• About
• Archives
• Contact
• Forums
• Projects

21. 

    Tim
    Member
    Posted 1 year ago #

    this one is a bit out of left field. my goal was to really try to "visualise" the stats and put a face to them. i took the state with the highest rate amongst kids, then looked deeper for the highest rate within the state. I started with county level stats but ended up using school districts. I don't know if the faces I found were those of kids from a family below the poverty line or not, but something about them really impressed me. in looking for all the info I became strangely drawn to this town and very curious as to what life is like there. For example, every single kid at one of the schools qualifies for the state's free lunch assistance program.

    [attachment=274,55]
22. 

    Tim
    Member
    Posted 1 year ago #

    Nathan, just a suggestion. It would be good to separate the entries into separate threads somehow. I found myself wanting to ask a few questions about certain entries, but i don't want to clutter up the thread. The single thread is good for scanning the images though.

    Perhaps a separate forum with list of thumbnails for each entry and then click thru to a separate thread for each entry?
23. 

    Tim
    Member
    Posted 1 year ago #

    Pat, I like what you did to show the actual number as well as the percentage. One question, for the US level data, both older brackets are 15%, but it looks to my eye like one is a larger percentage than the other because of the way the "shifting" is done. Can you explain more about it?
24. 

    nathany
    FlowingData Editor
    Posted 1 year ago #

    Good idea, Tim. I'll see what I can do for next time. Feel free to discuss here for now though. I think it'll be ok :)
25. 

    mrflip
    Member
    Posted 1 year ago #

    These are awesome, and I especially love Tim's (faces) and Pat's (boxes). They both brought out the most compelling part of this data: Yes, poverty exists in America, and it most affects those who have no control over it, children. Tim, your graphic humanized the numbers in an amazing way.

    @Tim -- I think yours would be even more powerful if it presented a macro or historical view to go along with the story, something that said "This is a /national/ problem, one that isn't going away."
    Ob nitpicky suggestion: The inline footnote numbers are distracting, and the full URLs at the bottom can't easily be typed in. I'd ditch the superscripts, change the box at the bottom to text and use a URL shortener: "National poverty statistics from Kaiser Family Foundation, http://tinyurl.com/statehealthfacts . Shelby's demographics from US Census Department via Wikipedia, ..."

    @Pat -- one of the other striking things about the data is its geographic distribution. Rather than split out into 50 boxes, could you group the states into either Census regions, or perhaps use @pragmatic's map to visually group the states into regional blocks with similar stats?

    Sorry to be critic and not participator -- I'm knee deep in another visualization and can't let myself play, but I'll definitely jump in on the next one.
26. 

    Pat McComb
    Member
    Posted 1 year ago #

    Tim asks, "for the US level data, both older brackets are 15%, but it looks to my eye like one is a larger percentage than the other because of the way the "shifting" is done. Can you explain more about it?"

    Good question. Both brackets should represent 15% of their respective populations. The thicknesses of the bars do not ideally match the population numbers. If, for example, I'd used just a few bars for the seniors, the 15% in the poverty row would extend off the left of the page.

    However, in the US total stats, there is enough room for adjustment which would bring those stats into closer vertical alignment. Since this wouldn't have worked at the state level, I didn't feel compelled. But if I do a round of corrections, I'll give it a try.

    Thanks!
27. 

    Pat McComb
    Member
    Posted 1 year ago #

    mrflip asks, "Rather than split out into 50 boxes, could you group the states into either Census regions, or perhaps use @pragmatic's map to visually group the states into regional blocks with similar stats?"

    I may try that. Thanks!

    I was impressed with hadley's elegant post.

    And Tim, yours hits home vividly. I'm reminded of the statistics professor at UofM who (then nearing retirement) told his students that people do not mentally retain statistics well at all -- that stories make more of an impression.
28. 

    whatype
    Member
    Posted 1 year ago #

    Hi there,

    The first thing I wanted to avoid was showing this over a map. This, for me, is a common mistake, people tend to show numbers using maps. But maps are the visual representation of geographic data, therefor there’s no real relation between the size of the states and the number of people that live in those states (this was a common mistake on most maps during the last USA elections). By doing a cloud of words, where each state name was represented with the type size relating to the number of it’s population, one can be more accurate on the info.

    Another thing to take into account is that 17% of the California is very different to 17% of the Wyoming population, so by charting just percentages without taking into account the number of people each state has is not the best way to approach this. So I went to the Census Bereau and got the population by state data.

    But even after doing the word cloud, I wasn’t representing the amount of people that lives in poverty, at least not in a visible way. I then made the circles in the background all in proportion to the total of people living in poverty in each state. Please notice that the intention of this circles is to give a visual understanding of the number of people, but not to give the precise number of people that live under poverty.

    Well, this is my approach to this ‘problem’ that Nathan presented. My guess is that there are a number of questions that my visualization didn’t answer and that there are much better ways of solving this. I wanted to show people and not numbers, it's possible that I've over complicated it. Please let me know what you think of this. You can download the pdf on my blog http://whatype.wordpress.com/

    [attachment=281,58]
29. 

    Tim
    Member
    Posted 1 year ago #

    @mrflip thanks for that, good ideas. The timeline would be interesting. It would be great to see pictures of people as they get older and the effect poverty has on them.

