From CS 294-10 Visualization Sp10

Jump to: navigation, search

I wanted to see what effects different presidencies had on the population, or whether it is safer to live under a democratic or republican president. I used the Human Mortality Database information for the USA. I downloaded the information in text format, imported it into Excel which did a great job seperating columns, I did some minimal cleaning for duplicates and number formatting, and then imported the information into Tableau.

I first graphed simple population against time, and colored it by age. I had a stacked polygon effect in mind that I could not accomplish, so instead I simulated it by widening stacked bar charts. You can make out the contours for age groups, widening as there are more, and shrinking as there are fewer people in that age group.

There's an interesting anomaly in 1945, and if I switch to a line chart you can see a very clear and sombering dip in 18-35 year olds during WWII.

The dip is even clearer when only viewing the male population, as there is no decline in female population during this time.

The trends from total population weren't helping me answer my question as much as I'd like, so I took the deaths data from the Mortality database and joined it with a table I created showing the party of the president every year. The graph that resulted showed, basically, a continuous increase in deaths regardless of party presidency (although there was a curious leveling off during the Nixon/Ford Vietnam era).

I assume this increase was proportional to the population increase, so I decided to normalize the deaths based on population. I joined all the tables I've worked with so far and graphed the deaths as percentage of population.

I was still curious as to the low death rate during times of war, so I checked some of the specifics of the data. As it turns out, the death data only take into account people who died on in the country, and after 1970, only count US residents (so non-residents who died in America don't count to the total). I decided that the information was still useful so I pressed on.

I decided to add another dimension to the data by graphing deaths of both men and women. A curious feature of this graph was that more men have died than women throughout American history (presumably) up until 1997, when women surpassed men in total deaths. I tried to break this down further by looking at male and female deaths now split up by age.

Looking at the right side of the graph, you see there are more circles (for women) on the top ridge of the graph. At the left, the top ridge (not the outlying stretch, which I will address soon) are mostly males. Furthermore, most of the males dying earlier on were in their 60s, and most of the females dying later were in their 80s. It would seem that after a surge in population at the start of the 20th century, the women lived longer than the men, which would help to explain the overtaking of men by women in death rates in 1997.

More interesting on this graph are the unexpectedly high infant death rates (deaths at age 0) in earlier years. It would seem 3 or 4 times more infants died around 1935 than the nearest age death group. I decided to investigate this further by removing the gender aspect of the graph, and just looking at total deaths by age. I highlight the infant deaths so they are easier to see, and I try to get closer to answering my question by shading the graph red during a Republican presidency and blue during a Democratic presidency.

While thankfully infant deaths decreased, there's no discernable pattern yet between deaths and party presidencies. Note: The infant death data was indeed surprising, so I verified its accuracy with academic and CIA World Factbook information.

I next tried to create heat maps, one showing deaths by the percent of the corresponding age population, and one showing absolute deaths by age.

The percentage heat map is not quite so interesting, since as you would expect, a large percentage of old people die while only a small percentage of young people die.

The absolute heat map also gives common sense information. Aside from the decrease infant mortality rate, which we have discussed, people used to die around their 70s, and now they die around their 80s. I have overlaid a trendline showing the general progression. While I am tempted to say that in this heat map it looks like more people die younger during Republican administrations, neither this nor any of the previous visualizations confirm this clearly, nor does a closer examination of the heat map.

So if a conclusion was made to the question of whether the party of the presidential administration has an effect on the mortal safety of Americans, it was that despite the doomsayers, it does not. The progress of medical science and law enforcement is obviously stronger than any political affiliation, and as the final visualization shows, an American in my age group (18-25) is less likely to die than pretty much any time in history.

This visualization graphs the percent of people of age for each gender that died every year. I restricted the ages to just 18-25, and they are drawn overlapping for each gender in the graph. This is the result of restricting my question of how politics affects my safety to: How likely am I to die compared to previous years? Since I found that political party had no correlation with deaths or safety, this became the pertinent question. The graph shows that the likelihood of a person my age dying this year is lower than it's ever been, especially if she's a woman. I also overlaid a trendline, showing the decline of death percentage.

[add comment]
Personal tools