When to use a table vs. a graph

hello my name is Nick Deborah I'm an independent instructor and bestselling author in the fields of data visualization and dashboard design and in this short video I want to talk about uh basically the first design choice that we need to make when we're creating

a new chart which is am I going to show the data as a table or as a graph and this is actually a question that I put to my workshop participants when I'm teaching my practical charts course uh

I'll ask them basically like when would they use a table versus a graph are there certain kinds of data which they would show as a table rather than the graph certain audiences uh or you know

what are the strengths of tables versus uh versus graphs and I get all sorts of answers and so I want to kind of run through the most common responses there's like eight or 10 of them that I get uh when I ask people this question

probably the most common answer is well when the audience needs to see exact values and of course the table is going to give us that right we're going to get the exact values right down to the last

dollar or decimal point or uh or whatever but when I'm in a situation like this I'll almost always also show a graph because as we'll see graphs have some pretty significant uh advantages

and in a situation like this we kind of get the best of both worlds right we get the advantages of a graph and we get the high Precision of uh of a table another common answer is when you need to quote

unquote just show the data without saying anything about the data because people generally think of tables as being kind of neutral or sort of interpretation free uh but I don't think

that's actually true uh even even with a simple table like this where we're just looking at our expenses and budget for five departments there were actually a lot of different ways we could have designed this table for example we could

have shown the uh deviation of expenses from the budget in dollars deviation in percentage we could have shown a breakdown by uh by region instead of by Department we could have added a total

or not or we could have shown uh breakdown by quarter right showing how these are changing over time or by month and on and on and on and so uh in that sense you know tables aren't really

different than graphs because there are all sorts of design choices that we need to make based on assumptions about why we think the audience needed to see this information and so uh I don't think

they're any more or less neutral than uh than graphs and they're just not the kind of objective or interpretation free charts that uh people often assume them to be uh

unfortunately another common answer uh is uh you know people say oh I need to use a table when I want to show both details or parts and totals uh in the

same chart right so here we have some uh some details and also totals and of course the the details or breakdowns or parts are always going to be smaller than the total so people think oh I have

to use a table here well actually there are uh stacked bar charts and stacked area charts and other chart types that can allow us to very comfortably show data as a graph even when we're showing

totals and breakdowns in the same graph another common answer is people say oh I need to use a table because my data has outliers right values that are much

greater than the others and they tend to think oh you know my chart is going to be messed up because the smaller values are going to be too hard to see however using one of my favorite data visualization tricks which are inset

charts uh we can actually show data without outliers as as a graph right so we kind of get the best of both worlds we can see the outlier but we can also kind of blow up the smaller values to see them clearly and this works with

pretty much any chart type right we can use this with line charts and Scatter Plots and like I said just about anything else so we didn't have to use a table here just because we had outliers or uh when the data contains

different units right and so here I've got some donation data and I have an average donation amount and doter satisfaction rating and average household income apples oranges and bananas and people think oh got to show

this as a table well no we could use uh what I call a merged chart in this in this case I have three uh bar charts that are merged together and so we didn't have to use a table uh and I

could even use uh what's called an index chart I can do some math and kind of normalize these values so that they're all in the same scale like for example change versus the you know first day in

this case so again didn't have to use a table just because I had different units uh often times people say oh you know uh we have to use a a table when our data has a more complex structure

right lots of variables in this case we have product categories and we have different uh units of measure like sales in dollars and units and and profit an average discount again apples oranges

bananas and we also have time right so lots of variables here actually we can show this kind of information as a graph this is trickier it does require some uh some practice and you know some thought

to be put into it but if we think about the alternative which is to show people a big wall of numbers something like this is going to be probably more useful even though it'll take them a little longer to grasp it it might take them 15

or 20 seconds to understand a chart like this but it's still way easier though looking at a big ball of numbers uh another common answer is when the audience prefers tables they tell us

say you know I'm a table person I've been looking at tables all my life this is often what people with financial backgrounds will say you know accountants CFOs but when I say that I I do a simple demonstration with them and

