For this weeks’ #MakeoverMonday Andy chose a visualisation originally published by online journal ‘The Pudding’. The viz was featured in an article entitled ‘The Timing of Baby Making’ which explored the theory that Hurricane Sandy (a substantial hurricane which hit the Northeast US in October 2012) resulted in so-called “Sandy babies” nine-months later.
The original viz is actually quite impressive (click on the image to explore the interactive version):
What I like about it
- Simple, clean design.
- Fantastic interactivity, including the ability to filter to display data which may be more personal to the end-user. There are some great annotations that appear to the right of the viz in the interactive version too.
- The events icons at the bottom are a really cool feature; particularly for somebody like me who wouldn’t necessarily know when major US sporting or weather events took place.
What I think could be improved
- The title is a little boring and gives no indication if Hurricane Sandy had any impact upon birth rates in Suffolk County.
- The colours used (blue/pink) have strong gender associations, especially in babies. Since this viz does not explore the gender of the babies born another colour palette may have been a better choice.
What I Did
Given the focus on a weather event in the original viz I initially thought it would be a good idea to attempt something similar. However, rather than hurricanes I wanted to explore snowstorms; my thinking being that these had the potential to trap people in their homes for longer periods of time. As the theory goes, when people are stuck in a confined space for a long period of time (especially if they are without power and therefore distractions such as TV, broadband, etc) they are more likely to resort to other forms of entertainment.
The Wild Goose Chase
After a quick search on Google I found an article which listed major weather occurences in the US in the last 20 years. I was drawn to the North American winter storm which brought up to 70 inches of lake-effect snow to the Central US and New England in November 2014. It was so severe remnants of the snow still remained in Buffalo, New York eight-months later. People were trapped in their homes, sporting events were cancelled and everything came to a standstill for over 7 days.
While the storm hit multiple states and counties, the city of Buffalo was one of the worst hit areas. Unfortunately the data set provided for #MakeoverMonday didn’t go any lower than county level. Nevertheless, I was able to focus on births in Erie County where Buffalo is located. Given the storm hit in mid-November 2014 I was hoping to see an unseasonal spike in births in August 2015; nine-months later. Unfortunately this wasnt evident from the data and births were no higher in August 2015 than they were in the previous year. I suspect this is because the data was at a county-level. If I had the data for births in Buffalo alone I suspect I would have seen a spike. Local news articles at the time made reference to a “Snowvember baby boom” so the storm must have had an impact on births to some extent but unfortunately I couldn’t show this in the data.
Rather than explore another weather disaster I decided to search Twitter for inspiration. I rarely do this as I prefer to come up with something original for #MakeoverMonday. It is very easy to be swayed by somebody elses design choices or their take on the data set, even if this is subconscious. For these reasons I try to avoid looking at other people’s work until I have finished my own if possible.
Searching through Twitter Daniel Caroli’s viz caught my eye:
Daniel used carefully shaded box plots effectively to highlight that mothers in the USA are getting older. I like how he did this using ALL of the data and therefore shows the range of mothers ages by year.
I wanted to create something similar that would also show the range of mothers ages but not by year but by state. I was curious to know if there were significant differences in the ages of mothers by state and which states had the youngest or oldest mothers. As Daniel had already shown that mothers were getting older I wasnt too concerned about showing this in my viz.
My Original Viz
After manipulating the data in Tableau I created the following viz:
I set out design something that was quite striking where it would be easy to identify which states had the youngest or oldest mothers. I played with the colour of the circles for some time before deciding to stick with white. I wanted to use a dark background to make these stand out as much as possible, especially considering there were some outliers in some states. I tried adding box plots but this made the viz look messy given the long list of states so I used BAN’s (‘Big Ass Numbers’) to indicate the min/median/max by state instead. I wasn’t keen on my final design and felt like it was ‘missing’ something but decided to stick with it regardless.
