Have you ever gotten frustrated adding an image into your report,
especially in word? The text boxes don’t line up quite right, images jump
around, things get anchored in a weird place, you can’t resize anything… and
you want to throw your computer against the wall. Well, maybe that’s just me.
Although I have been writing reports for nearly a decade, it
was only recently that I learned this cool trick that would have saved me lots
of grief and frustration.
Recently, Andrea and I were working in tandem on a report.
We each took a section, set a timer to draft it out, and got busy. I was in the
zone. Chugging away, racing to get as much done as I possibly could before the
timer went off. I decided not to get bogged down in adding figures, it was just
too frustrating for me to add them in word. Instead, I created a couple of figures
in excel, and added placeholders for them in word. I would deal with them
later. I glanced over at what Andrea was doing. Cool and confident, Andrea was two
pages in AND HAD FIGURES!!! She saw my eyes widen and said,
“Hey, have I ever shown you this cool way to add figures in
Word?”
“No…”
“You just add a three-row table in word and drop the figure
in. Make the top row the title, the middle row the image, and the bottom row for
the note or source.”
My jaw dropped.
GENIUS!!!
Here is a step-by-step process for inserting a figure in
word (or PowerPoint) using a table.
STEP 1: Insert a 3-row table
(Insert->Table)
Three-row table
Note: You can add columns if you are inserting a series of figures that all have the same title; the columns can be the same width and the rows can be the same height, ensuring that your figures are all the same size.
STEP 2: Drop in your figure, add your title, and your source/note.
Place the figure in the second row, the caption in the top row, and the note in the bottom row.
Image inserted into word using a three-row table.
I just copy and pasted my figure from excel into word. I can
easily make changes to the figure in excel and it will update in word. I can
also make minor edits to the figure in word (like color or font) as well.
Now these three elements – title, figure, and note – are all
grouped together. You can change the size of the figure as one unit, move the
figure around without throwing off anchoring. Text in the remainder of the report
can wrap around the figure in multiple ways, depending on the report layout.
Wait! Before you move on to your next figure, there is one final step. Obviously, you do not want the table borders to show up, but it is helpful to see the borders while you are laying out your report and copy editing. Let’s get rid of them.
STEP 3: Make the table border invisible for the final report, but visible while editing.
Select table, then right click and choose “No Border”
Now you’re done!
Here’s the final result:
Final result with borders only visible while editing.
A similar process exists in Microsoft publisher, but that
software is not available on a Mac, and since many of us are working from home,
we might not have access to our regular office setup. Also, sometimes, you just
need to use word because not everyone has publisher.
In any case, if you are using publisher, you can insert an
image placeholder and add text boxes for the title and note/source and then
group those elements together. The difficulty with that process, however, is
that you have to resize the figure in excel prior to inserting it into publisher
or your figure might get wonky. Translation: you need to know the overall layout
of your report so that you know how much space you have for your image prior to
inserting it into your document. In word, the image resizes appropriately with the
rest of the elements in the table.
Hope this trick helps you as much as it helped me!
I’ve talked to many, many people who analyze data as part of their jobs who have expressed an interest in R. But people are sometimes scared to take the plunge. Here are four reasons you might want to consider it.
1. You’re tired of being locked into $$$ contracts with SPSS, SAS, or STATA
When I went out on my own, I went from having a free version of SPSS (well, free to me!) to having to pay for it. I shelled out a not significant amount of money (for one year), for a basic version of SPSS then got to work.
On my very first project, I realized I needed to do a logistic regression – a pretty basic procedure. I’d done it hundreds of times on my work computer. But the menu option was not available on my screen. What the heck?
Contacted support. My issue? Logistic regresssions were not supported in the basic version.
Seriously? The module I would have needed to download was another several hundred dollars, for ONE year of access, and for literally ONE type of analysis. Craziness. And not doable when you are on a shoestring budget.
In fact, as of press date, to get the professional version of SPSS 26, with (I assume), all packages installed, the cost is $5730 per year. Per person!
That’s insane. Guess how much R costs: NOTHING. It’s free!
2. You realize there’s so much that your stats program can’t do.
Propensity Score Matching has become a big area of interest in quantitative evaluation work today. Can you do it in SPSS?
Nope. That’s right, you’ve paid potentially thousands of dollars for an annual contract, but the program can’t do what you need it to do.
[OK, you can go to this page and download several python extensions that maybe will work. Check out this long list of instructions you need to make that happen. I’m sure you have to pay extra for the python integration!]
In R, getting the software you need on your computer is literally as easy as typing “install.packages (‘MatchIt’) .
And when a new trendy statistical procedure comes around? You can bet it will be available on R before any of the other packages.
What’s available now? Pretty much anything you can think of.
There’s a package that helps recode data easily. There’s a package that will help you automatically write your results to EXCEL. There’s a package that automatically scores survey data, even reverse coding items on the fly. There are tons of customizable graphics packages, if you want to go down that rabbit hole.
3. You are tired of always starting at square one with your data.
When I worked for a local school district, we had a climate survey that was given to every student, staff member, and parent. Tens of thousands of surveys needed to be processed, the results tabulated, and distributed online in the form of reports.
Each summer I dutifully spent one to two months preparing all of the data so it could be placed back into our survey report software, which cost over $20,000 up front and came with a $2000ish yearly maintenance fee. I spent a lot of time writing down procedures in a handbook so I would remember what to do the next time around.
One year, we had a small subset of parent survey data come in a month or so after the official close date. Now we had to make a choice – redo all of that work again to integrate the new data (and delay the reports by another few weeks), or ignore it.
It didn’t have to be that way!
With R, you write a script one time and then you’re done. You can reuse – and modify – that same script over and over, saving jillions of hours of time, and reducing the possibility that you’re going to make a mistake and not notice it.
So that late data that came in would have been loaded in with all of the other data, and the results would have been ready immediately. All in all, it might have added an hour or two to the process at most instead of WEEKS.
4. You know there are things you’re doing in EXCEL that you shouldn’t be doing.
Do you go through your data set and use “Find and Replace” to change “Strongly Agree” to 5, or “Yes” to “1” before importing into your statistics software? That is SO easy to do in R.
A list of EXCEL things you can and should be doing in R instead:
Find and replace
Merging data sets (VLOOKUP, and manual ways)
Sorting data
Filtering data
Creating a clean ‘final data set’
Are you ready to give R a go? Go here to download RStudio Desktop (yes, there is a paid version, but the free version will do all you need and more).
During these trying times, I have been calling on skills I
learned in my previous role as the program evaluation coordinator of social and
emotional learning in the Austin Independent School District (AISD). Sometimes
I need help, sometimes my kids need help, sometimes my husband needs help,
sometimes a friend needs help… sometimes we ALL need help.
As is the case with all of us, these days have been an
emotional roller coaster. Some days are great, and some days are not so great.
I take on most of the childcare duties for my 5-year-old and
20-month-old and one thing that has helped me is establishing a routine for each
day. We weave in zoom meetings for my two children’s gather times, dance
classes, and a children’s time at our church. None of that includes my work! Finding
time for phone calls or Zoom meetings outside these limited hours is difficult,
but I make it work sometimes requiring a muted call, blocked video, or both. My
only alone time is working early in the morning before the family is awake so I
can have some solid uninterrupted work time (and sometimes, like today, that
only lasts until 6:30 because both kids decide to get up early). I work during
nap/quiet time depending on how long the baby naps – sometimes 45 minutes,
sometimes 3 hours. Sometimes I still need to finish up something after the kids
go to bed at night. This all makes for very long days for me without any
breaks. Why do I do this? So that I can continue finding meaning and purpose in
my life. Otherwise, everything in my day will be consumed by caring for others
and Covid-19.
Generally, I am doing okay, but there are definitely moments
when I am at my wits end, when I need to remember to breathe, when I need to
take some time for myself (even if it is just to take a shower!). I am not
saying this to make you feel bad for me, there are many people in a much worse
situation, and things will probably get worse for all of us before they get better.
I have it pretty good, all things considered. We are all together, we are all
healthy, and we all have a job (except for the kids and dog, but they are off
the hook). That isn’t to say that I am not dealing with an unanticipated life
change, job security issues, or family stress. This is a time of great anxiety
for all of us, regardless of race, class, or gender.
So how do I handle this and try and keep calm? If I’m being
honest, many days I don’t! Some days, I have a meltdown. Some days, it is just
too much. Some days I am just okay, and all of this is okay. What I do have are
skills, tricks, and activities that I have learned from the wonderful AISD SEL
team to help keep me grounded (also, check out these resources from CASEL, Panorama
and the American
Evaluation Association).
Cool tip: All SEL-related activities and tools for
kids can be used by adults. Do what works for you! Look over my list and see
what speaks to you.
Breathe. I know this sounds silly but breathing helps… a lot. Find an app and take a mindful moment if you can. Even if it is just for 2 minutes. Some apps have great meditations for kids and adults (like Calm, GoNoodle, or Mind Yeti) which can be helpful for the entire family. Do yoga with your dog, cat, or family! Do whatever you need to do to find some inner peace – if only for a few moments. AEA even offers meditation practices.
