Thread Closed 
Week 2: Nations' Ranks from 1960 to 2004 in Infant Mortality Rates
09-03-2014, 07:17 PM
Post: #1
Week 2: Nations' Ranks from 1960 to 2004 in Infant Mortality Rates
[Image: infant_stats.html]

Infographic source:http://www.nytimes.com/imagepages/2009/0...stats.html

This hierarchical table graph displays a comparison of infant death rate rank between nations from 1960 to 2004. One can quickly see for countries listed whether the rank went to a higher infant death rate or a lower infant death rate by following the country from the left side (1960) to the right (2004) & any change in vertical location. The focus of the United States (this is from the New York Times) also can not be missed.

Visual encoding includes the nation’s moves from 1960 to 2004 either to a lower infant death rate (towards the top) with a darker colored line to connect the nation’s 1960 to 2004 ranks versus moves to a higher infant death rate (towards the bottom) with a lighter colored line to connect the nation’s 1960 to 2004 ranks. The lines utilize a Gestalt principle of connection as far as tying in the nation’s ranks from both sides of the chart. Another visual encoding is the fact that not only is the United States in bold, the connection line from 1960 to 2004 is also a completely different, darker color than any other connection line which is a popout channel.

Pros for the graph are that there is a very high data density but not to the point of distraction. This leads to a very high data to ink ratio where there is virtually no chart junk present & no visual embellishments at all keeping the information clean & concise. Data variation is clearly represented between 1960 & 2004 as far as the nations’ rankings. Even ties are represented by simple grouping at the connection line endpoint (see Netherlands, Germany, Greece, & Italy in 2004). After discovering that towards the top means lower infant deaths rates versus the bottom with higher infant deaths rates, the graph is a simple & clear representation of the change in ranks (if any) between 1960 & 2004 for the nations listed.

As far as a con for the visualization, the labels, lower/higher infant death rate at the top & bottom, are slightly distracting for the actual placement since at first glance it appears to be actually connected to those rank lines towards the top & bottom for those countries. another issue is scale distortion as far as what the rates & ranks actually mean as far as a quantifiable amount. There is no indication between one rank to the next what the actual delta is & what it is compared to other nations. Though not a large of a con, not all nations are listed. There is no way to tell if the countries present are the worst or best for infant mortality rates & why they are even included in the first place.

Suggested improvements are adding a level of interactivity such as roll-over effects to more easily connect the nation hovered over form its 1960 rank to its 2004 rank. In other words, the connection line between the years could use a popout channel effect as well as the nation in question. The suggested rollover interaction could also include actual metrics as far as the number of infant deaths for that year (depending if you are on the 1960 or 2004 side) & what the actual numeric ranking is. Though not critical, all nations that existed between 1960 & 2004 could also be present after clicking a “Show All” link or button.

* graphic is also attached


Attached File(s) Image(s)
   
