How to use color in your data visualization

I want to kick us off today with a quick story I was setting up for a workshop recently for a corporate client and one of the participants came up to me before we got started and he wanted to share a project that he'd heard about that he was hoping to replicate this participant wanted to build an algorithm that would create a unique color combination for something like thousands of different segments within a given graph now he recognized that thousands of different segments was probably overkill so he was going to set his cap at something like 200 so 200 different combinations of varying hues and intensity so never would the same colors appear next to each other twice now he was so excited about this that I didn't have the heart to tell him at that point that color can usually be used more strategically than that now just give you a sense of what we're talking about here the algorithm that he wanted to build would allow the user to easily create graphs that look something like this these are pretty but is pretty the goal when we're visualizing data pretty is the goal when my two and a half year old wants to use every crayon in the box but pretty isn't the goal when we're visualizing data or it isn't usually the primary goal when we're communicating with data now I like to think that as I talk through the various lessons in the workshop that day that I saw a new understanding forming with this participant color when used sparingly and strategically is one of your most powerful tools for drawing your audience's attention to where you want them to pay it now this is one piece of advice I find myself giving perhaps more than any other be thoughtful and intentional when it comes to your use of color don't ever let your tool or an algorithm make this important decision for you so today to really drive at this point home I'm going to talk through several lessons on color I want to draw out one distinction at this point which is the distinction between exploratory analysis and explanatory analysis so I'm assuming in our conversation here that you've already explored your data you already know what's interesting about it that you want to communicate to somebody else so we're in that explanatory analysis space where you have something specific you want to communicate to somebody specific so have that lens on as we're talking through these lessons specifically we'll go through seven brief lessons start off by talking about how attention-grabbing color can be and how we can use that attention-grabbing power to signal to our audience via color where to look but to work as a signal color has to be used sparingly color can carry quantitative value it can also carry tone and meaning but not everybody sees color the same way and finally we'll talk about how color can be used should be used consistently let's kick this off with how color can grab our attention when I start off by talking briefly about how people see this is a super simplified picture of that process on the left hand side we have light reflecting off a stimulus it's gets captured by our eyes we don't fully see with our eyes rather it's what happens in the brain that we think of as visual perception takes place now in the brain there are a few types of memory that are important to understand as we are communicating with data we're going to focus on one of those today which is iconic memory iconic memory is super short-term it's shorter than short-term memory and information stays there for fractions of a second before it gets forwarded on to our short-term memory the really cool thing about iconic memory is that it's tuned to a specific set of what we call pre-attentive attributes pre-attentive attributes are huge tools in our visual design tool belt so it's actually pause here and do a little exercise so in a moment I'm going to put a bunch of numbers up on the screen what I'd like you do as fast as you can is count the number of sevens that you see note your process as you're doing this all right ready set go correct answer is five but this was sort of tough not technically difficult we have to physically read through these four lines of text look for the shape seven check out what a different exercise it is when we make one tiny change this is important because what this tells us is our pre-attentive attributes here I'm using pre-attentive attribute of color or hue to make the sevens the one thing that are different from the rest which means that you pick up on that cue without even really knowing that that's what's happening before you have time to blink before you have time to think now this is super important because what this tells us is our pre-attentive attributes specifically color can be used to help us enable our audience to see what we want them to see before they even know they're seeing it here are the attributes I won't talk through all of these but notice as your eyes scans across the screen how is just drawn to the one shape within each group that's different from the rest you don't really have to devote any conscious thought to looking for it now two of the pre-attentive attributes here are attributes of color hue which is what we typically think of as color red blue green yellow and intensity which is varying levels of saturation of a given color now one thing to keep in mind with the pre-attentive attributes is some carry quantitative assumptions and others do not so for example when it comes to hue if I ask you which is greater red or green this isn't a meaningful question this is important because it tells us which of the pre-attentive attributes can use to encode quantitative information and which should be used as categorical differentiators so hue is typically used as a categorical differentiator whereas intensity can carry with it some numeric value or assumptions of numeric value with higher intensity colors having higher numeric value and vice-versa we'll look at a specific example of hue and intensity when it comes to visualizing information soon via a heat map but in the meantime let's shift on to our second lesson of the day which is about how we can take this color grabbing attention that we just looked at and use color as a signal for where to look so the test that I will often employ when I'm trying to determine whether I'm using my pre-attentive attributes specifically color well whether I'm drawing my audience's eyes to where I want them to go and that is via the where are your eyes drawn test where you create your visual and you either look away from it and look back or you close your eyes and look at it and you just note where do your eyes go first because probably this is where your audience's eyes will go first as well