UX Chat

« first day (1106 days earlier) ← previous day next day → last day (3577 days later) »

Erics

12:55 AM

@Pdxd I'll see that bad UX and raise you with this: cobbler's children syndrome

1 hour later…

Erics

2:01 AM

I'm organising a UX movie night - what are some of your fave youtube/vimeo/ted/etc clips on UX?

6 hours later…

John Doe

8:23 AM

What do you think about this? ux.stackexchange.com/questions/55569/…

Dan Hulme

It's a bit odd, saying "Do you keep your position against ..." as if the question's addressed to one person in particular.

JonW

9:14 AM

Urgh. Bloody pie charts. Trying to find better alternatives to using pie charts and my findings so far are.. all the alternatives are just as bad.

Dan Hulme

what information are you trying to show?

JonW

It's for dashboard displays. How many people logged a call by email, phone, webchat etc. About 6 different types.

Pie charts are useless at that sort of thing.

It'll probably end up being a bar chart, but I'm trying to find better options really.

Dan Hulme

perhaps I've missed a detail, but that sounds like the ideal use of a pie chart

or a stacked bar chart, if you want the absolute numbers to be comparable too

JonW

Well as an example, here's why it's a problem:

Which one is bigger? Green, yellow or magenta?

And also; how do you represent 0% on a pie chart.

Dan Hulme

there's a green slice in there?

JonW

9:18 AM

Well, the one on the left at 9:00ish

Dan Hulme

the answer to both of those is that you include the numbers

JonW

Not sure what colour that is.

Dan Hulme

:-)

JonW

It's not very data-visualisation if you just write the numbers everywhere. If you need to include the numbers for it to make sense then you might as well use a table.

That's why pie charts are rubbish.

Dan Hulme

if things that are almost the same size looking almost the same size is a problem for you, I don't think you're looking for any kind of chart

JonW

9:20 AM

It's a dashboard so needs an 'at a glance' view

Tables aren't good for that though.

"the only thing worse than a pie chart is several of them" - E. Tufte

Dan Hulme

oh no, if someone said they are bad then clearly there is no possible situation when they can work

let's all just go home

JonW

Haha.

Well clearly they have their place.

Dan Hulme

yes, and so far I'm not seeing why this isn't it

but then I don't know what's going on in your dashboard

it just sounds like either a pie chart or a stacked bar is appropriate

and you can give the user more detail on mouse over, if that helps

JonW

Too much cross-referencing and staring at the thing trying to figure out which segment is which, and which is larger. And that they only work when each segment is a decent percentage size.

Dan Hulme

haven't you realised yet that your example pie chart is ordered in size order?

the slices get smaller as you go around clockwise

JonW

9:24 AM

Yeah, but how much bigger is one than the other? What percentage is UKIP compared to Labour in that chart?

Dan Hulme

there's no single chart that would give me that information

JonW

They're good for comedy purposes though.

@DanHulme That's my problem.

I was hoping there was. I thought 'pie charts are annoying, what should I be using? I'll have a look online and see what the other options are'.

Turns out; there aren't really any other options.

Dan Hulme

well, as I say, a stacked bar can be good if the absolute value of 100% is important or changes

JonW

Trouble with those is similar to pie charts though; comparing one value with another is difficult.

Dan Hulme

that's a problem with all charts

they're supposed to make similar numbers look similar

the best you can do is animate them so the user can choose what things to put side-by-side to compare

JonW

9:28 AM

I did toy with going with a waffle chart for a while. But discounted them.

For one thing, they were making me hungry.

Dan Hulme

I don't like them because the shape of a segment depends on the sizes of the other segments

JonW

Plus; accessibility. They're probably even worse than pie charts.

Dan Hulme

they don't even make identical numbers look the same

that's a pretty big failing for a chart

JonW

Yeah, they aren't really much use.

Thought of these too:

But have similar problems.

