We don't want to expose results data to non-logged in users.

Returns the error "HITnotfound" (obviously), but should probably respond to that error by clearing the HIT from the display.

When I click "Start Experiment" at the end of the preview phase, the experiment isn't starting. I'm not sure what, if anything, is supposed to happen. Look like this:

What we provide via the data explorer is likely going to coincide with what is provided with the API - and the data explorer needs more work before it is production ready. @poldrack and @rwblair let' discuss this at the same time as #31.

We will need to do this when there is a domain. I'm not sure how to set this up for a new user that is doing the deployment, other than to provide instruction.

letsencrypt.com
https://gethttpsforfree.com

we have documentation on read the docs, but we should have a simple version on the site, with links to readthedocs for "advanced" users or more info.

when time comes to test from the real mturk, we will want to check that the worker id exists before serving battery.

Add experiment is currently too cumbersome. Either:
(1) allow you to add multiple experiments at once by checkboxes (later going back to edit expvariables)
or
(2) stay on the same "add_experiment page" after you add them.

It seems to me that the Docker container (nginx) will need to be customized to connect to / know about / set up /serve the SSL Certificate. Will need to figure this out soon.

before the experiment styling, otherwise the font gets teeny :)

Errors with button on expfactory.org

Delete_Button:

New HIT:

Batch HIT:

Right now I have row IDs for debugging purposes:

do not forget to remove them.

When I create a HIT, though only one entry shows up in expfactory.org, two HITs are created on Mturk.

If I press "Expire HIT", it expires one of those two HITs but not the other. If I press "Delete HIT" it only expires the HIT, doesn't delete.

When I look at the HIT, no ad is shown. I can't start the HIT. I've included the following in the "Ad" box:

Thank you for your interest in our experiment!

This is a large experiment with many different tasks for you to do. The HIT should last around 10 hours and will pay $60 plus performance bonuses linked with some of the tasks.

Most of these tasks require that you press keys and sometimes we'll ask you to respond quickly. It is very important that you follow all the instructions and don’t mindlessly press the keys to finish the HIT more quickly! Not only will this hurt our research, but all of the tasks are designed to take around the same amount of time no matter how quickly you respond. The tasks will be more interesting to you and useful for us if you follow the instructions and try your best on all of the tasks.

If you accept the HIT you will see a start button. After you press start you will be shown a waiver of consent to read over followed by a more detailed description of this HIT.

(and come back to where he/she left off)

I received a number of errors when creating hits. First, when submitting a "New HIT" I get this error:

Regardless of the error, a HIT seemed to be created. However, I got errors when I tried to expire or delete the HIT:

Expire:

Delete:

we will want a set of standard analysis functions to work with data exports (either from a local battery, json, or from the docker instance, meaning tabular) and so the functions to generate the tabular data should be the basis for the analysis, and the functions to generate these tabular formats from the json data provided via expfactory-python (and not as a function in expfactory-docker).

The modular presentation of experiments will allow for (in the future) easy integration of other Javascript frameworks. I am creating this issue to remind us to implement this as an enhancement in the future.

This button currently doesn't work. I assume it should allow me to edit the battery variables, including the consent/information/ad text. If it doesn't, add that functionality.

Should be able to pay out "in proportion to" a value. So now just "if greater than", but the numeric value multiplied by some amount.

since this triggers a 404 in some browsers, it should be added.

Right now, anyone could go to a participant link to take a battery. We should implement some kind of protection against this, such as a password specified by the battery owner to get access to take a specific battery. @rwblair

Once we have a user agreement in place, we should discuss specific structure of the API. My thoughts are that we can offer summarized versions of data that are open for all to see. It might be overkill to try and do some kind of API authentication to programatically serve private (not yet public) data. @poldrack and @rwblair we will want to meet to discuss this.

Workflow should be like:

• person finishes battery
• assignment is submit to aws
• task is sent to celery to calculate credit

In celery task:

• look up credit conditions for bonus and rejection
• calculate bonus / determine rejection
• update assignment object
• if rejection criteria --> send to flagged list for user
• else: --> send updated object to AWS

In the case that the assignment is not approved from us, it is automatically approved after the time specified by the experimenter. This means that rendering of the assignments for review table should first obtain them from AWS to determine if they have still not been approved.

@rwblair there is currently a bug when I use AWS RDS that migrations aren't working. If you might have some time next week to take a look together, this would be very helpful!

jspsych display-stage, fullscreen, and presumably other init variables specified in the config file are not actualized.

It seems that the .click function for #start_experiment_button does not generate a jspsych.init that incorporates config variables. Here is the output:

$("#start_experiment_button").click(function(){
$("#instructions_modal").hide();
$("#bootstrap_css").remove()
jsPsych.init({
timeline: stroop_experiment,
on_data_update: function(data){
console.log(data);
expfactory.recordTrialData(data);
expfactory.djstatus = "UPDATE";
$.ajax({ type: "POST",
contentType: "application/json",
url : "/local/15/",
data : JSON.stringify(expfactory),
dataType: "json",
success: function(data){
console.log(data);
console.log("data update called")
}
});

Compare to the correct output from the same experiment run on expfactory.github.io:
jsPsych.init({
timeline: stroop_experiment,
display_element: "getDisplayElement",
fullscreen: true,
on_trial_finish: function(data){
addID()
},
on_finish: function(data){
jsPsych.data.localSave('stroop_results.csv', 'csv');
expfactory_finished = true;
}
});

Right now I have added 56 tasks, and am missing one (I should be at 57). It is now difficult to figure out what I missed.

must be encrypted, options could be s3, dropbox... @chrisfilo do you have suggestions for what you do for NeuroVault? Is it encrypted?

Researchers should also be able to click a button, and manually enter a participant ID, and run the experiment on the computer (the idea being they are running on computers in their lab by way of an internet connection to expfactory.org)

The second possibly desired integration is with facebook. If we could get this done for the initial go, this would be huge.

migrate old postgres database to RDS, test migrations

Three tasks currently are failing to install:
tower of london
psychological_refractory_period_two_choices
psychological_refractory_period_gonogo_choice

we need an admin button / view that will update files (and database options) for a given experiment. This is a significant decision as it will change all instances of the experiment, so it should be done with caution.

Additionally, it probably makes sense to end the preview with an "Accept HIT" button rather than "Start Experiments"

Test for bugs in running battery, report issues here @bissettp

This weekend. Me and @IanEisenberg !

it should be possible to get (de-identified) results by way of an API, instead of needing to download stuffs.

Now that we have fix #4 and have certificates for https, the server must be run in production mode in order to allow https to work with Django. We have two options here:

• deploy in production outside of Docker (faster and easier, as we are currently developing outside of it, but this means we will still need to find the correct nginx container image to somehow connect to /deploy the same certificates).
• deploy with Docker (harder as will need to figure out the above, but then it means that the piliot is more production ready).

When we decide on the above we can 1) deploy in production 2) pilot within lab to collect results and feedback, 3) meet with Russ and Ian to decide on new model with "forms" and make changes, and 4) pilot with larger group. Bonus / reject is in place #6 but we will still need some interface to manage the hits (and pending credits, etc) along with some kind of results interface. Right now all that can be done is download of results to tsv, which isn't so bad as this is the minimum requirements for this to work.

Both in the "add experiment" drop down and in the batteries list of experiments

Right now the link for an experiment that is added to a battery goes to the experiment template page, and the user would want to be able to review / change bonus / credit criteria for each experiment.

A user may want to deploy a custom experiment that is not part of expfactory-experiments, so we should allow for upload of a zipped file, and then have a "private" experiment only accessible to those with permissions for it.