and I want to do this with a series of pictures and talk about some considerations with each so just take note as I flip through these of where your eyes go first where do they go first here most people will go immediately to that big bold red stop sign at the bottom right because of the bright color because we're sort of trained over time that red means danger pay attention because of the big bold capital letters written on it and then we maybe back up from there and read some of the other signs about where do your eyes go here most people will be drawn to the Sun at the bottom left but if you're like me if you try to focus on the Sun you see the plane in your peripheral vision and pulls you a little bit there or if you try to focus on the plane you see the Sun in your peripheral vision and have a some tugging going on when it comes to your attention there this is just a illustrate they want to be aware of this tension that can happen if you're emphasizing multiple different things on a page or within a data visualization let's do another one of these where do your eyes go first here most people will go first to that perennial sale sign that the pink sign in the very middle because of the bright color because of the big bold black text now one thing to know about visual processing is most people when encountered with a page or a screen without other visual cues they'll start at the top left and do zigzagging Z's across the page or screen in this case though that perennial sale sign is so attention-grabbing that we start there and then continue on the Z downward to the right also because of the arrangement of the signs there notice that means that we've missed whatever is happening in that top-left quadrant it's just something to be aware of as you're designing your visuals and the pages that contain them where'd your eyes go here if you're like me they're drawn everywhere and nowhere all at once there are so many things competing for our attention it's impossible to know where to look and that's the downfall of the pretty graphs that we looked at in the introduction with so much competing for our attention we don't have any cues of where we should look color can be used more strategically than this check out the difference between that and this when the orange cran is the one thing that's different from the rest we can't help but look at it we can't help but have our eyes drawn there so when I think about how we can leverage this attention-grabbing power when it comes to our data visualizations first though let's look at how we can have color signal where to look when it comes to the use of text go ahead and give this text a quick scan now without other visual cues it becomes very much like the count the sevens example again where you're faced with this block of text that you pretty much have to read and put on the lens of what's important or interesting then maybe read the block of text again to put the important or interesting things back in context of the rest but I can use pre-attentive attributes color specifically we'll look at here to direct your attention to one part in the text or another or I could direct your attention to multiple places in the text help make it scannable right so we can have the full verbatim comments here in this example but some pithy phrases pulled out that we can see very quickly so we don't have to necessarily read all of the text that's there studies have shown we have on the order of three to eight seconds with our audience during which time they're deciding whether they're going to continue to look at what we've put in front of them or move their attention on to the next thing if we've used our pre-attentive attributes well even if we only get that first three to eight seconds we've gotten our main point across now as you can imagine pre-attentive attributes are also hugely useful when it comes to visualizing data so get a few examples first let's imagine that you manage a bus fleet one thing you might be interested in understanding is how cost per mile varies according to miles driven so here we could use pre-attentive attributes to quickly draw attention to the cases where a cost per mile is above average so we can see if we drive less than about 1700 miles a month more than about 3,300 miles a month our cost is above average for another example here we're looking at survey data customer feedback from our annual survey last year in 2014 versus this year in 2015 you can see how we've done across a number of categories if you want to quickly draw your attention to the category where we saw a decline we can do so through that sparing use of color so get one more of these imagine you work for a car manufacturer one thing you might be interested in is understanding the top 10 design concerns which is what we have listed here on the basis of concerns per 1000 now we could use color sparingly to draw attention to some of these right we might want to talk about some of the top design concerns whether these are acceptable to fault rates within this group I could further refine the story to point out some of the issues in common perhaps paired with some explanatory text now one word that I've been using throughout these examples when it comes to the use of color is sparingly because color really only works as a cue when it's you used sparingly it's easy to suck and a sky full of pigeons this is an analogy that Colin where talks about in his book information visualization perception for design the analogy is it's easy to spot a hawk and a sky full of pigeons but as the variety of birds increases that Hawk becomes harder and harder to pick out in other words the more things we make different the lesser degree to which any one of them stand out or if there's something that's very important we should think about making that the one thing that's very different from the rest now this idea applies very much in data visualization here's an example of color not used sparingly in this example imagine you work for a US retailer and you're interested in understanding the distribution of your customers compared to the general population so the segments listed on the left hand side here could be any demographic measure age groups for example if we think about applying the where are your eyes drawn test here similar to that box of crayons where there is so much competing for our attention our eyes go nowhere and everywhere sort of at the same time now if we stare at this for a while we might notice the red box on the right and use that as a signal to say okay I think I'm supposed to concentrate they're not totally sure why because we don't have any context but let's ignore that for the moment check out the how we can use this same construct