quetzalc0atl
Registered: 10/07/09
Posts: 3
| |
bpierce
Moderator
Registered: 06/18/07
Posts: 31
|
|
| |
Reply with quote | #2 |
Hi Awais,
I think the best way to help people see where their neighborhood falls compared to other neighborhoods and the city average is to use a modified version of a box plot, as shown below. I constructed this graph in Excel using the "Open-High-Low-Close" version of a Stock Graph and I just made up the "Worst Neighborhood" scores for this example.
-Bryan
Attached Images:
|
| Loading... | | |
quetzalc0atl
Registered: 10/07/09
Posts: 3
|
|
| |
Reply with quote | #3 |
Hello Bryan,
Wow - talk about thinking outside the box. This visualisation certainly makes things a lot simpler and there would be very little need for explanatory notes. I will forward this to my supervisors.
Thank you very much for your help.
Awais
|
| Loading... | | |
b_rossi
Registered: 06/05/09
Posts: 2
|
|
| |
Reply with quote | #4 |
Hello Awais,
Bryan's direction certainly is a great direction to follow.
An item I noticed from your original graphs was how far you had taken the decimal on the X-axis. From my experience, unless there is an explicit reason to need this level of accuracy, you would be better off taking it to the whole number (100 vs. 100.00). When I originally read it, I mistook the decimal mark to be a comma and wondered if the comma was in the wrong place, or if you had neglected to add the final zero to make "100,000".
Also, from my understanding of your intended audience (cross-demographic general public), you may also benefit from modifying the scale to depict 10% increments -- I have found it is easier for people of all educational backgrounds to grasp values that fall within a given 10 percentage point spread, than to have to determine a percentage that falls into 25% spreads.
Update, and a question for the board:
I thought about this graph again and wanted to ask you about the choice of "100" to represent the worst neighborhood score. Does anyone have an opinion on the emotion effect of setting "0" as the standard for the worst scoring neighborhood? In my mind, I see a powerful competitive drive to raise scores, especially when they follow suit with how Americans are conditioned. For example, "I scored 80 out of 100, but June scored 90 out of 100, so I need to work harder next time".
Best of luck with your effort. |
| Loading... | | |
quetzalc0atl
Registered: 10/07/09
Posts: 3
|
|
| |
Reply with quote | #5 |
Hello b_rossi,
Thanks for your feedback. The decimal places on the graphs were an error on my part - I agree with you that 100 would read better than 100.00. I will also follow up the 10% increments.
I read somewhere on this board (can't remember exactly where) - Mr Few mentioned something along the lines that people generally tend to associate the the best or something that is good as having a higher score, or an upward trend of some sort would imply showing improvement. Some of my more experienced colleagues believe that the terminology that we've used isn't particularly great. People may not react well to find out that their area is described as the worst in the city for a particular theme. They've suggested going with terminology along the lines of Most Deprived and Least Deprived which is less damning - however again we come back to the issue of whether 100 should constitute the best or the worst.
I received some feedback from my colleagues and supervisors on the different graphs that could be used here. Part of what we are trying to do is demonstrate relationships between the different theme areas, so for example if levels of education are poor in a particular area this can suggest links with high levels of worklessness and crime.
As a result they believe that the "Radar Graph demonstrates links visually as well as showing differences in areas effectively. It can also be adapted to display all findings from one theme if necessary". Although they believe this second point to be similar for the straight bar graph.
They also believed that the table of figures was far too simplified and would not be able to demonstrate the relationships between the themes as easily. And their opinion of Bryan's graph was that it would be too complicated for the general public to interpret due to there being too many factors on one graph and the relationships between the themes are not clear.
Any further comments or suggestions would be most welcome.
Thanks,
Awais
|
| Loading... | | |
b_rossi
Registered: 06/05/09
Posts: 2
|
|
| |
Reply with quote | #6 |
The major problem I have with the radar graph, in this context, is that you aren't encoding values that depend on being displayed serially (such as "crimes committed, by hour of the day"). I like the goal of building a visual display that allows potentially related or causal measures to be associated, but I don't know if it is possible given the current amount of information.
I don't know if it is possible to answer questions like, "Is higher worklessness related to higher crime?" or "What happens if a given neighboorhood's worklessness is higher but as an average, the educational score is significantly higher than it's peers?". My fear would be that the radar graph could lead to a set of false assumptions or correlations that's can't be proven.
I'll keep thinking about how to display this, it has been a great mental exercise.
-Ben
|
| Loading... | | |
Derek_C
Registered: 08/15/06
Posts: 34
|
|
| |
Reply with quote | #7 |
For bivariate relationships among the variables, I'd try a scatterplot matrix.
As for colleagues and superiors having strong opinions, I experience similar things in my work. While I sympathise with the authority issues there, as far as colleagues and superiors actually being in the right (as opposed to just in charge), the problem is that everybody has opinions, but not everybody has informed opinions. In fact almost nobody in the lay community has informed opinions about Visual Intelligence: it's a field unusually rich in counterintuitive facts and "things everybody knows that ain't so". Which accounts for the popularity of pie and radar charts :-)
|
| Loading... | | |