Go outside. See the beauty in nature. Walk, run, ride your bike and take a break from the media. Think of things bigger than yourself. [Andrea’s note: And, bonus, Vitamin-D is an important protector against COVID-19.]
Practice self-compassion. Talk about how you are feeling. Call a loved one, a friend, a family member, a colleague, anyone. We are all in this together, and chances are someone else is probably experiencing what you are experiencing and feeling what you are feeling. If you can’t talk to someone on the phone/FaceTime/email, write a letter to yourself as though you were a friend or loved one. What would you say? Treat yourself with the loving kindness with which your friend would treat you.
Get creative! Make something. Bake, cook, build, make or listen to music, create art. Creating something gives us a sense of purpose. Coloring is often meditative and therapeutic. Take some time and color – inside or outside the lines! This can also help if you are out of work or are no longer finding meaning in your work. We all are doers and want to feel like we are contributing, and sometimes that means baking cookies.
Take a break. Sometimes we just need to take a break. Those emails can wait. Ask your partner to watch the kids for a couple minutes. Reschedule a phone call. Take some time for you. Doing so will help you will return to your work, family, or friends in a better place than where you were when you left them. Sometimes this can even mean taking time to be a parent! Stop and look at trains with your kids, go on a scavenger hunt, play baseball, fly a kite. Do whatever you need to do to be present in the moment with them. They will appreciate it, and so will you.
Reflect. Take some time to write down your feelings. It is easy to get caught up in the negative stories and all of the uncertainty right now. Start a gratitude journal and think of positive things to counterbalance the negativity.
If you are experiencing severe anxiety or depression, reach out for help. Most therapists are offering telemedicine, so please do not isolate yourself. Contact 1-800-950-NAMI (6264). Or visit https://www.nami.org/Find-Support/NAMI-HelpLine
What other tips do you have? Let us know and we’ll add them
to our list!
One final note, since a majority of us are working from
home, the days can begin to bleed together. Do things to distinguish the days
from each other. At our house, we have family movie night on Friday which we
all look forward to. My daughter has dance on Saturday, and we do a “church” of
sorts with my parents on Sunday. Find a routine and keep to it. Find things to
look forward to, like movie night. Stay engaged with others in your family and
community. This is hard, and will likely get harder for all of us as the weeks
progress, but together, we can get through this.
Andrea and I have mentioned it before: we are in a golden age, albeit a morbid golden age, of data and data visualizations. One way we can stay ahead of Covid-19 is by reading the reports and analyses that are readily available to the public. Sometimes this information can be super helpful, and sometimes the story gets lost in the data.
Recently however, I stumbled upon a site which examines data on COVID-19 and consumer behavior. The referenced survey was administered in mid March.
Take a look at the below charts. (These are similar to the ones from the site but have been remade in Excel.) As you look at them, see if you can spot some areas for improvement.
Data source & graph inspiration from Astound commerce; figures remade from data Note. Survey administered in March 11-13, 2020Data source & graph inspiration from Astound commerce; figures remade from data Note. Survey administered in March 11-13, 2020 Data source & graph inspiration from Astound commerce; figures remade from data Note. Survey administered in March 11-13, 2020
There is a lot to take in. Like, a LOT. These figures are pretty overwhelming.
Before I get into it, let me be clear: it is not my intention to trash talk the folks who created this. I am sure they worked really hard on it and that it took them a ton of time. Not to mention that they are working on top of the stress of COVID-19, childcare issues and job security. We are all in this together. I’m simply looking at this as data and thinking of ways to make it easier for readers to understand the story.
What these charts do right
The percentages are right there on the chart and I like the use of icons. Focusing on consumers is an interesting reference point. I haven’t seen this side of the COVID-19 story before, which is why I thought it would be interesting to highlight in our blog. Overall this chart succeeds at drawing in the reader, but at that point things get murky.
What’s missing?
The use of different types of charts is confusing and might lead the reader to think we’re looking at different types of data, or data in a different way. Really, I think the graphic designer was trying to create visual interest by breaking things up a bit, which is admirable, but not at the cost of easy interpretation. Additionally, the colors used throughout are really similar and readers must carefully search the legend over and over.
We ended up with some questions. What are the key takeaways? What are we really comparing? What is that wheel shape? And why are the hands floating?? These are not questions that you want your readers to be asking themselves, because in all likelihood, they will just walk away.
Let’s Do a Makeover!
Let’s start with the donut chart and the bar chart.
Data source & graph inspiration from Astound commerce; figures remade from data Note. Survey administered in March 11-13, 2020
Here are
some ways I came up with to make the story pop a little bit more.