Find all posts by this user
09-04-2014, 05:37 PM
Post: #2
RE: Week 2: Nations' Ranks from 1960 to 2004 in Infant Mortality Rates
I agree that the labels to identify which end of the scale is higher mortality and which is lower mortality looks like it belongs to the lines under them. This information could have been moved to the top, right under "Lines show change in rank". You mentioned that no quantity data is included. I think at a minimum the ranges should have been included. With just the rank we don't know if the best got better or worse. We know how things changed relative to other counties, but we don't know how any one country did. For all we know, every country on the chart may have had much higher death rates in 2004 than in 1960. For example, did the US drop in the ranking because our death rate went up, or because lots of other countries had their death rates go way down?
Organizing this chart on a vertical scale would improve the data content by using white-space to encode data. The distance between two countries would represent the difference in death rates. This would use up more space, which would decrease the data density, but the data/ink ratio would go up, because the white-space would also contain data.
Find all posts by this user
09-05-2014, 02:42 PM
Post: #3
RE: Week 2: Nations' Ranks from 1960 to 2004 in Infant Mortality Rates
I think the designer want to make this graph as simple as possible, therefore some important information ( at least I'm curisous of) was not included in this graph.
First, the ranking number. I would like to see a number by the countries so that I don't need to count by myself if I want to know the exact ranking.
Second, the "Lines show change in rank" comment is not that necessary. At the first look at this graph I can tell the lines show how the rank changed. If we get rid of this it will help make the figure clearer. Also, lower infant death rank and higher infant death rank can be present as arrows as well. Since we already got so many letters as names of countries, less extra letters will make the graph neat.
Third, even though the red color for U.S is outstanding, but the grey and grey green are too similar and hard to tell. Use smoke yellow may solve this.
Find all posts by this user
09-05-2014, 06:28 PM
Post: #4
RE: Week 2: Nations' Ranks from 1960 to 2004 in Infant Mortality Rates
I agree that this visualization can be misleading. Just based off of the visualization you have a feeling that the United States has had a drastic increase in Infant Death Rate; however, this is probably not the case. If we had additional data like that actual statistic for the death rate we could make better sense of the visualization. I am sure the death rate for the United States has actually decreased since 1960 just at less of a rate than other countries that have moved up the list. Did anyone else notice that Russia is all alone on the right hand side? They could have added it to the list on the left hand side somewhere, even if it is at the bottom. If they are implying that they cannot show the data for Russia on the left hand side due to the fact it was the U.S.S.R. why not just leave it off entirely for the sake of continuity of data points.
Find all posts by this user
09-06-2014, 02:09 PM
Post: #5
RE: Week 2: Nations' Ranks from 1960 to 2004 in Infant Mortality Rates
Quote:Pros for the graph are that there is a very high data density but not to the point of distraction.
I'll have to disagree with you on this observation, as there is a lot of clustering towards the center of the graph. Similarly, since a lot of the lines are practically the same color this can lead to some confusion as ones eyes tend to be stray when following a particular line.

Quote:This leads to a very high data to ink ratio...
Again, I am going to have to disagree with you on this one. There are much better ways that you could present this graph that would lead to a higher data to ink ratio. For example, you could turn this into an interactive graph where countries could be selected/deselected at will, reducing the clustering problem. (Also, the scale size would need to be fixed too.)

Quote:..another issue is scale distortion as far as what the rates & ranks actually mean...
I agree with you on this matter, there is definitely no sense of scale for this graph, aside from being able to tell whether the mortality rate went up or down.

Quote:Suggested improvements are adding a level of interactivity such as roll-over effects to more easily connect the nation hovered over form its 1960 rank to its 2004 rank. In other words, the connection line between the years could use a popout channel effect as well as the nation in question. The suggested rollover interaction could also include actual metrics as far as the number of infant deaths for that year (depending if you are on the 1960 or 2004 side) & what the actual numeric ranking is. Though not critical, all nations that existed between 1960 & 2004 could also be present after clicking a “Show All” link or button.
I agree with all of these suggestions for improvement, as they would do much to improve the charts effectiveness.
Find all posts by this user
09-07-2014, 12:06 PM
Post: #6
RE: Week 2: Nations' Ranks from 1960 to 2004 in Infant Mortality Rates
The visualization presented is kind of cluttered and hence not easy to extract information pertaining to a single country. A better choice of colors would have improved the readability of the graph, i.e. It is harder make out different shades of grey representing lower and higher mortality rates. Instead they could have chosen two distinct colors or change thickness of lines or give dotted lines.

