Thread Closed 
Week 3: Girls Lead in Science Exam, but Not in the United States
09-08-2014, 11:43 AM (This post was last modified: 09-08-2014 11:46 AM by aedunn.)
Post: #1
Week 3: Girls Lead in Science Exam, but Not in the United States
This visualization is a scatterplot showing the difference between boys’ and girls’ test scores on a particular math test by country. Color is used to distinguish between different regions, including Western/Northern Europe and the Americas, Asia and the Pacific Islands, and Eastern/Southern Europe and the Middle East.

There are two things I think the visualization does well. First, it encodes the three regions by different, easily-distinguishable colors. Second, I also like the interactivity -- mousing over the dots shows the country and the scores for both boys and girls. And going through the text box on the top-right of the graph highlights and labels different dots, and also provides some interesting insights on that subset of the data.

What the visualization does not do well is scales and labeling. The administered math test is not named or explained, so I’m not sure what the scale is. I assumed by the marks shown on the y-axis that the score range is 300-600, but I could be wrong. The percents on the x-axis are also not labeled or explained. I assume they are the difference between boys’ and girls’ scores, but this is not made explicit anywhere. Furthermore, the scale is a little misleading because it only spans 6% both ways. If this scale were extended (maybe to 100%?) then the gender differences would essentially disappear.

In order to improve this visualization, I would just add labels to the plot to explain the meaning of the x- and y-axis, and I would implement more honest scaling that doesn’t exaggerate the gender differences quite so much.

http://www.nytimes.com/interactive/2013/....html?_r=0


Attached File(s) Image(s)
   
Find all posts by this user
09-08-2014, 12:22 PM
Post: #2
RE: Week 3: Girls Lead in Science Exam, but Not in the United States
The axis is not clearly marked. The administered test generalizes the test as science test. It would be an interesting study to see in which stream of sciences girls outperformed boys.
A stacked bar chart could be used to better represent the gender ratios.
It does well in the interactive labeling when you hover over the graph and highlights the different regions by hiding the others.
Find all posts by this user
09-11-2014, 11:23 AM
Post: #3
RE: Week 3: Girls Lead in Science Exam, but Not in the United States
I think the x-axis represents the difference between the mean score for girls vs the mean score for boys, but why did they use a percent instead of a point value? Also, was the test given to equally sized groups? Did some countries have more representation from boys or girls? I like the extra information when you hover over a data-point, but I tend to be very suspicious of averages when the study methodology is not very clearly stated.
Find all posts by this user
09-14-2014, 02:08 PM
Post: #4
RE: Week 3: Girls Lead in Science Exam, but Not in the United States
I agree that this visualization does very well in interaction. You can use the right side text box to show the performance of different regions which boys do better or girls do better. Besides this, if you are interested in a specific dot, just put your mouse on the dot, detailed data will be shown. This design is very readable.
But one thing I want to point out is: though the design use three different color to distinguish different region, what the standard of this partition? I'm confuse with the blue color regions, for example. Western, northern Europe and the Americas all are use blue dot. However, these three regions have different performance in Exams. If you don't read every picture, you can not figure out the trend that girls lead in Northern Europe and boys lead in Western Europe. Relatively, boys and girls go halves in America region. Maybe using three different colors to represent these three regions is better.

(09-08-2014 11:43 AM)aedunn Wrote:  This visualization is a scatterplot showing the difference between boys’ and girls’ test scores on a particular math test by country. Color is used to distinguish between different regions, including Western/Northern Europe and the Americas, Asia and the Pacific Islands, and Eastern/Southern Europe and the Middle East.

There are two things I think the visualization does well. First, it encodes the three regions by different, easily-distinguishable colors. Second, I also like the interactivity -- mousing over the dots shows the country and the scores for both boys and girls. And going through the text box on the top-right of the graph highlights and labels different dots, and also provides some interesting insights on that subset of the data.

What the visualization does not do well is scales and labeling. The administered math test is not named or explained, so I’m not sure what the scale is. I assumed by the marks shown on the y-axis that the score range is 300-600, but I could be wrong. The percents on the x-axis are also not labeled or explained. I assume they are the difference between boys’ and girls’ scores, but this is not made explicit anywhere. Furthermore, the scale is a little misleading because it only spans 6% both ways. If this scale were extended (maybe to 100%?) then the gender differences would essentially disappear.

In order to improve this visualization, I would just add labels to the plot to explain the meaning of the x- and y-axis, and I would implement more honest scaling that doesn’t exaggerate the gender differences quite so much.

http://www.nytimes.com/interactive/2013/....html?_r=0
Find all posts by this user
09-14-2014, 08:35 PM
Post: #5
RE: Week 3: Girls Lead in Science Exam, but Not in the United States
I think this is a fantastic piece of visualization.
The visualization takes care of many details a user might have while exploring it.
I agree with aedunn on various points:
1. The colors used to indicate different regions do not conflict in any manner and are easily noticed one against the other.
2. The interactive hovering over dots show the exact scores of each gender which is quite helpful and can actually be used to answer the query aedunn put forward for percents on the x-axis. The x-axis percent do indicate the percentage difference between the gender scores. For example, in case of finland it is (562-546)/546~2.9%
3. The top right corner of the graph is very useful in investigating the girls vs. boys performance for each region of the world.