The Updated Title
After publishing the viz to Twitter, Andy contacted me and suggested I amend the title:
When I made the viz I thought it was clear that I was referring to the mother’s age (my thinking being that men don’t experience childbirth – well not last time I checked anyway). However, in hindsight I agree Andy made a valid point and perhaps the original title wasnt 100% clear. Andy got a few cheeky remarks on Twitter following his suggestion but this was all in good spirit of course! I updated the title as suggested and republished the viz:
The Viz Review
I asked Andy and Eva to review my viz during the weekly #MakeoverMonday ‘Viz Review’ webinar and hoped for some constructive feedback I could work upon. Although I have joined all of #MMVizReview webinars to date this was the first time I had submitted my viz for review (not because I was reluctant to receive feedback; I actually haven’t finished my submissions for those weeks yet). It’s been great to see people update their vizzes following the previous reviews and I hoped to do the same with mine.
During the review Andy and Eva made some valid points about my viz. While they liked the overall design their comments were:
- The background is too dark for the subject matter. Babies are associated with joy and happiness so why the dark background? *
- Annotations would help to tell a story.
- Clicking upon a state in the list triggered the viz to drill-down to county which caused the viz to look messy meaning the State/County hierarchy needed to be removed.
- It would be nice if the state filter could highlight or change the colour of the state selected from the list below (I had set-up the filter to update the BAN’s only).
*On a side note somebody on the webinar commented to say that perhaps I had coloured the viz black and white to look like an ultrasound picture. I wish I could say this was my inspiration but unfortunately it wasnt.
Following the webinar I returned to my viz and updated it based on the comments.
I was reluctant to amend the background colour as I liked the black. However, I took Eva’s comments on-board and decided to make the viz more cheerful by the inclusion of the purple. I struggled to pick a suitable colour for some time but I went with purple as I thought it worked well with the white against the dark background.
I mentioned during the viz review that I didn’t know how I could make the filter highlight a state as suggested. However, in the updated viz I created a parameter which would select the state in all three graphs so I think this works better.
Given the design of the original I had little space to add any annotations. Also, I was conscious if I included a state filter any annotations to the side of the viz could get lost or messy when the filter was applied. After much thought I decided to include a few additional charts to help tell the story with some commentary for each of them.
- Hex Map – I have wanted to create a hex map for a while but struggled last time I attempted one. Thankfully this time I had more success using Matt Chambers’ Hex Map tutorial (it seems this is the go-to tutorial for hex maps). I wanted to show the median age of mothers at a glance and thought this was a good use case. I like how you can clearly see regions on the map where mothers tend to be older or younger than average.
- Line Chart – I was initially going to stop after building the hex map but I was conscious this only included a static view of the median age and did not show changes in the median age over time. By including a line chart to show the difference in the median age versus 2003 I could show this whilst also indicating any states where the median age didn’t increase over time.
- Original chart – This remained largely unchanged. I added some commentary and also included some arrows to make it clearer what the chart was showing.
The Final Submission
Here is the updated viz. Click on the picture to view the interactive version in Tableau Public:
I’m much happier with the final viz. I really like how the colours turned out and I think the new version tells a more insightful story.
The Power of Feedback
I’m so grateful to Andy and Eva for their feedback, without which I probably wouldn’t have made the changes I did. This really goes to show the benefits of seeking feedback (and acting upon it) for our work. While you may think you have created something brilliant it always helps to show somebody else to get their opinion and see if they can offer any suggests for improvement. This doesn’t always need to be somebody who is data-savvy; I often show my vizzes to my Mum. While she doesnt know much about data viz she can tell me if something make sense or if anything is unclear. Of course, this process only works if you evaluate and work upon any feedback given. There’s no point seeking feedback if you are only going to ignore it.
The new #MakeoverMonday viz reviews are a great way of seeking feedback from people who know what they are talking about! Don’t be afraid to put your viz forward for review. The feedback can help you to identify flaws you may have missed or even suggest quick wins which could take you viz design to the next level. Over the last few weeks its been great to see people working with the feedback provided and improving their work.
Thanks for reading.
Update: I’m happy to report that on 18th August 2017 my visualisation was selected as the Tableau Public ‘Viz of the Day’ and later featured as ‘Viz of the Week’ on the Tableau Public website!