Dot maps are a fun way of being able to show quantities over areas as a nice alternative to a choropleth map. Like the most common way that people tend to think of mapping values for areas is to use a choropleth map, when you do that, your standardizing the data in this case, population per square kilometer. So you take counts for an area, you divide them by the area itself to standardize or normalize it. That gives you a density, and that gives you a choropleth map like this where you're assigning the shades or colors to those classes in order to be able to show a pattern. Dot density maps don't work that way. With a dot density map, you're not doing any of that stuff. All you're doing is showing density visually. So, you have a dot that's worth a certain amount, in terms of the count of whatever it is. Here its population but it could be something else, and the more dots there are, the higher the density is. So, you're showing the density visually, you're not calculating it. This is something I find, especially once people have learned about choropleth maps is that they automatically want when they're creating a dot map to normalize it and they don't have to do that. The whole idea is that you're just saying, where there's more dots, there's more of that thing in this case people, where there's fewer dots, there's fewer of that thing. That's all there is to it. So, I think a fun or different way. It's eye-catching to show variations in values across an area. I think it's something that pretty much anybody could easily relate to without any instruction or experience with this is you can just look at it and say, "Where you see a cluster of dots, there's more of that thing." Dot maps are easy to create with a software, but it was not always so easy to make though. In the old days, pre-GIS, pre-software, pre-computer, dot maps were actually created manually. They were an incredibly time-intensive process. You had somebody who would literally place each dot on a map manually using a pen or whatever methods they were using, and there was a pro and con to that. So, the good thing about that was that those dots could be placed very intentionally or very intelligently. So, if you were say, making a dot map of the population in United States, you'd have more dots around the cities, because you would know where those cities were and you'd be able to place those dots accordingly. You'd have fewer dots say in the prairies, or desert areas, or whatever. So, you could use this auxiliary information as an intelligent human to decide where those dots should be placed. The downside of course, is that to create one dot map could take months. That's right, literally months and so, that's why you didn't see them that often. With the advent of computers, and software, and GIS, you can now make a dot map in a matter of seconds. The downside is, you don't necessarily have that same level of control about the dot placement. There is ways around that, I'll show you a little bit about how to do that, but it's not exactly the same thing. So, yes you can now make them quickly, but it's not quite to the same level of quality that you would get if someone was painstakingly putting those dots in there by hand. That's okay. I'm willing to live with that, I don't really want to be the person putting those dots on the map. So, here I have the City of Toronto. If I treat the City of Toronto as one polygon. So, there's one population value for the entire city, let's say. I then tell the software, "Okay, here's my population for this polygon. I'm going to assign a dot value of one dot equals 1,000 people." What the software will then do is randomly place those dots inside that polygon. So, it takes the population value that I give it, divides it by the value for each dot, which here is 1,000, comes up with a total number of dots that it needs for that area, and then just randomly places them on the map. So, if you look at this, even if you don't know Toronto, you probably could figure out fairly quickly that this isn't very representative of the population patterns of the city. Where's the downtown area? Where the suburbs? Why don't I see patterns and differences and so on about clustering of where people actually live? Well, if you just do it the way the software says, "Okay. Well, if this is what you want me to do." So, it's just do what you told it to do. You're not going to end up with a very useful or effective dot map. But there's a way around that. So, let me show you how that works. If you still have the dots randomly placed, but you tell it to randomly place them inside smaller areas, then you can start to get something that's a little more realistic. So, what I've done here is I have population values for each census tract in the city of Toronto. What I can do is, I'm still using one dot equals 1,000, but when I do that, now it's randomly placing them inside each census tracks. So, some census tracks are going to have a lot more people. Some are going to have a lot fewer people. So, they'll be more dots, where they're needed and fewer dots where they're not needed. So, now I've got a pattern that's a little more representative of the real population patterns for the city. One little tip by the way is when you're doing this method is that you don't actually have to show the boundaries of the areas that you're using to create that dot map. Like for here, it's the census tracks, but I could actually just make those invisible like that. So, that's actually much better. So, now I'm getting that effect of a dot map, I've told it to randomly place them inside each of those little areas. So now we can see that yes, there's definitely more people downtown where you would expect to have clusters of points on higher population density. As you move out to the suburbs, you're getting less population density so fewer points. So that works pretty well. We can actually take that a step further, though and use even smaller areas. So, in the Canadian census system, dissemination areas are smaller than census tracts, but we have population counts for those as well. So, I can use exactly the same method, I can tell it to randomly place the dots, they're still worth one dot equals 1,000, but now it's randomly placing them inside even smaller areas. So, if I do that and takeaway the boundaries again, I get I think an even better