configurable labels for geographic areas

[ewheeler]

Currently, any geographic organizations of tracpro users are labelled as regions

This is confusing in scenarios where a country has an official administrative level called 'region' but tracpro is configured using a different official administrative level, such as 'districts'

[nicpottier]

I think we use state and district are our level 1 and level 2 names.

I feel this one is slightly tricky, as yes, obviously every country calls these different things, but then how do we refer to them in documentation, in dropdowns etc?

[ewheeler]

Yes - very tricky. Using state, district, and region (which often have very precise meanings) as generic terms is very confusing without additional explanation for users. I think using deliberately generic terms like admin1 and admin2 would lessen confusion overall.

The case for tracpro is a bit different, because geographic areas can really be arbitrary as it just represents a group of users (that may or may not have some geographic affinity -- could be organizational divisions or ministry departments etc). So for tracpro it could even be free text entered by the admin.

[ewheeler]

Another thought: for tracpro it might make sense to generalize even further -- and have them be cohorts instead of having any explicit geographic anchor. Then maybe have a cohort type that is geographic (so a 'district' cohort can be associated with country x's admin2 level) that could then be displayed on the currently non-existent map..

[ewheeler]

panel is the term used in the market research industry to refer, more or less, to "a group of people we contact regularly (and know their demographics) that we can send surveys to"

kindof obscure but perhaps more precise.

cohort still might make more sense for our users, but wanted to throw this in the ring too

[daaray]

👍 for cohort in this context.

[rowanseymour]

TracPro is currently modelling two types of cohort and I wonder if naming shouldn't distinguish them. There is:

• region - a cohort of contacts (typically) associated by geography. Supervisor users are assigned permissions to particular cohorts.
• reporting group - a cohort of contacts associated by role (e.g. Males, Teachers, Farmers)

cohort is certainly an apt word for the former... but that would seem to make the latter less clear

[ewheeler]

would it make sense for them to be
cohort (region)
role (reporting group)
?

[rowanseymour]

Or consolidate them both into a single Cohort model. They're pretty close to consolidated already as they both extend AbstractGroup (see https://github.com/rapidpro/tracpro/blob/master/tracpro/groups/models.py)

Then let users provide whatever labels they want for both of these cohort types. That saves us revisiting this when someone starts using role for something that isn't a role.

[ewheeler]

👍 sounds good to me
do you think it makes sense to plan for having optional types of cohorts?

then we could have, for example, a geographic cohort that maps to a district/province/etc
or a facility cohort where all contacts are related to a specific facility

these might be useful when expanding the reporting/charting/mapping capabilities

[rowanseymour]

I think I want to backtrack a little... Currently we only provide a limited number of ways to view the data, e.g.:

We've filtered by a region cohort at the top, and then we are disaggregating by "role cohorts" to get each row of the table. I can envisage a day when we have a super-duper report builder that allows users to generate any table or graph based on any subset of cohorts... but we'll likely always want these common views of the data.

We also only allow polls to be started against region cohorts (they are our panels) and only model user permissions based on region cohorts.

I like the idea of generalising cohorts but it also seems useful to keep this distinction between these two types of cohort - even if they won't always map to regions or roles. Perhaps then...

• Region → Panel since these are the cohorts that polls are conducted against that's the usual name for that sort of thing, e.g. https://msurvey.co
• Reporting Group → Cohort since these are any kind of cohort that be used for disaggregation

Curious to know what others envisage for extended analytics in TracPro as that seems relevant to this discussion.

[ewheeler]

@rowanseymour i dig it. makes sense why these should be distinct and treated differently

[dpoirier]

@ewheeler Do we still want to change these names? It seems like it'd be confusing for them to suddenly change, and this issue has been open almost 2 years now without anyone considering it important enough to actually make the change.

[ewheeler]

@dpoirier i would really like to. this thread helped to get the tracpro group concepts to click with devartis devs and some unicef folks as well. region has been very confusing for folks involved with edutrac in uganda as they are using ugandan districts which are labeled as regions rather than ugandan regions.

[dpoirier]

Looking back at the title of this issue, do we want to make these names configurable by org, and default to the current name so that nothing changes until/unless someone wants it to?

[ewheeler]

I'm not sure its critical to make these configurable since we arrived on the generic Panel and Cohort labels that side-step the need to configure by org because Panel doesn't prescribe a geographic area, let alone a specific term for a geographic area.

We can communicate the change ahead of time to unicef users so it isn't a surprise. I think Panel and Cohort are more sensible defaults as they won't risk confusion that Region can bring. Additionally, Cohort is a common term in the international development vernacular and will probably click more readily with more people than Reporting Group which is not a term of art for anybody as far as I'm aware

[dpoirier]

I'm starting to look into doing the rename. There appear to be at least 2 kinds of groups, contact groups and reporter groups?  If so, which one is being renamed to "cohort", or will we have contact cohorts and reporter cohorts?

[dpoirier]

@ewheeler Is there a difference between contact groups and reporter groups? If so, are we renaming one, or both? ("contact cohorts" and "reporter cohorts"?)

[ewheeler]

@dpoirier i might need to dig around to confirm. but i believe that only reporter groups should become cohorts

IIRC, reporter groups is the classic tracpro conception discussed as role/cohort in this thread

contact groups were added with the trackers & alerts functionality bc for those features is necessary to be able to track a contact's membership in all rapidpro groups in order to trigger group membership changes based on contact field values. i'm not sure that it is necessary or desirable to treat this the same way as cohorts/roles. its been a while since i thought through this and don't recall any specific reasons why we would want to keep separate or combine these

am i remembering correctly @xkmato @mreartes @fara ?

[dpoirier]

I need to move ahead on this. In the absence of objections from @xkmato @mreartes or @fara I'm going to go ahead with @ewheeler 's response that reporter groups should become cohorts, but contact groups should remain contact groups.