Hmm. Quite like this one:

That gets around my issue of bar charts not illustrating that all the results total up to 100%

Dan Hulme

@JonW I think I'd sooner use two stacked bars for that

JonW

9:41 AM

This one is really showing the same dataset over two periods, whereas I'm only interested in a single snapshot, so would just use one chart, not two.

Dan Hulme

@JonW if you're only using one, what do you mean by this?

do you just mean that adding the extra %age scale makes it easier?

JonW

I mean that giving the bars an outline to show that they're 66% full is better than just having the bar on its own.

Dan Hulme

I see

JonW

and it helps illustrate that all the bars in the chart belong to the same scale, so overall they all add up to 100%

That is what pie charts do well; show that all the contents are part of one whole.

Dan Hulme

yes

JonW

9:45 AM

Not that any of this really matters. I'll spec up a decent bar chart or something and then the devs will go "we don't really have time to do it like this, we'll just chuck in a google charts version instead".

Dan Hulme

well, yeah

is a custom chart worth the extra time it will take?

JonW

Depends how long it'll take. I'd rather spec it how it would work best and then discuss compromises than go in with the compromise to start with.

Google Charts might not be an option anyway.

4 hours later…

Pdxd

1:27 PM

@DanHulme Look up D3 Charts, it looks very flexible. My team had used Google Charts and it seems flexible in the beginning but as we work with it more, feels very limited.

Ben Brocka

1:57 PM

App churn risk window

Wonder why iPad is so much higher--more forgettable apps perhaps, or download more

Dan Hulme

what is the time being measured on the x axis?

Ben Brocka

@DanHulme time between first and second app usage

Dan Hulme

yeah, I've skimmed the article now

I'd like to see quartile bars on those charts

Ben Brocka

> In examining global iOS (iPad and iPhone) app usage, the median user returns within just under 6 hours of his first app usage.

Dan Hulme

without knowing the SD of each thing, you can't tell the effect size

Ben Brocka

2:10 PM

I'm not sure why median is relevant...

Dan Hulme

and as they're pushing an analytics product, you have to assume they want to make the effect appear bigger

Ben Brocka

SD is, unfortunately, VERY rarely shared outside of academic sites for articles like this

The industry really deserves more and better data, but no one likes to share unless it gets people buying their stuff

Dan Hulme

If they can't see a connection between making their claims credible and people buying their stuff, I don't really care what they have to say about marketing.

Ben Brocka

I didn't even look at the conclusion, just found the stats interesting

Pdxd

@Erics I see your Cobblers children syndrome and raise you this: baddesigns.com

Nothing more ironic than a badly designed site about bad designs

JonW

2:26 PM

@BenBrocka as it's your question, and before this answer gets flagged - what do you want to do with it:

A: Has Fitts' Law been adapted to Touch screens?

This might be of interest to you: Bi, X., Li, Y., & Zhai, S. (2013, April). FFitts law: Modeling finger touch with Fitts' law. In Proceedings of the 2013 ACM annual conference on Human factors in computing systems (pp. 1363-1372). ACM. ISO 690 http://dl.acm.org/citation.cfm?id=2466180

Ben Brocka

@JonW I'd say wait and see if they update the answer--though I'm a bit doubtful what with brand new, unregistered user

I really wish we could leave a post notice that would ask for a follow up in a week or whatever

Lots of Not An Answers like this are more deserving of a followup in a few days rather than an on the spot delete/ignore

JonW

Sometimes I go through my old comments from the past week or so to see what happened to them.

I'd say it's more of a VLQ than NAA really.

Ben Brocka

well, per "links aren't answers". Could be fixed, which is exactly why I don't want to just delete it/convert to a comment

JonW

Annoyingly the link in that answer goes to a 'buy this paper' link with only a brief summary.

Dan Hulme

@BenBrocka I completely agree

Ben Brocka

2:32 PM

@JonW arg. Hate that about research

Pdxd

3:08 PM

There was a movement. I can't seem to find it right now but the movement entailed that if a research paper became popular (demand for it exceeded 1,000+ readers) it was forced to become open for the public.

7 hours later…