Is it possible to organize systematic randomization in ODK?
Within our humanitarian programme in DR Congo, we need to register several
thousands of beneficiaries with a short questionnaire, however we would
like to take the chance while registering them to also administer to a
random sample the baseline questionnaire for our monitoring.
Since we do not have the list in advance we thought of Systematic
randomization as a good option, so we would like ODK to automatically
switch from registration only form to registration and baseline every X
number of beneficiary being surveyed.
To implement your randomization scheme, you need to be able to count how
many instances of the Registration Questionnaire have been finalized and
refer to it in your form. I can't see how the original ODK is possible to
do this, I hope other folks in the community point out a way.
Nevertheless, our survey platform (rtSurvey) can do this quite easily. If
you want we can discuss and see how we can collaborate.
BTW, currently we are conducting an RBF IE survey in DRC, too.
Trung.
···
On Thursday, February 4, 2016 at 3:07:32 PM UTC+7, Arcc Drc wrote:
>
> Hi All,
>
> Is it possible to organize systematic randomization in ODK?
>
> Within our humanitarian programme in DR Congo, we need to register several
> thousands of beneficiaries with a short questionnaire, however we would
> like to take the chance while registering them to also administer to a
> random sample the baseline questionnaire for our monitoring.
>
> Since we do not have the list in advance we thought of Systematic
> randomization as a good option, so we would like ODK to automatically
> switch from registration only form to registration and baseline every X
> number of beneficiary being surveyed.
>
> Ex.
> Instance 1 Registration Questionnaire
> Instance 2 Registration Questionnaire
> Instance 3 Registration Questionnaire
> Instance 4 Registration Questionnaire
> Instance 5 Registration Questionnaire
> Instance 6 Registration Questionnaire
> Instance 7 Registration Questionnaire + Baseline
> Instance 8 Registration Questionnaire
> Instance 9 Registration Questionnaire
> Instance 10 Registration Questionnaire
> Instance 11 Registration Questionnaire
> Instance 12 Registration Questionnaire
> Instance 13 Registration Questionnaire
> Instance 14 Registration Questionnaire + Baseline
>
> Is anybody able to help?
>
> Many many many
> Thanks
> Gabriele
>
Gabriele,
If you were to do a random number generator from 1 to 7 go conditionally to
the baseline section when the random value is 7, then wouldn't that in
effect give you the same result in the long run. For every 700
registrations you should get around 100 Reg+baselines.
Curtis
···
2016-02-04 10:34 GMT+01:00 Trung :
Hi Gabriele,
To implement your randomization scheme, you need to be able to count how
many instances of the Registration Questionnaire have been finalized and
refer to it in your form. I can't see how the original ODK is possible to
do this, I hope other folks in the community point out a way.
Nevertheless, our survey platform (rtSurvey) can do this quite easily. If
you want we can discuss and see how we can collaborate.
BTW, currently we are conducting an RBF IE survey in DRC, too.
Trung.
On Thursday, February 4, 2016 at 3:07:32 PM UTC+7, Arcc Drc wrote:
Hi All,
Is it possible to organize systematic randomization in ODK?
Within our humanitarian programme in DR Congo, we need to register
several thousands of beneficiaries with a short questionnaire, however we
would like to take the chance while registering them to also administer to
a random sample the baseline questionnaire for our monitoring.
Since we do not have the list in advance we thought of Systematic
randomization as a good option, so we would like ODK to automatically
switch from registration only form to registration and baseline every X
number of beneficiary being surveyed.
Curtis BRODERICK
Head of technical training and user support
---------------------------EpiConcept---------------------------
Systèmes d'Information en Santé
Health Information Systems
47, rue de Charenton - 75012 PARIS
Tel : +33 (0)1 53 02 40 60 Fax : +33 (0)1 53 02 40 62
Thanks for your answer, how does your platform work? can you send me more
info at gerba@unicef.org?
