Health datasets - what do we want to measure?


#21

Thanks @Mor , I have a question relating to a submission I just made. So in the UAE, the infectious diseases data is available, but the hospital location data is not available as required (the opening hours are not available), so I did not put them as available to the public. I think that’s a bit too strict, but I think that’s how it should be done?

On a related note, I have been facing this problem with many of the submission I made: the system gives me an error when I indicate that the data exists, I don’t know if it is available in a digital form, and I know that it is not available to the public (So the order of the answers is Yes, Unsure, No). I tried this several times with many different submissions, the only way for the form to accept my answers is to choose “Yes, Unsure, Unsure”. Not sure if this is a problem others faced, but it affected many of my answers and I made a comment in the comments section below some of them.


#22

Hi @bluechi - Thanks for the update. I think that if hospital data don’t have one of the minimum criteria, we should still include it, we are experimenting in this category, so as long we are justifying it in the comment section, it is ok.

Your second part sounds like a bug in the survey logic. @pwalsh, can you have a look at this please?


#23

@Mor @bluechi

I see the issue with the Yes, Unsure, No bug.

The problem is that the changed methodology flow for this year (https://docs.google.com/drawings/d/1jj0mnDjvzz-O8SNBgEhNTqsXa8SNdMoMeoxPoy-tBfE/edit) deals with the conditions for Yes and No, but is unclear on the behaviour when answers are Unsure (it doesn’t address Unsure).

So, the implementation of this has some fixes that can be made to handle “unsure” on certain questions better. It might take some days to change it as I’m away right now, but I’ll look at it next week.

@Mor in general, I can fix this issue in code, but the survey flow might also need another look to understand the expected relationships between questions in some cases where a “parent” question is “unsure”.


#24

Unsure doesn’t have conditions. since it’s unsure. So I don’t think it
needs a logic like the “yes” or “no”.

(or did I miss what you are saying?


#25

The problem with the current definition of the health dataset is that it is
asking for two different data: location of health establishments AND
infectious diseases rate.

Very few country will meet the criteria.

Australia’s submission is a good illustration of the issue
http://global.census.okfn.org/submission/e608146e-b828-4958-bd38-7abd3d92e0b9

The solution would be to remove one criteria.

Regards
Pierre


#26

I had the same problem with the controls that hide some questions based on previous answers. My best guess is that they weren’t tested enough and there are some instances where the pre-defined answers do not fit the reality. My simple solution was to write a short message to the Regional Editor and explain where to correct the submissions.

The simple solution would be to remove the controls and let submitters answer all questions.

On the health dataset, it is a classical example of poorly designed item: two questions in one item. I can’t see why somebody would put list of hospitals and disease rates in the same dataset, since they have different reference systems. So it is likely that most people will have to use the comments box to indicate where the second dataset is located. There will be some complications where only one of the datasets is available, or even more complicated, when both are available but only one is updated. Since there are no clear guidelines, contributors may answer differently in similar cases.


#27

"data on infectious diseases rates in a country"
the index notes are not precise enough or narrow enough for crowdsourced contributors to make a dent or to know they have met the minimum requirements

notification of 32 infectious diseases is a legal requirement in the UK and national statistics are collated and published weekly:


#28

@Mor unsure must have conditions, otherwise it breaks the survey logic.

A simple example: if “exists” is “Unsure”, then without conditions, “digital” could be “Yes”, which would obviously be wrong.

Could you open up a GitHub issue for this?


#29

@ovoicu - great feedback, thank you!
We are aware of the situation of two datasets for one topic. It is some kind of an experimentation for us, so we are limited.

@pwalsh - sure thing. I will open an issue now.


#30

France : In this category, Ww have different datasets corresponding to the criteria on Health Performance with different license regime. We will answer UNSURE for the license but the situation is more: some are open, some are not.
What should we do? Thanks


Reflections on this year's index
#31

Unsure is a good solution. As long as everything is in the comments sections, we are good.


#32

I’d go with unsure, leave all the details in the comments, and reviewers can adjust as required.


#33

Hello @nehemiah_attigah,
Would you care to work with an organisation that is interested in working on such project? We could discuss about it more.


#34

Hi Emma Akin,

email me at nehemiah@odekro.org or nehemiah.attigah@hatuasolutions.com


#35

I am an entrepreneur in Training with MEST. I would like to work with you
on the datasets for Ghana and I think we have a couple of software
developers that would love to solve the pain point.


#36

reach out to me via the email provided and lets take the conversation further.