What do the stars mean?

[adieyal]

Is there any explanation about what the stars mean? At the moment they are pretty meaningless

[longhotsummer]

I'm not sure I agree with this. Stars are a widely used rating system, and while we may not be able to say something like "2 stars means drugs available within 2 hours", I think most users will understand more stars are better for each category.

What do you think we could do differently?

Greg

(iPhone)

On 17 Oct 2014, at 7:30 AM, Adi Eyal notifications@github.com wrote:

Is there any explanation about what the stars mean? At the moment they are pretty meaningless

—
Reply to this email directly or view it on GitHub.

[adieyal]

I don't know. Does the data explain what they mean? Should I worry if my
hospital has 3 stars for cleanliness?
On 17 Oct 2014 7:50 AM, "Greg Kempe" notifications@github.com wrote:

I'm not sure I agree with this. Stars are a widely used rating system, and
while we may not be able to say something like "2 stars means drugs
available within 2 hours", I think most users will understand more stars
are better for each category.

What do you think we could do differently?

Greg

(iPhone)

On 17 Oct 2014, at 7:30 AM, Adi Eyal notifications@github.com wrote:

Is there any explanation about what the stars mean? At the moment they
are pretty meaningless

—
Reply to this email directly or view it on GitHub.

—
Reply to this email directly or view it on GitHub
#34 (comment)
.

[longhotsummer]

This boils down to working out what a rating of 52 means for, say, "infection prevention and control", which I don't think we know. Any insight into the data @j-norwood-young?

Can we do better than "low is bad, high is good"? I'm not sure we need to. Part of the usefulness of a star rating system is that it boils complex details into something simple, albeit by losing detail.

[adieyal]

Don't get me wrong. I don't want to do more than the dept of health. I'm
just wondering whether they attach more meaning to it. I'll Look at the data
On 17 Oct 2014 8:59 AM, "Greg Kempe" notifications@github.com wrote:

This boils down to working out what a rating of 52 means for, say,
"infection prevention and control", which I don't think we know. Any
insight into the data @j-norwood-young
https://github.com/j-norwood-young?

Can we do better than "low is bad, high is good"? I'm not sure we need to.
Part of the usefulness of a star rating system is that it boils complex
details into something simple, albeit by losing detail.

—
Reply to this email directly or view it on GitHub
#34 (comment)
.

[j-norwood-young]

It's pretty complex - 30 pages of indicators.

[adieyal]

The original report is marginally more useful in that they say compliant vs
non-compliant. I have no idea what it is compliant with and whether
non-compliant is bad but at least there are some semantics.

How are the stars calculated? Equal intervals betwee 0-100?

One suggestion would be to use percentiles instead. That way if 20% of the
samples fall below a score of 45%, the stars are better allocated.
Regardless, that's only for correctness and doesn't help with usability.

How about saying that this hospital/clinic is better than 75% of all
hospitals/clinics

On 17 October 2014 09:38, Jason Norwood-Young notifications@github.com
wrote:

It's pretty complex - 30 pages of indicators.

—
Reply to this email directly or view it on GitHub
#34 (comment)
.

Adi Eyal
Data Specialist
phone: +27 78 014 2469
skype: adieyalcas
linkedin: http://za.linkedin.com/pub/dir/Adi/Eyal