Another suggested improvement is to represent mortality rates as well below the 'ranking' tab so that we can infer the actual mortality rate of a country and not just it's ranking which will help understand the actual difference in mortality rates.
Find all posts by this user
09-07-2014, 07:08 PM
Post: #7
RE: Week 2: Nations' Ranks from 1960 to 2004 in Infant Mortality Rates
First of all, a great critique by Kevin.
Some of the points on which I agree with him are keeping the information clean & concise,simple grouping at the connection line endpoint.The other point I agree with him is that there is no actual data in the visualization and as a result we cannot judge whether the countries included in the visualization are best or worst for infant mortality rates.
Some of the points I disagree are that the high data density do create a distraction. It is quite impossible to follow the lines in the middle of the bipartite graph.Also ties are represented by simple grouping at the connection line endpoint(Netherlands, Germany, Greece, & Italy in 2004). The factors that lead to this tie, is not very clear from the visualization.
Some of the changes I think can improve this visualization are:
1.The colors used for encoding can be made a bit more dark and separable.
2.There should be a rollover included in the graph to highlight the line from 1960 from 2004( Kevin has mentioned this in his critique)
3.The original data should be portrayed.
Find all posts by this user
09-07-2014, 07:08 PM
Post: #8
RE: Week 2: Nations' Ranks from 1960 to 2004 in Infant Mortality Rates
For the most part, I agree with the critique.
The graph seems easy to understand and the inking of lines based on changes (rise or drop) in the country between the years was a good idea. The hierarchical arrangement in the graph makes it easier to determine which country is doing well. In other words, comparing countries and even the same country at different years is much easier.

I, unlike u0909249, did not find the labeling in between the lines distracting. The connections however, seemed quite clustered at the center given the number of lines.

I also agree that the graph has a lot of scaling issues. I would prefer it if the graph could also indicate the percentage by which the rates have changed. For example: Sweden and Ireland might be at different levels of rank but the percentage by which the death rates have reduced in both might be the same. That is something what would be interesting to analyse. However, there is no data to in the graph to help with that analysis.

As an improvement, I would have also liked for the ranking to be numeric in order to have a precise understanding of ranks of countries rather than a vague one. A key indicating the meaning of each color used for the connecting lines would also be a good addition.
Find all posts by this user
09-07-2014, 08:03 PM
Post: #9
RE: Week 2: Nations' Ranks from 1960 to 2004 in Infant Mortality Rates
Firstly, kudos to a good yet simple design by the designer and a great criticism by Kevin.

I agree on points regarding the concise and somewhat clear representation of the change in ranks of the enlisted nations between 1960 and 2004. I also agree on the fact that the design does not provide any insight of the actual numbers of infant deaths of the countries, and as a result it is very hard to compare them based on the connecting lines.

On the other hand, I disagree on the point that the visualization is not distracting. At first glance, a viewer would not clearly picture the data as presented by the designer. The connecting lines, specially in the middle section, are very cluttered and its quite hard to follow their trail.

Following are some things which I think could improve the visualization
- Use of darker colors, like the one used to highlight United States, for encoding.
- Some degree of interactivity, like roll-over effects, would make the visualization much more engaging. The ties could be changed accordingly to only show the countries being hovered over, rather than the whole list.
- A range of numbers could be included showing the minimum and maximum infant death rates, which will give us the relative perspective between the countries.
- Another point I feel could make this visualization more informative is to provide a grouping of the countries based on region. For example, increasing death rates in countries with close proximity, like Italy, Germany, and Greece, could have been due to some environmental hazards.
Find all posts by this user
09-07-2014, 08:54 PM (This post was last modified: 09-07-2014 08:54 PM by soumyasmruti.)
Post: #10
RE: Week 2: Nations' Ranks from 1960 to 2004 in Infant Mortality Rates
I personally think Kevin did an excellent job in evaluating the visualization.

I agree to most of his critiques such as,
(09-03-2014 07:17 PM)u0909249 Wrote:  Pros for the graph are that there is a very high data density but not to the point of distraction. This leads to a very high data to ink ratio where there is virtually no chart junk present & no visual embellishments at all keeping the information clean & concise.
.

If we take a close look at the data, keeping an eye on starting points and mapping till the ending points, the picture(mapping lines) is clear and the labeling of the country names also helps. I don't think there is any amount of ambiguity present here.

The labeling of the lower infant death rates in between the mapping lines seems quite distracting but it becomes more clear when we take a look at the lower label of higher infant death rates.

Suggested improvements to this visualization would be that the representation of actual data should have been given. Even if the rank of a country has decreased over the years there's a clear possibility of percentage decrease in death rate as a part of increasing population. If most of the countries have done well with reducing the death rates then this visualization doesn't send a clear message by making an opinion just on the basis of ranking.
Find all posts by this user
Thread Closed 


Forum Jump:


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