization, experiment, auto fill etc. This question is for all respondents.

2. Data (Hidden question)

Email ............................

FirstName ………………………….

LastName …………………………..

RandomNumber ..............................

Candidate ..............................

Color ...............................

3. Can you see candidate _____ wearing _____ cap? (different text for diff respondents)

Yes

4. Define your relationship with the candidate.

Manager

Peer

Self

Other

5. Do you consider [Airline's] on-board food prices reasonable? (Q 5-7 for respondents with number 1)

Yes

6. Which of the following do you purchase most often during your flights with [the Airline]?

Food ........................................

Beverage ......................................

7. Please rate the following in-flight food and beverages.

Please mark the corresponding circle - only one per line.

Excellent Good Fair Poor

Selection of food

Quality and freshness of food

Selection of beverages (soft)

Quality of beverages (soft)

Selection of beverages (alcoholic)

Quality of beverages (alcoholic)

8. How satisfied are you with the following [education establishment] programs: (Q8-10 for respondents with number 2)

Please mark the corresponding circle - only one per line.

Very Satisfied Satisfied Neutral Dissatisfied Very Dissatisfied

Classroom aides

Science fair

Library

Music

Art

Computer/Technology

9. How satisfied are you with your child's progress in the following areas:

Please mark the corresponding circle - only one per line.

Very Satisfied Satisfied Neutral Dissatisfied Very Dissatisfied

Writing

Reading

Math

Computer Skills

10. Please indicate your level of agreement with the following statements about the [Educational Establishment].

Please mark the corresponding circle - only one per line.

Strongly Agree Agree Neutral Disagree Strongly Disagree

[Education facility] maintains a pleasant, warm and friendly environment

[Education facility] has a clean and well-maintained building

I have adequate opportunities to become as involved in [Education facility] activities

[Education facility] provides a safe learning environment

I feel comfortable recommending [Education facility] to other parents

11. What services did you deal with? (Q 11 – 13 for respondents with number 3)

Please pick one of the answers below. (Drop down list)

12. How satisfied were you with your medical care at [Hospital]?

Please mark the corresponding circle - only one per line.

Poor Good Excellent

Expected level of medical care

Actual level of medical care

13. Physician Staff

Please mark the corresponding circle - only one per line.

Disagree Neutral Agree

My doctors were skilled and experienced

My doctor was kind and caring

My doctor kept me fully informed

Cardiology

Other

14. Please provide your contact information. All fields are optional. (for all respondents)

Please use the blank space to write your answers.

Name

Title

Company

Street

City

State/Province

Zip/Post code

Country

Phone

Fax

15. Please provide any additional comments you may have.

Please write your answer in the space below.

16. Please select your gender

Male

Female

Here we will only give guidelines for constructing questions required for the experiment purpose.

The other simple questions are easy to create and can be done using ‘help’ in Key Survey.

INSTRUCTIONS FOR QUESTIONS WHERE DIFFERENT RESPONDENTS WILL SEE DIFFERENT TEXT.

3. Can you see candidate _____ wearing _____ cap? (different text for diff respondents)

Yes

In Key Survey the question will be constructed in the following manner:

Can you see candidate [Q2.A5] wearing [Q2.A6] cap?

· Here Q2.A5 refers to Answer 5 of Question 2 (this answer 5 was imported from the 5^th column of the sample file)

· Q2.A6 refers to Answer 6 of Question 2 (this answer 6 was imported from the 6^th column of the sample file).

When the respondent is answering the question he/she will see the corresponding data. The Q2.A5 and Q2.A6 portions will be replaced by data imported from the data source. In this case:

Q2.A5 will be replaced by D and
Q2.A6 will be replaced by blue and green.

(Look back at the sample file and question # 2)

One group of respondents will see:

Can you see candidate D wearing blue and green cap?

Another group will see:

Can you see candidate R wearing red and orange cap?

INSTRUCTION FOR DIFFERENT BLOCKS OF QUESTIONS FOR DIFFERENT GROUPS OF RESPONDENTS.

After you’ve created your questionnaire you will write logic in the previous question where the skip will take place. In this case it has to be written in Question number 4.

4. Define your relationship with the candidate.

Manager

Peer

Self

Other

The next few blocks of questions will be randomly chosen. Not all respondents will see the next few questions. The skip logics start here (Q 4) and you will create the logic here.

Go to the ‘Questions’ page and select the ‘Logic’ button of Question 4.

