Showing posts with label DataViz. Show all posts
Showing posts with label DataViz. Show all posts

Saturday, February 27, 2010

What Does the Obama Job Chart Really Mean?


An old joke says that there are three kind of lies.

  1. Lies

  2. Damn lies

  3. Statistics
To that list, I would add data visualizations, especially when seemingly misused to purposely misinform the electorate.

By now, you've likely seen the chart, courtesy of Barak Obama's Organizing for America (OFA), showing the miraculous reduction in job losses after President Obama took office and the Congress passed the $787 billion Stimulus plan.


This visually-striking graphic is such fantastic news that Organizing for American even made a video highlighting the chart.




Something about this jobs chart bothers me. As of February, 2010, unemployment is still stubbornly high. So, how can it be that the Obama Administration seemingly eliminated unemployment? From the chart, it seems that the President halted and reversed unemployment to roughly the same level as December, 2007. I showed this chart to a number of engineering and accounting friends--people accustomed to working with charts and graphs--and they came to similar conclusions based solely on the chart. However, it is not true.

This chart seems designed to purposely deceive. How? The chart shows the monthly change in unemployment, not total unemployment. Of course, the original chart does not indicate this, but I've been able to reproduce the chart using source data from the Bureau of Labor Statistics.



How is this deceptive? Allow me to illustrate with an example. A man goes out on the town with his girlfriend to celebrate her birthday with dinner and a movie. Over the evening, he makes a number of purchases as shown in the table.

Graphically, the evening's spending per transaction looks like the following chart. At first glance, it appears that this lucky man had a great night on the town with his favorite girl and even turned a small profit. From the chart, one might assume that the man spent some money but then started making money after dinner and ended the evening with more money in his wallet than when he started!

How is this possible? It's not. It's a bit of graphical and presentation trickery. The chart shows the change in the data, not the present state.

The previous table and chart shows the amount of spending per transaction. However, the real effect on the man's wallet is the cumulative spending, as illustrated in the following table and chart.


While the spending per transaction, or the change in spending, is dramatic during the middle of the evening, it tapers off at the end. In fact, because the man found a $5 bill, the evening apparently ends on a high note! However, the cumulative, out-of-pocket expenses for the evening totals nearly $300. In one presentation of the data, the man apparently ends the evening with $5 in his wallet when in reality, he is down $300.

How does apply to Organizing for America's jobs chart?

The original chart shows the number of jobs gained or lost per month, or the monthly change in the jobs number. However, if the unemployment rate remains unchanged at 5%, 50%, or even 100%, the monthly change in unemployment is zero. Actual unemployment is not zero, just the monthly change in unemployment. See the difference? The following chart presents the cumulative job loss over the same time period, using the same data, and more clearly shows that unemployment increased over time and has not returned to December 2007 levels. While it is good news that monthly job losses have slowed (what the Organizing for America chart actually shows), the unemployment situation has not miraculously disappeared (what the Organizing for American chart seems to imply).


There are many ways to distort the visual presentation to reflect a particular political view. For example, here is the same data presented in yet a different manner. In this example, the chart compares cumulative unemployment during the final year of President George W. Bush's presidency and the first year of President Barak Obama's presidency. While the original graph seems to indicate that unemployment improved after President Obama took office, in reality the job market continued hemorrhaging!

Here is a chart of the showing the total number of officially unemployed based on Bureau of Labor Statistics data. The number is even greater than the cumulative chart before as there were over 7.5 million unemployed when the crisis deepened.



To put this into historical perspective, how does the current unemployment trough compare to those from prior recessions, measured in months from the beginning of the recession? The current trough is deeper and it appears will last longer than prior recessions.

The original Organizing for America chart appears to be purposely deceptive. It falsely appears to indicate that the Obama Administration has completely reversed unemployment when clearly, this is not the case. With the nation already split on so many issues, the electorate needs to operate from a truthful set of data without willful distortions. Sure, there can be multiple interpretations of the same data, but those interpretations should clearly indicate what data is presented.

See also ...

Provides previous examples how Organizing for America attempted to distort open and honest debate

Sources:

Bureau of Labor Statistics

Spreadsheets and Chart Data

  • All spreadsheets and charts used in this blog posting
    http://www.editgrid.com/user/soquel_by_the_creek/Obama_Admin_Job_Chart
  • BLS 1-month unemployment data
    1-Month Net Change
    Series Id: CES0000000001

    Seasonally Adjusted
    Super Sector: Total nonfarm
    Industry: Total nonfarm
    NAICS Code: -
    Data Type: ALL EMPLOYEES, THOUSANDS


    Year,Jan,Feb,Mar,Apr,May,Jun,Jul,Aug,Sep,Oct,Nov,Dec,Annual,
    2000,249,121,472,286,225,-46,163,3,122,-11,231,138,
    2001,-16,61,-30,-281,-44,-128,-125,-160,-244,-325,-292,-178,
    2002,-132,-147,-24,-85,-7,45,-97,-16,-55,126,8,-156,
    2003,83,-158,-212,-49,-6,-2,25,-42,103,203,18,124,
    2004,150,43,338,250,310,81,47,121,160,351,64,132,
    2005,136,240,142,360,169,246,369,195,63,84,334,158,
    2006,262,326,304,174,31,69,232,141,100,43,201,177,
    2007,194,104,239,92,149,55,-20,-71,52,86,128,70,
    2008,-10,-50,-33,-149,-231,-193,-210,-334,-458,-554,-728,-673,
    2009,-779,-726,-753,-582,-347,-504,-344,-211,-225,-224,64,-150(P),
    2010,-20(P), , , , , , , , , , , ,
  • BLS Unemployment Data
    Series Id: LNS13000000

    Seasonally Adjusted
    Series title: (Seas)
    Unemployment Level
    Labor force status: Unemployed
    Type of data: Number
    in thousands
    Age: 16 years and over


    Year,Jan,Feb,Mar,Apr,May,Jun,Jul,Aug,Sep,Oct,Nov,Dec,Annual,
    2000,5708,5858,5733,5481,5758,5651,5747,5853,5625,5534,5639,5634,
    2001,6023,6089,6141,6271,6226,6484,6583,7042,7142,7694,8003,8258,
    2002,8182,8215,8304,8599,8399,8393,8390,8304,8251,8307,8520,8640,
    2003,8520,8618,8588,8842,8957,9266,9011,8896,8921,8732,8576,8317,
    2004,8370,8167,8491,8170,8212,8286,8136,7990,7927,8061,7932,7934,
    2005,7784,7980,7737,7672,7651,7524,7406,7345,7553,7453,7566,7279,
    2006,7059,7185,7075,7122,6977,6998,7154,7097,6853,6728,6883,6784,
    2007,7085,6898,6725,6845,6765,6966,7113,7096,7200,7273,7284,7696,
    2008,7628,7435,7793,7631,8397,8560,8895,9509,9569,10172,10617,11400,
    2009,11919,12714,13310,13816,14518,14721,14534,14993,15159,15612,15340,15267,
    2010,14837, , , , , , , , , , , ,
Posted by at 10 comments:
Labels: , , , , , , , , ,
Subscribe to: Posts (Atom)