I also agree with aedunn that from this visualization we don't get any clear idea about the total score of the unnamed maths exam. But when I ponder further upon this, these details are more or less irrelevant for when our main goal is to compare gender performances. Nonetheless, this information could have been added.

I also feel the total no. of girls and boys appearing from each country for this maths exam could have been an useful data.
Find all posts by this user
09-14-2014, 09:32 PM
Post: #6
RE: Week 3: Girls Lead in Science Exam, but Not in the United States
There are some viewpoints I agree with aedunn.
First, The y-axis labeled as score is confusing and difficult to understand by reader. Second, the writer didn't show scale of this chart, it's hard to be convinced.
The title of this article is Girls Lead in Science Exam, but Not in the United States, however, this chart just reflect performance of girls and boys at age of 15 and also didn't provide the scale of data. So this article mislead readers on the methods and techniques.
Find all posts by this user
09-14-2014, 10:12 PM
Post: #7
RE: Week 3: Girls Lead in Science Exam, but Not in the United States
I agree with the x-axis being problematic. I wonder how many of these points fall under what we consider statistically significant. It seems that the majority of the data points fall right around the average. You would have to really break down the statistics of the data to see how many of these points can actually tell us something useful. It may be that the data points that are less than 1 percent on either side of the average cannot not really tell is anything. This would make much of this data irrelevant.
Find all posts by this user
09-14-2014, 10:18 PM
Post: #8
RE: Week 3: Girls Lead in Science Exam, but Not in the United States
The bubbles are split into three categories with highly saturated distinctive colors, which makes channels (color and position) easily discriminable and effective. Also, interactivity and comments on the top-right corner provide rich extra information which are good complements to the simple graph. Still it has space to further improve on.

- I partly agree with aedunn's point about the image's interactivity. Interactivity makes it good, but it's also the right place it needs further improvement. That, it could be better to make bubbles with interested color to pop out. For instance, if the mouse is on a purple bubble, all purple bubbles should pop out to the top layer. Clicking tabs on the top-right comments has similar function, but it's for another purpose. As you may see Australia is totally blocked by Estonia whatever you do. Further, I think there is enough white spaces to put a label for each bubble of interested color (e.g. name of country and test scores). Doing this could make the design more expressive.

- As for the X-Y axis, I don't agree with aedunn. The X-axis, on the left part, represents the percentage of boy's average score surpassing that of girls. So, granted the amount of difference would be become insignificant if the X-axis is measured up to 100%, it's unlikely to happen in practice. Also, the measure of Y-axis doesn't interfere with the reader's interpretation, since the topic is about difference rather than absolute value.
Find all posts by this user
09-15-2014, 12:19 PM (This post was last modified: 09-15-2014 12:22 PM by mcarter.)
Post: #9
RE: Week 3: Girls Lead in Science Exam, but Not in the United States
The intent of this graphic is to convey the relative performance of boys vs girls per country. My support for this assessment is the captions that go along with each of the 5 interactive frames almost exclusively discuss gender disparity in the score, as well as the fact that the article text only discusses gender disparity in scores.

To this end, I don't think that the graphic is particularly effective in doing so. Why muddle the point of the graphic with an extra dimension (namely, average score)? If you are trying to demonstrate the relative performance of boys vs girls per country, why confuse this with average score information?

Also, I think an important point that has been missed in this discussion so far is the consideration of audience with respect to the quantitative nature (or lack thereof) of the visualization. This isn't a graphic in a scientific journal. If it were, many of the points regarding the graph's axis, sample sizes, etc. raised so far in this discussion would be important. However, the audience of this graphic is the general population. The point of the graphic is to convey a general sense of the performance of boys vs girls in science exams for 15 year olds to a genera audience. To this end, a qualitative, relative comparison (as mentioned by t.li), I feel, is sufficient.

However, there is one huge gripe that I have with this visualization - the partitioning of the countries. The pink and yellow group make natural subsets based on geography, however, the blue set includes all of Northern and Western Europe plus all of the Americas. Why would you group these this way, when the others are grouped geographically? The only reason I could see is because these data points share similar properties. However, in the 5 interactive frame analysis, there is no discussion of the blue points as a whole. They are separated out into Northern Europe in frame two, and Western Europe plus Americas in frame four. If you are going to analyze these subsets of the blue group, why not break the blue dots into two distinct groups?

With all that said regarding the colorings, even if you did break the blue into two groups, the groupings are inconsistent... Some of the groups would be based on geographic similarities, while some of the groups would be based on statistical properties. This makes it difficult for the reader to determine what is being emphasized with the colored subsets.
Find all posts by this user
Thread Closed 


Forum Jump:


User(s) browsing this thread: 1 Guest(s)