Change to a Lollipop Chart & Add Icons
First, I redid the donut charts – which are not a great substitute for a bar graph – and used my favorite chart type: a lollipop chart. I added images of each country/region for some flair and to make it easier to see which country was which. Now readers can more easily find the story behind the data. It is easier to see that respondents from the Middle East expressed greater fears concerning COVID-19 than did survey respondents in other countries.
Add a Graphic
Andrea edited this to include a graphic (a scared face) because we planned to use the chart again, and she wanted to make sure the two charts were distinctly different at-a-glance. I think you could leave out the graphic out though.
I also made the chart title more informative. I gave a short statement of what I felt was the main message of the chart. Then I added some subtext. Readers should get enough information from the title to be able to easily interpret the chart.
Middle East Citizens Report Highest Fear of COVID-19.Collectively, three quarters of all people surveyed had fears concerning COVID-19. Fears were similar for the US, Canada, and Europe, but much higher among those surveyed in the Middle East.
Let’s take a look at the second chart with the floating hands. We could do another lollipop chart. The use of the same chart type informs the readers we are looking at similar types of data and it invites them in to make comparisons. I bet that the story will pop here for you, even without a helpful title.
Before:
[Before makeover] Data source: Astound commerce survey – March 11-13, 2020
and here’s the after:
Middle East Citizens Report Largest Behavior Changes Due to COVID-19.Around half of people surveyed in the US, Canada and Europe had altered behaviors, but over 80% of those in the Middle East had made changes.
[After makeover] Data source: Astound commerce survey – March 11-13, 2020
The response options are included as a bar within the chart, making it difficult to read. It took me a while to even notice the survey questions, probably because the text is insanely small, making it difficult to read (particularly when you have old eyes and got up at 5:30 AM to get some work done before the kids wake up and chaos ensues… but maybe that is just me!).
It is also difficult to know what we are supposed to compare. Are we interested in comparisons across country? Are we interested when participants think COVID-19 will be under control? Do responses within country and across survey options add up to 100? All of these questions are difficult to answer.
Slider chart!
For this chart, I am going to use a slider chart (which is basically two lollipop charts combined). This makeover was a bit more complicated, but I think it tells a better story. First, I combined categories such that we now only report “within 1-2 months” and “6 months or more” (combining 6 months, one year and over a year). I felt like these two categories were more meaningful than looking at each one separately. I then created the two lollipop charts, added text boxes for labels and added my icons.
Color!
I selected the two tones from the green color family since the data were depicting participants’ likelihood to spend money. The main point is that we want two different tones in the same color family so people can distinguish them. I chose a deeper color for the “a year or more” category because that response was more intense.
Now the story is much more clear.
Now we can clearly see that most survey participants believe COVID-19 will be under control within 6 months. As always, a descriptive title helps.
Across regions, half of participants felt that COVID-19 would be under control in 6 months or more. Only one-third of participants across regions felt that COVID-19 would be under control in just a couple of months.
As a business owner, I would find this information helpful. I could mentally prepare for customers to not feel comfortable coming back to my business until 6 months at the earliest (but maybe more). This information might help me plan what to do to pivot my work for the next 6 months to keep things afloat until our economy starts to normalize. Obviously, opinions may have changed since the time this survey was administered, but it probably still gives business owners a good ballpark.
Regardless, this story got lost in the jumble of infographics. I’m sure there are other stories buried in the data, but that is what is fun about creating data visualizations. Sometimes you do not know the full story until you see it.
What are some key takeaways?
Multiple pie charts (and their cousin, the donut chart) are difficult to interpret, and it is usually best to represent the data differently.
Make data visualizations clean and not too busy. Use white space, and make sure the reader can walk away from your visualization understanding the information without having to read a full report.
If presenting several visualizations together, make sure the images/graphics/colors all work well together and tell a consistent story.
Use icons and images to draw interest and help the reader interpret the graph.
Those of us who have little kids have become a tribe through this COVID-19 pandemic. A tribe of multi-tasking, always on, juggling being full-time parents, working, and attempting to be teachers, too. It’s a lot.
Sometimes, balls get dropped.
Even my 3-year old has Zoom meetings, now
Here’s an example from my very first at-home meeting during Covid-19.
It was a morning meeting, and my 3-year old son had just woken up. My husband had stayed up way too late working and was fast asleep in the bedroom.
“I’ve got this,” I thought. I poured a bowl of cereal for my son, went to my office, and opened my computer, ready to log into the Zoom meeting. As I was about to click the link, I heard:
“MOMMY! I need you!!!”
OK. Change of plans. I decided that instead of logging in to the computer, I would dial in so I could be more mobile. I grabbed my cell, logged in, and walked to the kitchen. I found my son on the verge of a meltdown.
