I'd like to get your ideas on the design for the site when it graduates from Beta. I really know nothing about statistics, besides the couple courses I took in college. So I'm definitely all ears! :)

When I design a site, I design for a "feeling." As in the feeling a brand new user visits the site, or a long term "feeling" seasoned users establish. For example, a person should know right away this is a "cooking site," because it's warm and inviting, or this is a "gaming site" because of the 8bit retro graphics. etc etc. I also try not to let the visual elements get in the way of content, since they should fade in the background after repeated visits.

So what's that "feeling" for the statistical analysis page? I recently designed the Mathematics site. It uses a familiar grid pattern as the Beta theme, sans the "sketchiness." The design goal was to make it clean and simple, but academic looking. Should the SA site be something similar?

I'd love to hear your ideas. Normally if I know a subject matter somewhat, with some extra research I can pin down the correct feel. However in the case of Statistical Analysis, I'm rather lost.

Edit: I have changed the post to only ask for over all design suggestions, for logo ideas, please use the other thread.

UPDATE: design mockup screenshots

Thank you for your input on design ideas, both here and in the chatroom. The consensus was for a clean and simple design. Also, it seems we're pretty set on a logo design from the logo thread.

I learned from some of you that infographics falls under statistics too, which gave me some design ideas. I've been a long fan of http://flowingdata.com and http://infosthetics.com. I decided to add some subtle infographics influenced elements to the design to spice it up a bit.

Color wise, to go with the clean look, I decided to go with more muted and pastel colors that are soothing on the eye. Of course, picking the right colors that would please everyone is rather difficult. I believe the palette I chose is rather safe.

Here are couple design mockups for both the home page and the question detailed page. (note: not all site elements are present, this is mostly for look and feel).

click images for full resolution versions.

alt text

The faint graphics in the header and footer background are from a program that tracks and plots my mouse movement. I like the randomness of it and it gives that infographics feel. I'm using it for pure decoration purpose to keep the design from being too bland, since the body section is all text.

alt text

Please let me know your thoughts. I'd love to get the site launched for you this week!

  • $\begingroup$ Why will the site be launched as Statistical Analysis? Looking through other threads, Data Analysis seems to be favored by almost everyone (including a data.se url). $\endgroup$ – Peter Smit Oct 28 '10 at 5:54
  • $\begingroup$ See for example: meta.stats.stackexchange.com/questions/379/… $\endgroup$ – Peter Smit Oct 28 '10 at 5:55
  • $\begingroup$ @Peter, I wasn't aware of that. I'll ask Robert about this. $\endgroup$ – Jin Oct 28 '10 at 6:00
  • $\begingroup$ @Jin, please have a proper look at the logo thread you linked above. Although some of the ideas there are specific to the 'CrossValidation' name, others aren't at all. In fact, some were proposed after it became clear that CrossValidation wasn't going to happen. @Moderators: I've no idea if maybe it's possible to migrate answers between questions to help distinguish the two?? $\endgroup$ – onestop Oct 28 '10 at 8:22
  • $\begingroup$ @Peter I have also mailed Robert to justify the name issue. $\endgroup$ – user88 Oct 28 '10 at 12:15
  • 2
    $\begingroup$ @Peter, @Jin Looks like Robert agreed we can have CrossValidated in a same way as SeasonedAdvice; so the previous thread applies. $\endgroup$ – user88 Oct 29 '10 at 7:18
  • $\begingroup$ (In case anyone wonders why both my above comment and one of my answers below seem to make little sense, that's because the question has since been radically changed.) $\endgroup$ – onestop Oct 30 '10 at 12:42
  • $\begingroup$ @Jin Looks great! $\endgroup$ – Shane Nov 3 '10 at 16:54
  • $\begingroup$ The logo has a glaring flaw. The trace of a matrix, which I presume is what the logo is depicting, runs from the top left to the bottom right, not the bottom left to the top right, which has no interpretation. Please correct that. Thank you. $\endgroup$ – gabgoh Nov 4 '10 at 18:02
  • $\begingroup$ @gabgoh The logo is representing cross-validation, where every time a different part of the data set is left out for evaluation, not the trace of a matrix. $\endgroup$ – Peter Smit Nov 4 '10 at 18:41

Love the look of the normal distribution for the accepted answer, but I am not sure about it when it comes to the voting itself (2nd screenshot you posted). I understood that those were buttons to up/downvote because I know how stackoverflow (and related sites) work. But if I were a new user who was not familiar with the way the site works I wouldn't immediately recognise them as arrows to vote.

Maybe red and green (for bottom and top respectively) border would help in that sense?

  • 3
    $\begingroup$ @Matt yes that was my concern as well. The concept was first proposed in the logo thread. I just implemented it in my mock to see what people think. I've talked to MBQ further in chat.stackexchange.com/rooms/18/ten-fold about the usability of this. We decided to go with the standard arrow shapes(but in a slightly curvy form). $\endgroup$ – Jin Nov 3 '10 at 20:36
  • $\begingroup$ Sounds good to me. Thanks, Jin! $\endgroup$ – Matt Parker Nov 3 '10 at 21:10

Re similarity or otherwise to Mathematics StackExchange site:

I think the 'feel' should also be clean and fairly simple, but less academic looking than Mathematics. In particular, we seem have established a preference for a Sans font rather than a Serif font. We make less use of TeX, so if the TeX font can't be Sans the mismatch is less of an issue here than it would be for Mathematics.