but just use our color more sparingly to make it very clear where to look now this example has been fairly generalized so that we're not sure still why we're concentrating here but we at least know very quickly because of the sparing use of color where we're meant to pay attention something to keep in mind with color is that it can carry quantitative value we talked about this briefly when it came to hue and intensity and we were looking at pre-attentive attributes I want to dig into this idea a little bit more so what we're looking at now is country level sales rank of top five drugs so this is an example from the pharmacy industry free read the subtext at the top it says rainbows distribution and color indicate sales rank in given country from number one in red to number ten or higher in dark purple here we're using color as a categorical differentiator if I apply the where are my eyes drawn test I go first to number one in red which is good but then I go to ten and twelve and eight these purple and dark blues because they're in higher intensity of color but when we think about where we want our audience to pay attention that's not necessarily the order we want them to process this information for those of you who grew up in the 80s like me you may appreciate this reference so they table that Rainbow Brite would love but I don't like it so much because we can use our color more thoughtfully than this specifically instead of using color as a categorical differentiator through varying hues we can think about varying the intensity which does carry with it some quantitative assumptions so in this case I've made the market leaders the highest saturation of blue color going to the lowest saturation so now if we consider how you process the information I go first to the ones then to the twos then to the threes and so on this is a more thoughtful use of color so color can carry quantitative assumptions color can also carry tone and meaning which means we want to think about what is the tone that we want to set with our data visualization or with the broader communication in which our data visualizations sit and think about how we can use color to underscore that desired tone now any search on Google for something like color and tone or color and meaning or color and emotion will pull up visuals that look something like this now if we look at some of these different colors black is elegant bold powerful blue confident classic red aggressive speed danger yellow down at the bottom so funny story I had a rental car when I was traveling in Seattle last week and all I had to do is look at this car and it made me happy which was sort of bizarre because you wouldn't expect that reaction with a car necessarily but this particular car was a bright yellow VW Beetle now I would never buy a bright yellow car but it made me happy every time I looked at it and when we look at yellow down here tends to represent things like youth-friendly positive feelings sunshine and so on we get over to the other side though we get to hot pink and it's expressing ideas like exciting playful tropical flirtatious which may have its place probably isn't in our quarterly board report for example so you just want to be aware of this tone and emotion that color can evoke when it comes to your data visualizations let's take a look at this in action so here is a generic graph I'm going to render it in some different colors and just notice how the graph feels different when we use different color to emphasize as I typically design my data visualizations in shades of grey and then use a medium blue very sparingly to draw attention I like blue because you avoid colorblind issues which we'll talk further about momentarily it also prints well in black and white if that's a concern but one time I got told by a client who had done some makeovers for that my visuals look too nice as in they were too friendly they were reporting the results of statistical analysis and were used to and desired a more clinical look and feel so in that case I remade my own makeovers substituting the blue with both black and the graphs as a result of that simple change felt and looked totally different from the originals you could think about leveraging just an outline to have an assumption of opportunity my latest blog post is actually considering that idea further notice that red trending upward feels like a negative thing whereas green trending upward feels like a positive thing and I've done nothing to change the actual data here the only thing I'm playing with is color now if you have a company logo or brand colors or your client has brand colors that you want to work within that's awesome it's one way to give personality and cohesion to the data visualizations that you create for your brand or your clients brand but if your brand is colorful just be aware that it doesn't mean you have to use every brand color in your graph alright pick one or maybe a couple brand colors to use to draw attention and here's the flirtatious home version of the graph rendered in hot pink slipping through a magazine one time and there was a sort of fluffy article on online dating and their visuals were entirely hot pink and teal and that works given the sort of fluffy nature of the content but as we mentioned before not appropriate for something like your board report probably unless it's a brand color which then is a different consideration now we talked about some of the tone that color can have one thing to be aware of is the tone the assumed meanings associated with color can vary depending on where you grew up where you're from so especially in the case where you are communicating with an international audience and make sure you take some of those nuances when it comes to color into account now what we have in front of us now is the colors and culture it's a color wheel done by David McCandless that's both at the same time sort of beautiful data visualization as well as a useful tool for visualizing data so if we look at this for a moment we're in western Americans that's a which is the outermost ring around the circle here and we can look along it to see what different colors represent here so red is 41 so red means Heat blue 39 means healing yellow is happiness right thinking back to my yellow VW Beetle and and so on and so forth and so you just want to be aware of the tonal and meaning impact that color can have both on your data visualizations as well as the broader communications in which they sit and be thoughtful about what colors you use also should be thoughtful that not everybody sees color in the same way specifically about 10% of the population 8% of men and half a percent of women are colorblind which typically manifests