“Mommy,” he said, through teary eyes, “I need you to help me eat my cereal!”
Meanwhile, the call had gone through and I could hear other people on the line saying good morning. [I wisely stayed on mute.]
“Look buddy,” I said, reasonably, “I am on a call for work, and I know you can eat cereal all by yourself. Can you try, and be quiet so I can hear?”
His response? “NOOOOOO!!!!! I CAAANNNNNN’T!”
Ugh-of course.
I tried to walk away for just a second so I could at least say hello, but found I had a screaming toddler barnacle attached to my leg. Meanwhile, the host of the meeting was saying, “I see someone else on the call. Andrea, is that you? Andrea? Andrea?”
And like any disaster, there are management practices that can mitigate the challenges. Since that first day, I haven’t had any other full-blown crises. Here’s what has helped:
Set a consistent schedule. My husband has a demanding job working with many clients and suppliers in Asia & California, so he’s often up late. But, luckily, business doesn’t really get rolling until 10 or 11 AM. After that first challenging meeting, we set up a schedule. I start work at 6 AM. We tag team from 11 – 2, and then he gets the late shift. Getting up at 6 AM is no fun (I am decidedly not a morning person), but it’s working.
Try for no (or limited) screen time in the mornings. Of course this is easier said than done, but I do feel this is helpful for (1) my guilt level in letting your child watch too much TV/iPad (2) giving him something to look forward to in the afternoon and (3) setting up some general rules. This has helped him learn to play by himself better, as well, which has been an unintended bonus. My (4) is that I only let him watch iPad after nap, which means that he has an incentive to take a nap (always a battle in this household).
Work in a room with the door closed. I’m lucky enough to have an office, even though the door doesn’t lock. The physical separation is good for many reasons, including that it curbs the (otherwise constant) question, “Are you done working?” You may not have an office, but there’s probably somewhere you can get away. I know of a popular podcaster who records all of her podcasts in the closet. You’ve gotta do what you’ve gotta do.
Undershare. Strategic use of the mute button is extremely helpful if you’re trying to make lunch, direct an art project, or change a diaper and be on a conference call. Many of our meetings are now on Zoom or another video conferencing platform. You can turn off your camera, too, even if everyone else is sharing theirs. I recently watched a savvy mom who showed herself quickly on camera, then turned off the camera for the rest of the meeting. I think this made her look very professional, while minimizing her stress level and visual distraction for the other attendees.
Overshare. Let your boss or clients know that you have children in the household and ask them to help you during this time. I feel that most people are very understanding. My Director of Evaluation, Lindsay, has TWO littles to take care of. We try to schedule meetings before the kids wake up or during nap time.
Be flexible. Sometimes, nothing else is going to work but pulling out the iPad or altering your schedule for the day to take care of your kiddo(s).Realize that this is not forever! You can knock off of work early to work on a project with your kids or even skip all or part of a meeting if the day isn’t going well. This time, too, shall pass — and — as my mother-in-law likes to remind me – “you’ll miss it.” While it’s hard to imagine missing the awful impacts COVID-19 is having on the world and our daily lives, there might be elements that are good in all of this. For our kids, who often get shuffled from bed to school to activities to dinner to bed, this might even have the opportunity to be a pleasant memory. Maybe.
The New York Times is one of the best places to look for good data visualization design. Some of their visualizations are truly mind blowing and thought-provoking, like this text analysis of presidential candidate song preference during their rallies (back when rallies were still allowed, and there were many more presidential candiates).
But sometimes, they miss the mark a bit. Take the graph below. The headline reads “U.S. Jobless Claims Soared to a Stunning 6.6 Million Last Week.”
Graph courtesy of The New York Times, April 2, 2020
This graph isn’t terrible, but it’s not great either. Three things I dislike:
Grey bar in the middle – I get that it’s supposed to indicate when the recession took place, but there’s something that feels really off about it.
Lots of irrelevant data in the middle – again, it’s there to show the general trend over time, but there’s just too much of it.
The relevant data is squished WAY over to the side. This is the main issue — we’ve got so much data, that the real story gets lost. And the orange color isn’t helping, either.
COVID-19 Job Losses 10x Greater Than the 2008 Recession
How could we better display these data? Let’s try a hybrid graph.
Source: The New York Times, April 2020
Just to keep things simple, I used the left side of the initial NYT graph, and added on to the right side. Instead of just having one skinny line to represent the jobless claims of today, I added a large bar to show the new data. The large, red bar draws the eye to the important information, and the change in style also makes it clear that there’s been some sort of break here.
Then I added arrows, descriptions, and labels. The graph still isn’t perfect, but I think it’s better, and, importantly, it tells the story more clearly. And it wasn’t hard to do. With my reuse of the left side of the graph, and a very simple bar, this makeover in EXCEL and Word took less than 5 minutes.