Best
Gabriele
···
On Thu, Feb 4, 2016 at 11:34 AM, Trung wrote:
Hi Gabriele,
To implement your randomization scheme, you need to be able to count how
many instances of the Registration Questionnaire have been finalized and
refer to it in your form. I can't see how the original ODK is possible to
do this, I hope other folks in the community point out a way.
Nevertheless, our survey platform (rtSurvey) can do this quite easily. If
you want we can discuss and see how we can collaborate.
BTW, currently we are conducting an RBF IE survey in DRC, too.
Trung.
On Thursday, February 4, 2016 at 3:07:32 PM UTC+7, Arcc Drc wrote:
Hi All,
Is it possible to organize systematic randomization in ODK?
Within our humanitarian programme in DR Congo, we need to register
several thousands of beneficiaries with a short questionnaire, however we
would like to take the chance while registering them to also administer to
a random sample the baseline questionnaire for our monitoring.
Since we do not have the list in advance we thought of Systematic
randomization as a good option, so we would like ODK to automatically
switch from registration only form to registration and baseline every X
number of beneficiary being surveyed.
Yes indeed in the long run it would be like you said. However i do not see
to generate randomly the interval between instances to do either to short
or the complete survey, i would have a fix number depending on the
population size
Thanks
Gab
···
On Thu, Feb 4, 2016 at 5:37 PM, Curtis Broderick wrote:
Gabriele,
If you were to do a random number generator from 1 to 7 go conditionally
to the baseline section when the random value is 7, then wouldn't that in
effect give you the same result in the long run. For every 700
registrations you should get around 100 Reg+baselines.
Curtis
To implement your randomization scheme, you need to be able to count how
many instances of the Registration Questionnaire have been finalized and
refer to it in your form. I can't see how the original ODK is possible to
do this, I hope other folks in the community point out a way.
Nevertheless, our survey platform (rtSurvey) can do this quite easily. If
you want we can discuss and see how we can collaborate.
BTW, currently we are conducting an RBF IE survey in DRC, too.
Trung.
On Thursday, February 4, 2016 at 3:07:32 PM UTC+7, Arcc Drc wrote:
Hi All,
Is it possible to organize systematic randomization in ODK?
Within our humanitarian programme in DR Congo, we need to register
several thousands of beneficiaries with a short questionnaire, however we
would like to take the chance while registering them to also administer to
a random sample the baseline questionnaire for our monitoring.
Since we do not have the list in advance we thought of Systematic
randomization as a good option, so we would like ODK to automatically
switch from registration only form to registration and baseline every X
number of beneficiary being surveyed.
Curtis BRODERICK
Head of technical training and user support
---------------------------EpiConcept---------------------------
Systèmes d'Information en Santé
Health Information Systems
47, rue de Charenton - 75012 PARIS
Tel : +33 (0)1 53 02 40 60 Fax : +33 (0)1 53 02 40 62
Hi Trung, I found a short term solution by looping different set of
questions for the short questionnaire and the run the complete
questionnaire before closing the instance, but of course this as the
setback of grouping more observations into one instance exposing some
observations to the risk of not being saved. Do you have any better
solution? please do reach me on gerba@unicef.org
hi,
Hello,
(If I understood your question), I would apply the following solution: a dice game with a result of 0 to 9 (the random () * 10) function of ODK + "relevant" on the second questionnaire if the answer Is "6" or chosir 2 results ("4" or "7") or 3 results ("2", "3" or "9") ...
.int(random()*10) to have integers only.
If the random function of ODK is correct (I can not answer this statement), the draw will be random and the number of prints selected will allow you to continue the questionnaire every 1/9, 1/5, ..... person surveyed
The draw can even be doubled but there you have to do a bit of stats
This technique does not give results in relation to the total number of persons surveyed but only to the persons surveyed.
(I am not a specialist of the subject)
Dan
with a result of 0 to 9 (the random () * 10) function of ODK + "relevant" on the second questionnaire if the answer Is "6" or chosir 2 results ("4" or "7") or 3 results ("2", "3" or "9") ...