I can do the same thing with you you can try it yourself I'm going show a simple table like this um we're just looking at monthly revenue for two regions across

12 months so 24 values pretty simple uh but now I want you to try to spot any sort of patterns or relationships that you think might be interesting in this even if they're fairly obvious right

like for example uh the East is always higher than the west and if you like you can actually pause the video and see if you can spot maybe three or four potentially interesting patterns or

relationships uh within this data so uh if you give that a try now I'm going to uh give you the same task but I'm going to show this exact same

data as a graph instead of a table right Okay I lied I'm actually not going to ask you if you see any insights here because you're probably seeing like four or five of them right off the bat within the first one second right you're seeing

that the East is uh trending slightly upward whereas the West is flat the East is more volatile than the West there's that kind of double dip in September uh the East is has that sort of cyclical

quarterly pattern which the West doesn't have right and so we're much not only much faster but we're also much better at spotting patterns and Rel relationships in graphs than we are in

tables and cognitive scientists have actually understood why this is the case uh for for quite a while really since the 1950s or so and so when I do this demonstration with people have financial backgrounds even they're often like oh

my goodness you know what have I been missing by only looking at tables another common answer is oh we have to use a table when we have lots of values when you think about it that's

actually probably a good time to use a graph because if you look at a you know a tree map like this which showing a 100 values imagine what this would look like as a table right be a huge wall of numbers very difficult to read and in

fact that's the case you know whenever we have a large number of values the alternative here is a massive table that is going to be extremely difficult to interpret so I think this is really a reason to actually use a graph rather

than using a table uh or when none of the values are actually directly comparable to one another and this is a case where I think actually using a table would make sense it'd be a situation like this right

we're showing a bunch of different key metrics different kpis and we have funds raised in millions and we have you know admin overhead percentage donor satisfaction on 10 apples oranges

bananas and if we were to show this all within at least within the same graph it'd be kind of confusing it'd be kind of weird looking right you know we're we're sort of invited to compare these

values directly with one another and yet they're not really directly uh directly comparable and so this is a case where I think yeah using a table with no accompanying graph would be probably the

right choice and so where does this leave us are there times where it would make sense to use a table only with no

accompanying graph yes but pretty rarely in fact uh myself I will only use a table with no graph in one of three situations the first is the one we just

saw right where there none of the values are directly comparable with one another and so uh it's going to make sense to use a table with no graph uh the other case would be when the sole purpose of the chart is to look up values right I

know that this chart is only going to be used in a meeting where people are going to be asked right specific questions like what were our insurance costs in Q4 of last year it's just easier to look up values in in a situation like that as

long as they're not going to be trying to compare values with one another or look for patterns which of course um you know would be much easier in a graph so it doesn't happen that often and then

the third one is where the sole purpose of the chart is to allow the audience to import the data into another application right so so sometimes the audience might ask us for a chart but they're just going to like copy paste it into Excel

or or some other application in which case of course you know don't waste your time making a graph because they're not even going to look at it so ultimately not very often right that we're going to

have a table with no accompanying graph so uh by the way this is kind of just a little tiny piece of my practical charts course which lasts two full days

in person or four half days online right we talk about table and in fact many other things but like how to design tables well for example if you're interested in uh maybe you know improving leveling up your data visualization skills by taking the

course there are three ways you can do it uh I now have an on demand version of the course 6 and a half hours of video and 45 lessons at a affordable price uh I typically do three or four live uh

public workshops every year open registration uh some of them in person some of them online uh but most of the workshops that I do are actually private workshops for 15 to 60 employees so same

organization either live uh typically via Zoom or at your location anywhere in the world I'm up to I think uh 16 countries at this point information about all of these can be found on my website practical

reporting.com thanks very much for listening