itself as difficulty in distinguishing between shades of red and shades of green which means in general we should avoid using shades of red and shades of green together or if you do have to use red and green together right if you want to leverage that that connotation that red it went down that's bad green it went up that's good you can do so but make sure you have some additional visual cues also present make the numbers bold ensure that you have the plus or minus sign in front of them do something else to set them apart visually so you're not inadvertently disenfranchising part of your audience which brings us to our final lesson of the day on the consistent use of color so if you have five sales regions and you show them in one color scheme in one place in your report or presentation you want to try to maintain that same color schematic throughout the rest if you're lucky your audience will familiarize themselves with the details of what they're looking at once and assume those same details apply and in general if you can you want to try to avoid using those same colors for other purposes so care to a recent example from the media so these are some graphs from The Economist they accompanied an article on the changing role of women in Latin America and I like these graphs overall they they're nice they are clean looking they do a nice job of capturing the information but there are some slight changes that we can make that will go a long way in even more effective the sacred closer look at what we're looking at here so the graph on the Left shows labor force participation rate as a percent of total we start with the developed countries on left Middle East North Africa East Asia on the Pacific and then finally Latin America on the right for each of those we have time points in both 1990 and to 2013 red depicts the percentage of women in the labor force and blue depicts percentage of the men in the labor force and one of the things I really like about this is how the gap between the two is highlighted so we can see for example the gap is narrowing when it comes to develop countries in Latin America we see a sharp increase in female participation rate in the labor force which is awesome but still falling short of males one thing to a couple of things to notice here one is that women are rendered in red and men in blue and then secondly within Latin America we see darker colors used to emphasize they're also with the bolding of Latin America versus brighter colors were used in the other regions now if we come over to the right we have our first horizontal bar graph shows women in Parliament percent of he seats held and then at the bottom of women CEOs again on a percentage basis now whereas women were rendered in red on the left now we're showing this in blue which seems like it should be tied to men on the left but we're talking about women so that's contradictory the other thing is the use of bright and dark colors are switched here so on the left hand side in the slope graph Latin America was rendered in the darker colors versus on the right hand side and horizontal bar graphs Latin America and the Caribbean on the top and Latin 500 on the bottom are the brighter color not the darker color so again minor but it increases the cognitive burden of the processing of this information and there's no reason not to use consistent coloring here so if I were remaking these visuals I'd want to do them in red I would want to have Latin America be what's in the darker color to make it tie back to how we're using how on the left-hand side of the visual now for me this is a little tangential but these only tell part of the story the way they're depicted currently when I look at the top graph looks to me like Latin America in the Caribbean is doing great when it comes to percent of seats held in Parliament by women if I come down on the bottom graph it looks like you've latinum 500 not doing so well when it comes to CEOs of public companies but if we add in the full picture so let's add in the men component in both of these cases now when I interpret this top graph I see yeah Latin American Caribbean has a higher percentage of women than these other areas but there's still a lot of opportunities when it comes to balancing the men and women proportions in Parliament and when I come down to the bottom no longer do I say whoa Latin America is doing horrible rather I say everybody's doing horrible when it comes to this there's room for improvement everywhere yes the Latin America lags but it doesn't like buy so much so just changes the way I interpret the numbers and again here I've used blue now to tie back to the blue that was used on the Left who represent men and continued with that darker emphasis of color in Latin America to tie with how color was used on the left hand side so that brings me to the end of our lessons on color today just to quickly recap we started off by talking about how people see and how we can leverage the pre-attentive attributes of color to grab our audience's attention we can use that attention-grabbing power to signal both in text as well as in data visualizations where our audience should look but to be a clear cue on where to look color needs to be used sparingly color can carry quantitative value can also carry tone and meanings you want to think about what is the tone and meaning you want to convey with your overall data visualization or broader communication think about how you can use color to underscore that recognizing though that not everybody sees color in the same way and also that color should be used can system so what we've looked at today a couple key lessons storytelling with data visualizing data effectively these are much broader lessons that can be covered in much much more depth and I've actually covered them in much much more depth in a forthcoming book that will be out at the end of October called storytelling with data and this will just give you a sense of what the chapter breakdown looks like within that six key lessons and then a whole lot by way of case studies and examples and insight into my design process the lessons that we've looked at today on color are just a subset of what you'll find in Chapter four on focusing attention so if you like what you saw today definitely check this out for many more examples and lessons when it comes to the effective use of data in communicating with that I say a very big thank you for joining

Channel: storytelling with data

Views: 33,466

Keywords: Color (Quotation Subject), Data Visualization (Industry), Communication (Industry), storytelling with data, how to use color in data visualization

Id: AiD6etOB6qI

Length: 29min 36sec (1776 seconds)

Published: Thu Oct 01 2015