Making the graph above, I realized that there was another big story buried here. I looked up some jobs data, and found that, in Feburary 2020, there were 129.73 million full-time employees in the United States. I realized that the job losses from last week were huge considering that fact, so I thought about how to best visualize it. Here’s what I came up with:
Many Americans Have Lost Their Jobs In the Past Two Weeks
Source: The New York Times, April 2, 2020
I made this graph using a super-fancy proprietary software that I will be happy to sell to you for $199 per month.
Just kidding, y’all – I made this in PowerPoint.
Far from being sophisticated, these are just icons that I copy-pasted onto the screen a few dozen times. Add a couple text boxes, and it’s a pretty nice looking graph. Definitely better than a plain bar chart. Importantly, both graphs use the same information from the New York Times article, but tell distinct, and sobering stories.
A few weeks ago, we talked about one of the big
controversies in data visualization – the pie chart. When used properly, pie
charts can be a great way to display certain types of information. Sometimes
it’s hard to know what those rules are, and which graph should be used where.
Are there guidelines on the best ways to present data? The
answer is yes – there are many great
tools, like Stephanie Evergreen’s chart
chooser, that we often use to help me make these kinds of decisions.
One rule of thumb we use in our data visualizations is to only
use line graphs when displaying time-related data.
Here’s an example based on confirmed Covid-19 cases (as of 4/1/20):
Confirmed cases for selected countries
Horizontal axis shows the number of days since the case count exceeded 500 in each country
These data are perfect for a line graph. Readers naturally start
reading the graph from left to right; without reading anything else we can
quickly understand:
the number of cases are increasing over time.
some lines have a steeper slope compared to
others,
some countries have had confirmed cases of
Covid-19 for a longer period of time than other countries.
We can visualize in our mind
projections of each line (unfortunately for us here in the United States, that
line seems to be trending straight upward, but who knows! Maybe you are reading
this blog in the future and are saying, “What are they so worried about?!? We
found a cure, and everyone was fine!” Let’s hope so).
What would the data look like if it were presented differently? Let’s look at the same data in column chart format (updated 4/1/20).
Source: Johns Hopkins University CSSE data map Note: Data in this graph were pulled on 4/1/20. Some numbers are estimates, but numbers from days 15 and 20 were accurate as of the date and time in which they were pulled.
This graph just doesn’t tell the same story. We don’t get the same sense of time, or quite frankly, urgency. We cannot see how some countries flattened the curve compared to other countries. We cannot see where the U.S. is in terms of our trajectory compared to China and Italy. Sure, this graph is informative, and we do see that South Korea has far fewer cases over time, compared to other countries, but the power in that story is gone. Having the lines, slopes, and trajectories for each country in one graph together tells a much more compelling story. This is exactly why line graphs, or time series graphs, are so powerful. The time series graph created by Johns Hopkins CSSE contains a lot of valuable information all in one place (if you like data, I highly recommend taking a look at their dashboard). You can look at the graph and understand the story at first glance, without having to do much reading.
Now let’s walk through an example from a report we were working on with one of our clients. In this example, a line graph did not make sense with the data we had.
When I looked at this figure, as with the Johns Hopkins
figure above, I started reading the data from left to right, and assumed the
data progressed over time. Instead, the data were based on categories of
information (e.g., scores on different components of the ACT) and displayed
scores for treatment and control group students in each category. It was
difficult to compare the two groups within each category. I was spending too
much time thinking about the data within the figure, and if I was thinking too
much, our clients certainly would be as well.
I talked to Andrea about the graph (I’m calling her out – it
was her graph, originally) – and she said that she had considered a column
chart, but it just wasn’t having the same impact as a line chart.
We needed a different visualization.
I reworked the data using one of my favorite graphs I learned from Stephanie Evergreen – a lollipop chart!
Now, the reader can look at the data and quickly compare the two groups (i.e., treatment and control) within each category. It is easy to see that the treatment students outperformed matched control group peers in each subject.
Want to know how to make a lollipop chart? Check out Stephanie Evergreen’s website where she walks you through an example. This is the type of skill we include in our telling stories with data training we are excited to offer. This training is available online so if you are looking to hone your skills while quarantined at home, let us know!
Looking for something to take your mind off of the corona virus? Wanting to brush up your skills in presenting data? Our office is going to strive to offer you fun blog posts to increase your knowledge and skills in data visualizations, data analysis, and SEL evaluation techniques. Read below for our current tip!
