STUMP » Articles » Mortality with Meep: Guessing at Trends » 1 July 2019, 12:24

Someone on facebook linked to the Storyline site, and I’ve really enjoyed it.

The concept is very simple — there’s a time series of some sort, based on historical data — generally, there are points for every decade, or maybe every five years. You are given one point on the time series, and you have to draw the rest in. You can do a start/end point and then grab individual points and move them around.

I tried a couple so far — I’ve generally done okay on them, but… (well, you’ll see).

My main complaint is that I want to see a link to the underlying data… in a few cases, I’m a bit suspicious about data sources and/or definitions.

The other thing is how they “spline” points together… looks like it may be cubic splines, and that doesn’t work well for certain data patterns and makes people think there’s more variability than there is.

Here’s the death quiz. I actually hadn’t done this one before … you’re going to have to believe me that I’m doing screenshots from a real attempt.

Well, you’ll see I definitely didn’t do this one before.

First question:

So here you see my guess. The point in 1985 was given to me, and this question isn’t exactly fair… (I’ll say why in a moment)

My result:

I forgot to click on the question mark, but it’s a hint as to a driver for rate changes: “In 1963, the Catholic Church lifted the ban on cremation.”

Now, the reason that this question wasn’t fair for me to answer was not that I knew that factoid (I didn’t know when the ban had been lifted, but I did know there was no cremation ban in the Church right now.)

It was because I actually had to look up cremation stats for one of my research reports at my job. I had just been revising that report this morning. I knew what current stats were like, and I knew how cremation had taken off.

For the remainder, I will give the result page, plus the “hint”. You can ask for the hint before you guess, fwiw. I didn’t… and it shows for one of the graphs.

Question 2:

I used to live in NYC, so I had an idea of when murders peaked.

Question 3:

I had a good idea about what happened post-USSR, as well as some Putin-stuff.

Question 4:

I knew the general trend, but not the specific numbers here.

Question 5:

Yeaaaah, I sucked on that one. I really should have thought through that one more.

Final result:

Not that great, but I’ll take it.

WHY DO THIS?

So here’s the point on this one — there really is a good reason for you to try this out, and, more importantly, ask people to try something like this if you have a data story to tell.

If you quiz before you learn a fact, you actually remember the fact better.

I can’t find the specific research I’m looking for, but here’s something related: NYTimes: To Really Learn, Quit Studying and Take a Test and another: Studying With Quizzes Helps Make Sure the Material Sticks

So that’s about studying after you’ve been taught something, and you want it to stick in your memory.

But it also works if you ask questions before — you can get assumptions out of the way, etc. It’s about focusing people’s attention and getting them to think about what you want them to think about. This is more concrete than asking a question — just have people make a guess.

And here’s the thing: you’re not going to notice so much when you got the pattern right. I already knew about some of these trends, in terms of direction… the specific numbers? Not so much.

But that AIDS deaths pattern? Yeah, I screwed up big time there, and now I really notice how I was wrong… and I’m paying attention to the graph.

And that’s one of the issues with putting out graphs. You want the person’s attention. Most graphs – like these – are actually pretty boring. It’s just a line indicating something changing over time. Visually, it’s not all that compelling (thus the old USA Today scheme to get people to pay attention by putting on irrelevant graphics).

But if you make people guess where the line is going to be before you show them the line… they’re thinking about your topic, and they will really notice where the trend diverges from what they expected.

Try out Storyline! (I see where one can make one’s own…. I may very well do that.)