    I did think about showing it compared to poverty in a different country. Not to diminish the problem, but here's poverty and then there's poverty! The poverty line in developed countries is still something that other people would aspire to.
30. 

    Tim
    Member
    Posted 1 year ago #

    This one shows the percentage plotted against the total number of people in poverty by state for each age group. I colored them based on the latitude of the states capital city which is good enough to show the north-south effect.

    Imagining that I had to try and address the poverty issue, I drew a green box at 25% and 1M people. The approach would be to come up with a strategy for each cluster of states outside the green zone, and pull those outliers back to the same level as those inside. From there I would start contracting the green zone and repeat.

    That seems like a fair way to bring everyone along.

    In the spirit of supporting flowing data, I did this one with a trial version of Tableau. It was pretty easy even without prior experience of the software. I am impressed. Now can someone send me a free copy cos the joy expires in 14 days? :)

    [attachment=284,60]
31. 

    JonPeltier
    Member
    Posted 1 year ago #

    So far, I like Hadley's charts the best. They are sorted by a meaningful parameter (i.e., by the value being reported(, and not by the accidental alphabetical order of the states. The dot plots show the data with a minimum of distortion. I haven't posted anything yet, and may not if I don't get some time for it; I made some preliminary graphics using bar charts at first, then switched to dot plots, which are better for showing multiple series at once.

    I agree with whattype that maps can distort the data, but I find them better then his bubbles for showing geographical and demographic effects (e.g., north-south, northeast vs. midwest vs. west coast, bible belt). pragmatic's map would have been better in gif or png format. At first I thought the bars in whattype's graphic related to population, but on closer inspection it is related to the extrinsic property of the length of each state's name.

    The spider chart didn't do much for me. The Wordles do even less for me. I've blogged against both of these formats recently.

    Pat's mosaic suffers from the 15% of ages 19-64 being wider than the 15% of ages 65+. I fear the same distortion probably exists through his 50 alphabetically arranged states.

    Tim's scatter plots of # of people vs. % of people is rather interesting, and the color coding is a good first cut at geographic dependencies.
32. 

    nathany
    FlowingData Editor
    Posted 1 year ago #

    Jon, I am 100% unsurprised by your favorite :)
33. 

    silasfisher
    Member
    Posted 1 year ago #

    Hello all,

    I think this graph is nice because it reflects the key data points in one visual and is relatively easy for any viewer to digest.

    It's my first submission so hope everyone enjoys and welcome any feedback.

    -Silas
34. 

    nathany
    FlowingData Editor
    Posted 1 year ago #

    Hey Silas, did you have trouble attaching your file?
35. 

    silasfisher
    Member
    Posted 1 year ago #

    Oops, wrong format. Attached is the jpeg version.

    [attachment=293,63]
36. 

    uuilly
    Member
    Posted 1 year ago #

    Sorry can't seem to delete this errant post.
37. 

    uuilly
    Member
    Posted 1 year ago #

    sorry, can't seem to delete this post.
38. 

    uuilly
    Member
    Posted 1 year ago #

    Both images and videos were made using http://www.uuorld.com

    videos:
    http://www.vimeo.com/2929769
    http://www.vimeo.com/2929764
    http://www.vimeo.com/2929753
    http://www.vimeo.com/2929778

    and stills:
    [attachment=297,64] [attachment=297,65] [attachment=297,66] [attachment=297,67]
39. 

    Pat McComb
    Member
    Posted 1 year ago #

    Here is a revise on my earlier submission. I was able (with the luck of the numbers) to get the US 15% bars closer to flush. An advantage in the wider US graph is that it freed up some vertical space for better placement.

    JonPeltier is right that the same proportion issue applies to the 50 states. It is a consequence of having each full square equal 100,000 people. One way around this would be to use a non-integer number of horizontal bars. But I think that would create too much visual clutter. Also, none of the states created the same kind of visual distraction that was present in the earlier US graphic.
    The criticism of alphabetical order has a point. I chose to alphabetize because I expect potential viewers of the graphic to zero in on their own state.
    Views by rank allow for easier comparison, but they also tacitly imply a range for "normal." The US has a much larger poverty problem than other industrialized nations.
    (This is a difficult area, though. The UN's human poverty index uses a few criteria. The income criteria is the percentage of people in a nation who earn less than 50% of the median income. Arguably, this is just as much a measure of income disparity in a population.
    The US and the UN measures appears to be the standards. I'm not an expert on poverty so I do not know any other measures which are widely accepted.)

    whatype's graphic is interesting. I was impressed that the rough placement by location worked so well. A couple suggestions.
    The black bars are visually jarring and dominate the graphic at the expense of the statistics.
    Also, it appears that the circles are proportional to the diameter instead of the area. Since they are indicating size in two dimensions, pi*r^2 would be more accurate. It also may make it easier to read.

    [attachment=300,68]
40. 

    martian
    Member
    Posted 1 year ago #

    Nathan, When clicking on the attachments, it appears the MIME type is set to application/octet-stream on all files, including images, so my browser (Firefox) forces me to save the images to my hard disk to view them rather then just opening them in the browser like I'd expect. I'm not sure how hard it would be to add the appropriate MIME types to the attachments, but if you could I think it'd be much appreciated.

Reply »

You must log in to post.

About this Topic

• RSS feed for this topic
• Started 1 year ago by nathany
• Latest reply from nathany