Ever been faced with creating a PowerPoint but felt stymied by fear (How do I get started? How do I make something look cool? What are my key points?)? Or do you just have writers block when it comes to presentations in general? I get it! I went to an Edward Tufte course wherein he spent a good chunk of time hammering home the message that PowerPoints are terrible and have dumbed us down! However, we live in a world of PowerPoint, so how can we make them work?
His advice, and my advice to you: keep them simple. Make the meat of your slides limited to things like data visualizations, images, movies, and icons. Limit the words on each slide (his recommendation is to use NO words).
This is all well and good, but sometimes, it is hard to even
get started or know what to do.
A few weeks ago, my colleague Andrea contacted me to help save a PowerPoint presentation other colleague worked on and would be presented in just a few hours. She said it needed some help to make it pop, to be trimmed down, icons and other data visualizations. I had just two hours to help her out. I held my breath and waited to hear my email ding announcing the arrival of the presentation.
Ding!
I opened it up and read… and read, and read, and read.
Slides were plain or had walls and walls of text.
The default powerpoint title slide doesn’t make your audience excited to hear your content.
Look at this slide.
Walls o’text are NOT a best practice.
First off, how do you feel when you see this? I felt a
tightness in my chest. How was I going to fix this??? How would you fix this?
Can you read anything? What are the key take-aways?
I felt the fear and writers block kick in. I took a breath, checked the time, and got to work!
My first step, something I learned from my graduate school advisor, was to start at the beginning and make the structure of the presentation look professional. I quickly found a photo relevant to the study and made it the title slide. I found logos for each school highlighted in the presentation. I even used the Design Ideas feature in PowerPoint. I added our business logos. I started to feel better. It felt official. It felt real. It felt comforting. I could do this!
Transforming the title slide by adding an image professionalizes your presentation right from the start
I then started working through the rest of the presentation. I used the Design Ideas feature when it helped tell the story. What were the key points? If I was listening to this presentation, what did I need to know? In a way, it was helpful that I had not written the first draft because I was able to read through the existing slides, find the story, and trim it down. There was a pattern. I could do this.
I found icons available through PowerPoint and the nounproject.com (check them out if you haven’t done so already! They have an icon for just about everything!). I created a template and used the same icon for similar themes so participants could anticipate information. Remembering what I learned in a training provided by Stephanie Evergreen while I worked for the Austin Independent School District’s Department of Research and Evaluation, I made sure to add white space. We are visual learners, and adding white space helps us see the main points more clearly. I cut, and then cut some more.
Adding color and breaking up the text was extremely helpful
I looked at the clock.
I chugged away.
That wall o’text is now an attractive list.
I couldn’t get to everything I wanted to, but I distilled the information, made a template for how to present the information (e.g., consistent colors, layout, font), added informative icons to trigger what information was coming, and made more white space to draw readers in. I followed advice from Tufte and Evergreen as best I could and sent my slides to Andrea.
Within seconds I got a phone call, “ This is AMAZING!!!”
Andrea then shared slides with our other colleague who said, “OH MY GOD, WHERE DID THIS GIRL COME FROM?!?! WOW. ”
So, my friends, you can do this too, I know you can! Remember these general rules, and they will take you far:
Powerpoint Best Practices
1) Keep your slides simple – less is more! No paragraphs if possible (put text in the notes if needed)
2) Use consistent branding and color schemes, and don’t use PowerPoint’s default of blue/orange if possible
3) Remember that PowerPoint is a visual medium. Increase the use of images, icons, graphs, and other data visualizations.
4) Decrease the number of words on each slide (add white space!). Some of the slides above could have been further augmented by moving each point to its own slide.
As a golden rule, TELL your audience key points with your voice, and SHOW them key take-aways in the slides; the two go hand-in-hand.
You don’t need to search very hard on the web to find people that cannot stand pie charts. In fact, one of the first posts I found when searching for the topic was The Five Stages of Grief Over The Death of Pie Charts.
(Great read, by the way.)
But I think, as with many hated data visualizations, pie charts have their place, and today I’d like to share some things I’ve learned using Covid-19 as a timely example. (If you’re reading this way in the future, Covid-19 or Coronavirus was a nasty lung-infecting virus that swept the world in 2020, devastating countries and entire economies.)
I’m going to give a few examples of best practices for making pie charts useful. Much of what I’ve learned came from Ann Emery’s amazing course, Great Graphs. At publication time, Great Graphs wasn’t open, but I highly recommend following Ann and trying out her other courses. She’s at Depict Data Studio.
Do’s and Don’ts for Pie Charts
Don’t: Use too many categories.
Top 10 Countries with Most Confirmed Cases as of March 8, 2020
This is sourced from a real pie chart, and I’m assuming the author is normally a great data visualizer who is too stressed and overwhelmed to use best practices at the moment.
