As a Consumer, I want to understand more context about an institution's metrics to see if maybe there was an unaccounted for anomaly that would have influenced them.

[ErieMeyer]

As a Consumer, I want to understand more context about an institution's metrics to see if maybe there was an unaccounted for anomaly that would have influenced them.

On an individual school page, this would be called "College Response," and consist of a link to page on institution website where the institution could respond to the data presented on the individual school page.

So for example, Erie University's institution page would have a College Response link, which would be generated by (URL in IPEDS for the school)/StandardName, [Ex: Erie.edu/DataResponse] which would not display if the school page resolved in a 404.

Types of things that would go here could be "School was shut down for the entire year because of a natural disaster" etc.

Note: I don't think the /StandardName has been chosen yet. Separate issue tracking that.

[LisaGee]

school.edu/CollegeScorecard should be URL.

[LisaGee]

Marking this For Launch, even though this is technically required to be considered Feature Complete, but institutions won't have URLs set up to test against by Beta.

[LisaGee]

Dupe of #361
See #361 for design and placement.

[shawnbot]

Resolved: just put the link up and let it 404.

[shawnbot]

@jjoteal how's this look?

[ultrasaurus]

@jjoteal @LisaGee do we want this to be at the top of the page when most links will be 404s? when we talked about it with @LisaGee yesterday, I thought she said it would be less prominent, at bottom of page

[LisaGee]

Agreed. In the original designs it was at the bottom of the page. It probably got moved up in the spirit of trying to give institutions’s reponse a little more visibility.

Given what we’re implementing, I agree that it’s safer to put back at the bottom of the page, at least for this first release.

Thanks for catching.

--Lisa

From: Sarah Allen [mailto:notifications@github.com]
Sent: Thursday, September 03, 2015 7:47 PM
To: 18F/college-choice
Cc: Gelobter, Lisa
Subject: Re: [college-choice] As a Consumer, I want to understand more context about an institution's metrics to see if maybe there was an unaccounted for anomaly that would have influenced them. (#199)

@jjotealhttps://github.com/jjoteal @LisaGeehttps://github.com/LisaGee do we want this to be at the top of the page when most links will be 404s? when we talked about it with @LisaGeehttps://github.com/LisaGee yesterday, I thought she said it would be less prominent, at bottom of page

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/199#issuecomment-137603215.

[shawnbot]

OK, moving it to the bottom of the page!

[LisaGee]

After further conversation, we actually need a few things to change for Launch:

1. Remove the College Response button from the UI altogether. We will settle on the date this should actually appear in conversations with ED and then add it back to the UI after Launch.
2. It should be ready to go, in short order, after Launch.
3. To that end, the wording should change from: "College Response" to "More from the College".

[barkimedes]

@abisker @shawnbot understanding that this may be a larger ticket depending on how we can deal with the fact that we don't want to show the link if the school doesn't provide one, let's consider this a time boxed "think about it for a little while" kind of state. Shawn, whatever you can come up with for how we may want to deal with it, I would say toss those ideas in this issue's comments. Once we have some ideas/options, we can figure out how we want to move forward with this. Sound good?

[shawnbot]

Sounds good! What we need is a way to detect whether each school's response link is valid or not. Here are the best ways that I can think to do that, with advantages and disadvantages:

1. Write a script that generates a static index of which schools have valid URLs, and use that on the school page to determine whether to show the link.

   • ➕ no new infrastructure
   • ➕ easy to test
   • ➖ needs to be manually updated

2. Create a web service that fetches the URLs we care about dynamically, and tells us whether they're available.

   • ➕ doesn't need to be manually updated
   • ➖ involves creating a new piece of infrastructure
   • ➖ creates another path for integration testing

   This wouldn't necessarily be critical infrastructure, as we could have the school page just show the link if the service itself isn't available. So we might not need it to have dev, staging and production environments, blue/green deployments, New Relic alerts, and all of that.

The more I think about it, though, the more attracted I am to the manual option. School webmasters could even file pull requests to add their school to the index.

[abisker]

Thanks @shawnbot! For both options (but particularly #1 since you're leaning that way), do you have an idea of a) how long it would take to build out initially and then b) how intensive any manual updates might be on an ongoing basis?

[shawnbot]

I would probably do the following:

1. Get all of the school IDs and URLs and put that data (probably a CSV) in git. 1 hour
2. Write a script that fetches the response link URL for each school and checks to see if it's a 404, then write out those results to an index by school ID (probably as JSON). 2-3 hours
3. Output that JSON data on the school page, and update the school page script to check the index before dropping in the link. 1 hour
4. Write some tests for some links that should show up and some that shouldn't. 1 hour

So that's about 5-6 hours, all told. Updates should involve just re-running the script that updates the index and updating the tests if necessary, so hopefully no more than an hour (plus the requisite time to merge to staging and production).

[abisker]

So realistically, allowing for any surprises/bumps, this would take up a week of @shawnbot's time, i.e., if we do a week long sprint, this will be THE issue he works on -- with that in mind, @barkimedes do you want to prioritize this for next sprint, or wait on it?