FlowingData Forums » Data Visualization

Decon/Recon: Showing suicide rates in Japan

Started 3 years ago by nathany / 5 posts

  1. nathany
    FlowingData Editor
    Posted 3 years ago #

    Can you improve this graphic showing suicide rates and unemployment rates in Japan?

    There are quite a few things that we can do to represent this data a little better. For starters, something needs to be done with those labels. What else? What is the graph trying to show? Does it succeed?

    Unemployment data is from the OECD Factbook 2007 and the suicide rates are from the World Health Organization. For your convenience, here are suicide rates and unemployment rates by year in one spreadsheet. I've included population just in case.

  2. JonGoldhill
    Member
    Posted 3 years ago #

    Presumably the intention is to show a correlation between unemployment and suicide. I would question whether it does so.
    It doesn't help that it shows annual figures for unemployment and 5-yearly ones for suicides.
    Finally the y-axis label ought to make it clear whether we're looking at numbers or numbers per 100,000.
    I am not a statistician, but I thought you could represent correlation by a single number (a sort of degree of correlatedness)?

  3. James Scott-Brown
    Member
    Posted 3 years ago #

    This graph could be improved in several ways. The lines should be labelled better: their labels are separated from the actual graph, and placed in a separate "legend" box, so that it is not immediately obvious which line represents employment, and which represents the suicide rate. The actual units of the y-axis should be stated clearly (a rate is not a total, and to simply refer to a "rate" is inadequate). Perhaps the graph could be labelled to show the times at which event with a significant effect on the economy or suicide rates occurred.

    Alternatively, an entirely different presentation could be used. Rather than plotting time on the x-axis, and using the y-axis to represent both unemployment and suicides, the x-axis could represent unemployment, and the y-axis suicides. This would require plotting distinct data points, rather than lines. By doing this, you would show how the variables have changed together, rather than how they have changed over time. This has some disadvantages (such as not being able to see whether increases in unemployment precede increases in suicides), so it may be best to present both graphs along-side each other.

  4. James Scott-Brown
    Member
    Posted 3 years ago #

    Here's a rough illustration of what I mean: http://skitch.com/jamesscottbrown/1c8u/japan
    I've plotted the data for all 6 years for which both suicide and unemployment statistics are available.

    It looks as if there are two clusters: those up to 1995 and those after. Up to 1995 it looks as if there is a negative correlation between unemployment and suicides. After 1995, this relationship breaks.

  5. nathany
    FlowingData Editor
    Posted 3 years ago #

    hey james, thanks for sharing. i was hoping someone would touch on that.

    the original wants to show correlation, so why not do the scatterplot like you did? of course, as we see in your plot, there's really not enough data to make much of a conclusion

  6. admin007
    Member
    Posted 3 years ago #

    Hi, dear friends

    Broadcast yourself for free on Japanese YouTube with amazing and exciting video clips with friends & family across the world.

    you find here --- japan youtube, tokyo japan, japan hotels, japan, Japaneses, tokyo, travel, asian,vehicle, tokyo, china, new movies, funny videos. you tube, video, cars, car insurance, student loan, consolidating priv.

    More information kindly visit below here ----

    http://www.japanyoutube.com/

    Cheers
    Admin


Reply

You must log in to post.

About this Topic