This one is a rainbow of awful. What’s wrong with this chart? Several things, but the most important is that it has far too many categories. There are 11 in total, and all but the top 5 are almost impossible to see. Go ahead, try to locate Switzerland.
I’ll wait.
This chart forces you to use the legend to understand what’s going on, and in the end, the legend becomes the only thing that actually gives you relevant information. So what’s the point of a pie chart here? I would recommend converting this instead to a bar chart if you want all of the information. But first think – do my readers need all this information? Why am I presenting data about 10 countries instead of 7, 11, or 200?
If you can lower the number of categories, I give you full permission to If use a pie chart.
Do: Keep the number of categories to 5 or fewer
Top 10 Countries with Most Confirmed Cases as of March 8, 2020
This pie chart only has 5 categories and gives the message much better — Mainland China has the vast majority of cases, and Italy, South Korea, Iran and all other countries split the remaining 26% almost equally.
This is the same data, but with the number of categories kept to 5. Now we can actually interpret the pie chart itself, without relying on the legend.
Honestly, 3 or 4 categories is ideal. Keep in mind that if some of your categories are really tiny, unless that is a fundamental part of your message, it is likely better to use another type of visualization.
Do: Include the percentages and labels inside or next to your pie chart
Second, include your percentages and your data labels IN the pie chart. Ditch the legend. A good rule of thumb: if, in the end, you need a legend to interpret your pie chart, consider using another form of data visualization.
Do: Let the title tell the story
This is a point I try to make with every graph, and it’s so important. When you make a data visualization, you are telling your reader a (hopefully compelling) visual story. You are editing what you show to make a point. Why not explain the point so that readers don’t have to figure it out (or misinterpret the findings?)
The Vast Majority of COVID-19 Cases in the US are from an Unknown Origin
Source: CDC, March 18, 2020
The pie chart above is pretty clear on its story, but having the title spells it out for the reader. Sometimes the mere act of typing the title will also help you to make your graph more clear to make sure the message is consistent.
Do: Get creative
I used an exploding pie chart in the above visualization. Data visualization experts will tell you that you should never use exploding pie charts, but don’t listen to those spoilsports. I used one here because I tend to think of viruses and the immune system as a Pac Man-type system. I think it works. Anything you can do to keep your reader interested is always a good idea.
(By the way, it is absolutely maddening that as of time of publication, it’s almost impossible to get tested for the virus in Central Texas and other places in the U.S. unless you have been to a place with a high rate of outbreak or have had close contact with a confirmed case. As you can see, this practice makes no sense whatsoever.)
In conclusion: don’t be afraid of pie charts, but don’t make terrible pie charts either. With just a few easy rules of thumb, you can make great ones. Want to learn how? Contact me to enroll in my 2-hour in-person class, Telling Stories With Data , where I walk you through the principles of good graph design in EXCEL. Or take Ann Emery’s course when it’s available. It’s fabulous.
Adding icons to your reports not only looks better, but actually helps your readers retain information. I learned about this easy fix from Ann Emery’s amazing class “Great Graphs”, which I highly recommend (more at the end of the post).
Sometimes ‘serious’ evaluators or researchers brush off these techniques as being too warm and fuzzy, and feel it is more professional to present everything like a journal article. I feel that these techniques can actually make your presentations look MORE professional, in addition to ensuring that people actually read your report.
Imagine a quick summary box like this one. Adding text boxes like this one can help break up your reports and draw the reader to the important conclusions. Still, as is, it’s a wall o’text:
First thing: break this into bullet points. Bullets are much easier to read and give the user immediate information — you can quickly see there are three major points.
Now, we’ll add some icons. But first, we need to get rid of those bullet points we just so carefully added. We’ll keep the spacing the same, but make the bullet points invisible, so they won’t interfere with the graphics.
I’m assuming you’re writing this report in Microsoft Word—so here’s what to do:
Highlight your bullets, right click, and go to “Bullets and Numbering”.
Click “Customize…”
Click “Font…”
Under “Effects”, click “Hidden”
Now your box should look like this:
Let’s replace that empty space with icons!
Go to Insert->Icons
Scroll through the icons or search for something specific (see examples below!)
The icon dialogue box in Microsoft Word (Mac)
Choose 3 icons that go with the above message and insert.
Choose each icon, right click, and go to “Wrap Text”; choose either “Square” or “In Front of Text” – this will help keep your formatting in line.
Now, move to the text, and voila!
The icons really make this section pop, and it looks more professional, and less like this report was made in MS word..
Check out Ann at Depict Data Studio. At time of publication, her Great Graphs course was not open, but she’s got other options.