version of that dot map, where you really get this nice clustering taking place in different parts of the city. So, you can see there in there. But then there's areas that are much fewer dots so we have lower population densities. So, depending on the size of the units that are available to you and the data that's available, these are ways that you can try to simulate that manual dot placement method is by making those randomly placed dots constrained to much smaller areas. So here we have census tracks versus dissemination areas. Both of them are not bad, but I have to admit I think the dissemination area map looks a little bit better and does a better job. Here's a comparison of a choropleth map to a dot density map, and there's nothing really wrong with a choropleth map. There are very popular. They're used all the time. Most people I think are fairly comfortable looking at them now. Sometimes it's nice to mix things up to try something that's a little off the wall, not even off the wall really but just a little bit different and to use something like a dot density map. So don't always automatically go to choropleth, if a dot density might work for you. There's really not a lot to play with, with a dot density map in terms of the settings. Really, all you can do is set the size of the dot and the value of the dot. So, here the dot size is three points. So, that's what a three-point size dot would look like on the map, they're giving you an example of that. Then the dot value here is 1,000, you can set that to whatever you want. So, between those two things, that's really your main way of varying the way your map is going to look. So, let's have a look at the effects that can have on the interpretation of your dot map. If your dot value is too high. So, here we have one dot equals 10,000, you end up with dot map of the population of Toronto, that makes it look like there's nobody there. That the city's practically deserted. I don't know if there was some zombie apocalypse or something, but whatever happened, the city is practically deserted. We just have these stray dots wandering around the city, wondering where everybody went. So, that's not exactly the impression you want to give somebody is that, just by making the dot value too high, you end up with too few dots, then you end up with a map that doesn't really have any density at all, and doesn't really show people a useful pattern to the data. Here the dot value is pretty good. We have one dot equals 1,000 which is pretty reasonable for this data set. But the dot size is too small. So, even though we have, I think a good number of dots. If this dots are too small, they're hard to see. So, again, remember the whole idea of the dot map, is that you're trying to show density visually. If nobody can see anything that looks like a dense area, if you don't have clusters of dots that are easily visible, then again you're getting this idea that it's not very dense in the city and you're going to have people misinterpret that dataset. So, if you have too high of a dot value, it's not going to work well, and if you have too small of a dot size, it's not going to work very well either. Also, if the dot sizes are too big, you end up with something like this, in which obviously this is a little bit extreme. But the interesting thing is that when they've studied how people interpret maps like this, if these dots are too big, people actually tend to think that it's a crude map. In other words, they start to question the validity of the data. Even though the data is exactly the same, they look at that and say, "Yeah whoever made that probably didn't know what they are doing, it looks crude." So, I don't know if I would really trust the data with this map so you don't want to do that. Conversely by the way, if the dot size is too small, it looks like there's this pinpoint accuracy to the location of it and people will start to think that it's more accurate, when it's really just the same data and the same dots, it's just the way that's it's being shown is different. So, I'm showing you some extremes to give you a sense of what's possible or what's good or bad, but you want to avoid a situation like this. The problem of course with this, is that there's too much density going on. Because the dots are all starting to meet each other, you've got huge parts of the city that all look like they're really dense. So, again, you're losing some of that differentiation, or pattern, or clustering, so the people can look at that and say, "Oh, here's a higher value or lower value." It just all starts to look the same, whether it's high or low, and then it's losing its value as a good map. This map is pretty good. So, we have a dot value that seems to work well. We have a dot size that seems to work well. What you're striving for as much as possible is to have some areas where the dots cluster together and start to coalesce, and other areas where they're still separated out. But you can see these areas where there's definitely some clustering happening, but not too much. It's a bit of experimentation. It depends on the dataset and both the statistical and geographical spacing or distribution of those values, but you have to work with a little bit. So here, you can see that there's definitely some good clustering going on, some areas that are not as clustered. So overall, I think this works fairly well. What you're striving for as I said, is to have coalescence of dots. So, that's what we would call it. In other words, where the dots start to overlap with each other in the densest parts of your map. You don't want to have too much of this, and you don't want to have not enough of it. You're striving for this enough coalescence as I've shown in the example here, where you have some areas where again, remember, you're showing density visually. So you have to have that coalescence in order for people to get that idea, that there's a higher density taking place. So that's basically it for dot maps. There's lots of ways you can experiment with this, with different datasets. Essentially, one thing I didn't mention was colors. You have dot size and dot value, and of course the color in relation to the rest of your map. But I think if you play around with this you'll find that dot maps can be a fun and interesting alternative to just your typical choropleth map.