  • $\begingroup$ TeX font can be sans. $\endgroup$ – user88 Oct 28 '10 at 12:00
  • $\begingroup$ Oh good, very glad to hear it! That's something Wikipedia has yet to manage. $\endgroup$ – onestop Oct 28 '10 at 17:14
  • $\begingroup$ thanks for your input. I'll definitely come up with a clean and minimalist design for the stats site. $\endgroup$ – Jin Oct 29 '10 at 16:56

Overall, I like the mock-up a lot - it's very easy on the eyes, the mouse-click background is a cool-yet-subtle effect, and you've made mbq's already-awesome logo look fantastic.

I did have a problem figuring out the up-and-down buttons in the question screen, however - first, I thought they were missing; then I realized they were these little green slivers. After I thought about it for a moment, I realized that they must the tails of a normal distribution. Only now, after mulling that over for hours and having my eyes dilated by my ophthalmalogist have I actually seen the rest of the distribution behind the vote count. It looks like you're using the same grey there that you put as the backing for the tags, etc. - it might be easier to see if it were darker, or perhaps a different color than grey (would it work to take the same green of the toggled up-vote and just use a lower alpha setting?).

Otherwise, looks fantastic.

  • $\begingroup$ Oops - I see nico was posting on the same thing while I was still writing. $\endgroup$ – Matt Parker Nov 3 '10 at 20:25
  • $\begingroup$ see my comment to Nico regarding the voting widget. $\endgroup$ – Jin Nov 3 '10 at 20:46

Scatterplot with fitted straight line Ok, well, here's a very sketchy logo idea, based on my comment to Shane's answer on the other thread:

"I've started thinking a scatterplot with a fitted line is a good idea. As Shane said, it communicates the purpose of the site in a generally recognisable way. Clearly the [example posted by Shane] is too complex though. One with around 20 points would probably be good. Should be real data i think, preferably historic or iconic (in both senses!)."

I'm not really suggesting this particular plot - it's a scatter of CO2 emissions per capita vs. GDP per capita for the 19 individual countries of the G20,. I'm sure someone could find a less potentially contentious data set, perhaps from a classic textbook?

  • $\begingroup$ @onestop Thank you for the info. I'll read all the logo submissions from the other thread. I do like the chart logo above. One thing to consider is that we'll use the logo for site Favicon and Apple touch icon as well. So something overly detailed may not show up well. Maybe an abstract/symbolic version of a scatterplot? $\endgroup$ – Jin Oct 28 '10 at 9:21
  • $\begingroup$ An abstract/symbolic version of a scatterplot would be fine, but if we can find one that has real data behind it that would be even better! Maybe even 20 points is too many? How many would you suggest would work? Sorry I've no idea how big a touchy apple icon is. $\endgroup$ – onestop Oct 28 '10 at 9:40
  • $\begingroup$ @onestop favicon is 16x16 pixels. But I think maybe we can do a detailed version for the site logo, and for the favicon we can do a simplified version of it. We use 158x158 px for the Apple Touch icon. $\endgroup$ – Jin Oct 28 '10 at 12:32
  • $\begingroup$ Just to demonstrate a favicon-able scatterplot is a possibility (this e.g. lacks a fitted line though): sacmeq.org/statplanet/images/icon_graph2.png $\endgroup$ – onestop Oct 29 '10 at 13:33
  • 1
    $\begingroup$ FWIW, I really think that @mbq's idea will work better as a logo and favicon: meta.stats.stackexchange.com/questions/269/…. Also, so far as doing a scatterplot, I really prefer the idea of having a non-linear model (i.e. not a straight line). It communicates the complexity of our problems a little more. $\endgroup$ – Shane Oct 29 '10 at 17:20
  • $\begingroup$ @Shane I agree that mbg's idea will work better as a logo and favicon IF we're using the name CrossValidation. When I posted the above answer I was still under the impression that had been ruled out, however, so I spent a while thinking about alternatives. Seems I wasted my time. I realised that StatExchange was ruled by an oligarchy, but considered that has some advantages over direct democracy when it comes to consistency and speed of decision-making. Seems I was wrong. $\endgroup$ – onestop Oct 30 '10 at 12:40
  • $\begingroup$ @onestop Sorry that there has been so much confusion over the decision. Regardless, I do like the idea of scatterplot (having proposed a similar one), but it may be too complex as a logo, even with a small number of points. We'll see what everyone else thinks... $\endgroup$ – Shane Oct 30 '10 at 15:41

I'm not a particular fan of the way the Math SE portrays when a question has an accepted answered as a faded checkmark. Specifically I don't like how the number of answers is portrayed in a similar green color, and I don't like how much it is faded into the background. I can see it but it isn't immediate recognition of an accepted answer like it is currently on the Stat site.

I like the way it is now with a entire green background, but MBQ had a suggestion of using a normal distribution as the accepted answer symbol which I think sounds pretty cool. We could also potentially incorporate many of those same cross validation logo ideas as an accepted background answer (hence the link back to cross validation again).

  • $\begingroup$ Andy, I will tweak that check mark soon and make it more readable on the math site. That normal distribution curve is a great idea for this site's accepted answer bg. $\endgroup$ – Jin Oct 29 '10 at 17:42

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .