Hi all,
I was wondering how people go about collecting personal information needed for payment in oTree. In our lab, we are moving from cash payment to bank transfers, and for this I need to submit names, e-mail addresses, and the amount for each participant to my accounting department. Of course I could just include this as part of an experiment but we also promise participants that their behavior in experiments will not be matched to their personal identity, so there is some conflict there. I remember that Ztree has a specific routine to collect payment information in a file separate from the experimental data, but is something similar possible with oTree?
 
One workaround that I could come up with is asking people for their personal info in a separate online survey, ask them to add an individual code that we provide from oTree, and then link their amounts to their personal data using this code. Of course there is still some trust involved (I *could* match their names to their behavior this way, but I promise that I won't), but it's still better than having names and behavior in the same table to start with. But perhaps others have better solutions?

IN-PERSON EXPERIMENTS

We generally pay participants of in-person experiments in cash or immediate payment via PayPal. 

@Rense wrote:

> In our lab, we are moving from cash payment to bank transfers,

I have — successfully — lobbied with our university’s administration that we not replace immediate payments by bank transfers for in-person experiments because in my view, paying in cash or PayPal generates the maximum amount of trust: Participants can check right away whether the amount that they hold in their hands or have received via PayPal is identical to the remuneration that they were promised. And if they suspect that it is not, they can turn to the experimenter for clarification or correction.

This is different when the remuneration is paid via bank transfer. In that case, participants are usually unable to contact the experimenter directly, and they e-mail me instead has to forward their message to the respective experimenter.


ONLINE EXPERIMENTS

@Rense wrote,

> for this I need to submit names, e-mail addresses, and the amount for each participant to my accounting department.

When conducting online experiments, we pay participants via bank transfer. For this purpose, our administration requires us to collect each participant’s first and last name, postal address, and, of course, the amount paid. When collaborating with researchers from other universities, we are sometimes additionally required to collect participants’ tax IDs (Steuer-IdNr.).

> Of course I could just include this as part of an experiment

I do not see any issue with this approach. We inform participants in the invitation messages to all online studies about all the information that they will need to provide.

> but we also promise participants that their behavior in experiments will not be matched to their personal identity, so there is some conflict there.

It is true, of course, that this lifts anonymity. We therefore make it crystal-clear to participants that for the respective study, data collection is not anonymous. So far, we have not had any issues whatsoever finding enough participants for online studies.

> I remember that Ztree has a specific routine to collect payment information in a file separate from the experimental data, but is something similar possible with oTree?

Yes, you can generate a pseudonymized dataset of the experimental data using oTree. oTree allows you to download all collected data separately per app via the “Data” section of the admin interface. Hence, if you program a dedicated app for collecting participants’ bank details, you can download those data independently from the experimental data.

> One workaround that I could come up with is asking people for their personal info in a separate online survey

Of course, one can use a separate (oTree, Qualtrics, ...) survey for collecting the payment data. I do not consider this to make too much sense, though, as long as it is the same person who downloads both datasets.

> but it's still better than having names and behavior in the same table to start with.

I don’t agree.

I do think, however, that a separation into two datasets makes sense if the payment data are accessed by a different person than the experimentally collected data. That is, for instance, the experimenter exclusively accesses the experimental data, and a research assistant accesses the participants’ bank details. Via a shared identifier, the researcher can let the research assistant know which participant is supposed to receive which amount of money. This way, anonymity is preserved.

Thanks a lot for this very elaborate and helpful reply! I think the separation of the data between apps is precisely what I was looking, as this would allow to keep behavioral data and personal separate (of course we *could* still match them, but subjects will have to trust that we won't).