Logic button of question 4

In the logic page select the ‘Advanced Logic’ option and it will bring to the following page

where you will write all your logic instructions. Click ‘Add New’ to create a new logic rule.

Add New button

Advanced logic button

This will take you to the following page.

Enter the following Logic instruction in text box 2

Select the question (from the dropdown box) to’goto’ if this condition is fulfilled.

Q2.A4 == 2 go to Q8. How satisfied are you with the following …..

o (it means in Question 2, if the 4^th answer is 2 then skip to Question 8)

o Respondents with the numbers assigned 2 jumps to Question 8.

· Click ‘Save’ to save logic rule

Click save and keep adding more instructions as required

Select question to jump to

Write instructions here

In the same manner add more logic instructions to the same question (Q4):

Q2.A4 == 3 go to Q11. What services did you deal with? …….

o (for respondents 3 to jump to Question 11).

Save the instruction
Click ‘Activate’ to turn on all logic rules.

This is how your Advanced Logic page will look after saving the instructions:

Click Activate

The above instructions will take

· all respondents to Question 4,

· respondents 1 to go to the next question, which is Q5 (no instruction needed for this).

· respondents 2 will jump to Q8,

· respondents 3 will jump to Q11.

More logic instructions:

Now we need more instructions so that respondents 1 don’t get the questions which respondents 2 and 3 are supposed to get and vise versa.

Respondents 1 should get questions 5, 6 and 7 and then jump to question 14. So in question 7 we write (in the advanced logic section)

o Q2.A4==1 go to Q14

Respondents 2 will get questions 8, 9 and 10. Similarly, in question 10 we write:

o Q2.A4==2 go to Q14

Since respondents 3 will get questions 11, 12 and 13 and then go to question 14, no instruction is required here (since that’s the next question in the survey).

Your Questions page will look like this after applying the logic rules.

The above instructions will take

· all respondents to Question 4,

· respondents 1 to go to the next question, which is Q5 (no instruction needed for this).

· respondents 2 will jump to Q8,

· respondents 3 will jump to Q11.

More logic instructions:

Now we need more instructions so that respondents 1 don’t get the questions which respondents 2 and 3 are supposed to get and vise versa.

Respondents 1 should get questions 5, 6 and 7 and then jump to question 14. So in question 7 we write (in the advanced logic section)

o Q2.A4==1 go to Q14

Respondents 2 will get questions 8, 9 and 10. Similarly, in question 10 we write:

o Q2.A4==2 go to Q14

Since respondents 3 will get questions 11, 12 and 13 and then go to question 14, no instruction is required here (since that’s the next question in the survey).

Your Questions page will look like this after applying the logic rules.

Instruction for resp 1 to jump to Q14

Instruction for resp 2 to jump to Q8 and 3 jump to Q11

3. PRE-POPULATE DATA FROM THE SAMPLE INTO THE

QUESTIONNAIRE.

Now you will import data from the sample file by uploading the file. The first step is to go
to the Launch page by clicking the ‘Launch’ button.

Launch button

Once you are in launch page click the ‘E-mail survey distribution’ button.

Email Survey distribution

In this page you will select the option: ‘Import e-mail list for bulk survey distribution’.

Then browse and select the sample file. After that you will click the ‘Format Data File’ button.

1. First select Import e-mail list for bulk survey distribution button

3. Last, select Format Data File button

2. Then select sample file here

It will take you to the 'Format Data File’ page.

In this 'Format Data File' page simply select questions responses to which you want to import data from the sample file and use the [>] button to move them from the 'All Survey Questions' text box into the 'Autofill These Questions' text box. In this case it is the question called Data.
Just below the right box it says Enter field number for email address. Indicate the number of Email (column) in the 'Enter field number for email address' box. In this case it is 1 (because you have the email addresses in column 1 of your data file and question #2 should correspond with that). Make sure that columns of this table strictly correspond to fields of the Data File.
Click [Save] to confirm that you want to import responses to the questions listed in the 'Autofill These Questions' text box. A table representing the format of your Data File is automatically created.

Click [Back] to return to the 'Email Survey To List' page. There you can import your Data File.

All Survey Questions

Enter field number for email address

Autofill These Questions

Back in this page you will select the ‘Import Address File’ button to import data from the sample file, along with the email addresses.

Select the appropriate choices for your survey and then click the Send Email button.

Send Email button

Import Address file button

The email invitations will be sent immediately. It is always a good practice to test the survey
first with yourself and friends, before sending it out to the actual respondents.