Can ODK support discrete choice experiments?

I am looking to use ODK for a standard self-reported surveys, but would also like to use the same platform to complete a Discrete Choice Experiment. I have designed the survey and just need to be able to format the presentation for the presentation of a 4 x 4 matrix and radio buttons for respondent to nominate their choice across the 4 alternatives.

Backend formatting of data is another issue. The characteristics of the selected alternative, plus those of the unselected alternatives needs to be captured too.

Suggestions/advice would be helpful

Hi, ILeswsu,

would you please check this link

Best Regards

And if you need to see both the XML and XLS, should help too!

Thank you Nader and yanokwa for your suggestions, however, I don't see references to discrete choice experiments by following the links you provided. I see the multiple choice answer options, but DCE requires additional information be collected with each choice made. It seems unclear when ODK does this. Please correct me if I am wrong or missed something. The data collected and analysed is conditional, my choice of option A is conditional on the relevant attributes levels provided in a specific choice task. As each choice task changes, my conditional choice may change too.

Any other suggestions?

I read the following paper

that suggests that DCE can be run in ODK.

That article is a great resource! It looks like they don't provide a form but perhaps they'd be willing to share it if you got in touch?

I'm not very familiar with the DCE technique but I believe that a key component is randomizing the combination of values in the grid and knowing which choices were displayed to the survey-taker, is that right? If I'm following that correctly, I would recommend some of the techniques described here. In particular, I would create groups for each possible grid and then use a random number to choose which one gets presented. That way I would know exactly which grid the enumerator saw and which combination they chose.

Thanks, I have emailed the lead author and will share what information is given (with permission). There are several statistical designs to seek to maximize different statistics features - orthogonality, efficiency and also bayesian. I will through the link you provide. Cheers.

1 Like

it's look like one of thes

hie ileswsu, am having a similar concern, did you get a response on running DCE in ODK from the paper above?

Yes. One can insert a choice task as an image then use ODK's multiple answer survey question type to record the alternative selected. This is cumbersome as each choice task must be loaded independently into the ODK survey.

Hope this helps.

Richard Iles

thanks, but alitle more explaination with an example will do if you dont mind please