Project prep and discussion

Thoughts about visualization and storytelling

Visually attractive graphics also gather their power from content and interpretations beyond the immediate display of some numbers. The best graphics are about the useful and important, about life and death, about the universe. Beautiful graphics do not traffic with the trivial.

• Edward Tufte, The Visual Display of Quantitative Information, Chapter 9: Aesthetics and Technique in Data Graphical Design

Antipatterns in visualization

via Mushon Zer-Aviv, "Disinformation Visualization: How to lie with datavis"

Gallup: Pro-choice vs Pro-life

Riffing off of Mushon Zer-Aviv's critique of the various ways the Gallup abortion poll can be remixed visually to favor a certain political viewpoint.

Gallup's graph 1995 - 2010

The AP skew

(the AP article was originally published in 2009)

via Zer-Aviv:

Back in 2009 when the AP wanted to use the data to tell a good story, all they needed was to skew the graph’s proportions a bit, change the starting point of the vertical axis and switch to thicker, less nuanced lines that based on the very same data emphasizes a clear trend in the graph. And a clear shift in opinion in such a huge ideological debate makes for a much more interesting story.

The LiveAction.org skew

Now here's LiveAction's take on it

via Zer-Aviv:

Earlier in 2009 when pro-life site LiveAction.org wanted to graph the same data to preach to the choire and say “we’re winning!”…This time, no need to admit the public opinion was ever at the 1990’s low pro-life mark, so starting from 2003-2004 is fine. Focusing on a shoter time frame to tell a story of consistent increase helps too, as does the choice of 18-29 years old age brackets. There’s nothing about the data that isn’t true, but it is very selective.

Zooming out: Gallup 1995 - 2014

A much more muddled picture when you look at how opinion has changed from 2009 to 2014:

Adding nuance of which circumstances

Moving beyond pro-life versus pro-choice, let's look at the responses for "Do you think abortions should be legal under any circumstances, legal only under certain circumstances, or illegal in all circumstances?"

Variations in charting

Let's see what another level of granularity in data can show.

For respondents who said that abortions should be legal in certain circumstances, Gallup then asked them if they meant "in most circumstances" or "only in a few circumstances".

Part of the data is reproduced (by hand) below:

Date	Always	Mostly	Few	Never
5/8/2014	28	11	37	21
5/2/2013	26	13	38	20
5/3/2012	25	13	39	20
5/5/2011	27	10	39	22
5/3/2010	24	15	37	19
5/7/2009	22	15	37	23
5/8/2008	28	13	40	17
5/10/2007	26	15	40	18
5/8/2006	30	13	39	15
5/2/2005	23	12	40	22
5/2/2004	24	13	42	19
5/5/2003	23	15	42	19
5/6/2002	25	12	39	22
5/10/2001	26	15	41	15
1/13/2000	26	17	39	15
4/30/1999	27	12	42	16
1/16/1998	23	16	42	17
8/12/1997	22	12	48	15
7/25/1996	25	13	43	15
2/24/1995	32	9	41	15
9/6/1994	33	13	38	13

Stock area chart

When charting in Google Spreadsheets, the default chart choice is Area:

Area re-colored

The colors are randomly chosen, and in the default assignment, are pretty confusing: blue is for Always, but red is for Mostly.

So let's switch Always to blue, make Mostly a lighter blue, and Few and Never shades of red:

Area color, politicized

Now here's where judgment can apply color choices in such a way to skew the perception of the graph. Someone who sympathizes with the pro-choice side might argue that Few should not be considered a pro-life proponent, on the logic that you're either pro-life or you aren't. Such an advocate may then choose to make the Few area more neutral in appearance:

By making Few much less associated with Never, the situation makes the pro-choice contingent much more prevalent.

Line graphs

The area chart is useful for showing parts of a hole. But let's go with Gallup's choice and use a line graph.

Again, the default choices by Google are not helpful:

Line, re-colored

The problem with area charts (similar to pie charts) is that the eye has a harder time distinguishing differences between amounts of area. With the line chart, however, we see a new insight: the respondents who believe abortion should be highly restricted, i.e. Few circumstances, are by far the majority. And pro-lifers may claim that as a victory in the battle for opinion.

Line, compressed vertical axis

Changing the minimum of the axis to 5, and the maximum to 50, we see more dramatic changes in opinion, though that just serves to make the polling results seem more jittery:

Area, with less nuance

One more iteration: an area chart with the two in-between opinions combined into a Sometimes category, i.e. the same as Gallup's original graph, except as an area chart: