Has anyone tried to randomize the response options to a question? I am trying to collect social network data so I want to ask each respondent some questions about a few people in the village, chosen at random. The only option I can find is to write questions to ask if they know every single person in the village, and then randomize whether the question is asked, but with ~1,000 respondents this would be very tedious. Anyone done this before?
If you're using SurveyCTO and pre-loading your network roster from a .csv
file, you can draw random respondents from the .csv file. The general idea
would be to include a numbered column in the .csv file, draw a random
number with random-once(), and then use the random number to draw from the
.csv file.
Best,
Chris
···
On Sun, Jan 5, 2014 at 12:07 PM, wrote:
Has anyone tried to randomize the response options to a question? I am
trying to collect social network data so I want to ask each respondent some
questions about a few people in the village, chosen at random. The only
option I can find is to write questions to ask if they know every single
person in the village, and then randomize whether the question is asked,
but with ~1,000 respondents this would be very tedious. Anyone done this
before?
If you're using SurveyCTO and pre-loading your network roster from a .csv
file, you can draw random respondents from the .csv file. The general idea
would be to include a numbered column in the .csv file, draw a random number
with random-once(), and then use the random number to draw from the .csv
file.
Has anyone tried to randomize the response options to a question? I am
trying to collect social network data so I want to ask each respondent some
questions about a few people in the village, chosen at random. The only
option I can find is to write questions to ask if they know every single
person in the village, and then randomize whether the question is asked, but
with ~1,000 respondents this would be very tedious. Anyone done this before?
For what it's worth, we abandoned random() because your random draw gets
recalculated at various points, wreaking havoc if you've used it to
randomize question or group relevance. I've created an issue in the ODK
issues system, with our proposed random-once() implementation:
If you're using SurveyCTO and pre-loading your network roster from a .csv
file, you can draw random respondents from the .csv file. The general
idea
would be to include a numbered column in the .csv file, draw a random
number
with random-once(), and then use the random number to draw from the .csv
file.
Has anyone tried to randomize the response options to a question? I am
trying to collect social network data so I want to ask each respondent
some
questions about a few people in the village, chosen at random. The only
option I can find is to write questions to ask if they know every single
person in the village, and then randomize whether the question is
asked, but
with ~1,000 respondents this would be very tedious. Anyone done this
before?
For what it's worth, we abandoned random() because your random draw gets
recalculated at various points, wreaking havoc if you've used it to
randomize question or group relevance. I've created an issue in the ODK
issues system, with our proposed random-once() implementation:
If you're using SurveyCTO and pre-loading your network roster from a
.csv
file, you can draw random respondents from the .csv file. The general
idea
would be to include a numbered column in the .csv file, draw a random
number
with random-once(), and then use the random number to draw from the .csv
file.
Has anyone tried to randomize the response options to a question? I am
trying to collect social network data so I want to ask each respondent
some
questions about a few people in the village, chosen at random. The only
option I can find is to write questions to ask if they know every
single
person in the village, and then randomize whether the question is
asked, but
with ~1,000 respondents this would be very tedious. Anyone done this
before?