The Obama administration just posted a graph showing monthly job loss from December 2007 (Bush in red) up to last month. Discuss.
Update: There's a video version now [via infosthetics].
Maybe I’m just being overly picky, but I made some alternate versions to show how the intensity of colors chosen creates some skew. Important especially when you’re using color to separate who is responsible for what. Luminosity matters.
Color choice here is related to convention: red for Republican, blue for Democrat.
Although it follows the convention. The intensity is still an issue. This blue is very low weight, where as the red is very heavy. I’ll bet if you put the same graph side by side with these two colors, the red would look way worse.
I think we all understand the convention but the problem was more with the intensity than it was with the hue. Check out my version of the graphs particularly the black and white version of the original and you can see how they’ve given emphisisbto the bush years using intensity.
Of course its also true that on comparison graphs red is often used to highlight the negative parts so that’s a minor problem too but the intensity is what I was complaining about.
It may be more then just the color of the political party. Those colors look very similar to the color theme of the barakobama.com website.
Take a look at http://my.barackobama.com/page/content/commitforhealthreform?source=VIC-021110
— The dark red on the thermometer and on the ‘donate’ button look similar to the red used on the red side of the graph.
– The light blue is similar to the blue used for the words “Organizing for America Blog”.
My Apple DigitalColor Meter shows a similar (but not exact) RGB values for those colors.
Yeah I think its pretty clear the color choices were right out of the official campaign palate. Just to clarify my position I don’t know that this was intentional skew the graph probably just a lack of understanding.
Its a common problem actually where a marketing team comes up with a great set of colors for marketing materials and then they need a graph and use the same style.
Further to your point of red symbolizing negative (or the corrective pen of our school years) blue represents creativity -so it’s not only a negative on one side, it’s a positively charged colour on the other.
Mr. Obama is a master of the tools in the old book “How to Lie with Statistics”. Let me highlight a few issues, starting by pointing out that this plot does not represent “jobs”, but rather the change in jobs (math wonks call this “the first derivative”). The correct plot requires a plot of actual jobs; you must add up all of the negative bars on the left. Then you notice that the job “gains” on the right are negligible.
We also must remember that the important issue is POLICY, not PARTY. Following the election of a Democrat Congress in 2006 (off the graph to the left), Bush adopted the Keynesian and corporate socialist policies of the Democrats (as Richard Nixon “We are all Keynesians now” did, creating the Nixon Carter recession).
Keynesians borrow money from the private sector and spend it on politically favored economic activity. The overwhelming body of peer reviewed economic scholarship shows that this loses jobs (tax cuts cause job increases). The first Keynesian “stimulus” bill was under Bush in early 2008 (jobs dropped). The second is under Obama. Jobs dropped horribly.
In short, the policies still being advocated by the Democrats, implemented by a Democrat congress with the complicity of both Republicans and Democrat presidents, have lost nearly 7 million jobs, an average wait of 27 months for reemployment, and a 20% underemployment. Now, the job loss has stopped, for a moment, if we include census employees.
Of course, the jobs loss HAD to stop somewhere; everyone can be unemployed. But this is not a vindication of the policies.
Admittedly, any Republican who does NOT run on “Repeal and Replace” will NOT have an effective message against the misgovernment that Democrat policies constitute. They MUST run as Reaganites, recognizing that “government is the problem”. restoring the rule of law to economic transactions (no more taxpayer money to unions who supported Obama), and lowering the burden of government on people who create jobs.
The Democrats could ALSO be successful if they also support such sensible policies. If they do not, they will need to lie with statistics, as in this plot.
>>> Now, the job loss has stopped, for a moment, if we include census employees. <<<
Not true. Even if you exclude the 58,000 temporary census workers, or exclude ALL government workers, the private sector still added over 100,000 jobs in the last few months.
I agree with the Libertarian. The chart is misleading to say “Administration” as it implies that the movement in the chart is due to actions of either the Bush or Obama administrative policies (also because it represents job losses not jobs).
Rather, the chart should say “Bush Years” and “Obama Years” thus divorcing policies from economic movements, which move only in part due to policies reacting more to things like demographic shifts (e.g. Boomers passing peak spending years), major events (e.g. 9/11), and global economic developments (e.g. the rise of China as a manufacturing powerhouse).
heh. and the location for those is
Is it just me, or does it look a little bit too perfect to be believable?
It’s a clear message well told— and it needs to be told! I urge everyone who values the truth to link to it and get the message out before the big lie being pushed by a noisy misinformed (or misinforming) rabble becomes received wisdom.
Remember Yeats words:
‘The best lack all conviction, while the worst
Are full of passionate intensity.’
Yes, the message is clear: There was a recession during the Bush years where jobs were lost and now that time has passed, job loss has slowed as the pace could never be sustained indefinitely.
Oh nonsense…. it’s about the cumulative unemployment and the deep deep hole. Velocity is easily manipulated and the sheer symmetry should clue you to the fact that this thing is a kludge.
Strange how the wondrous people of November 2008 have become, almost by magic, the “noisy misinformed (or misinforming) rabble” by February 2010, don’t you think?
Ah, how easy to paint a rosy picture when you hand select the “Bush years” data to only include the worst. Job losses less negative does not a victory make.
@quantacide — The recession began in December 2007 according to the National Bureau of Economic Research (They make the proclamation), so it makes sense to begin the graphs in December 2007. The NBER has not proclaimed that the recession ended.
And… job losses more negative make what?
The graph shows job losses per month over a certain amount of time.
The fact is that there are fewer people employed who are eligible to lose their jobs – therefore smaller bars closer to Jan 2010.
Recovery is not illustrated here. What is illustrated is a very simple way of distracting people with charts
There are still plenty of people with jobs to lose. Its not like US Employment has reached -zero-. It shows that the bleeding is starting to stop but I do agree…a path to recovery has to be the next step and you can’t just look at this graph and say “Well everything is ok now!”
Job loss as a percentage of the working population would tell a better story.
This is deliberately misleading.
Thanks martin for the mathematical terminology – I didn’t have my Maths hat on this morning!
That all depends if you normalise it to the available working population or as a percentage of those left. What would you recommend?
For the record there is an approach adopted globally for reporting unemployment: Thirteenth International Conference of Labour Statisticians (Geneva, 1982) @ http://laborsta.ilo.org/applv8/data/c3e.html
If everyone were unemployed, there’d be no further job losses. Only cumulative losses or percent unemployed tell the story.
Cumulative would show a slowing in the rate of job losses. If you’re attempting to say that its *not* better under the Recovery Act this would fail.
Percentages would indeed be interesting – both in terms of cumulative (well, really that would just be the *state* of unemployment) or monthly changes.
The truth is that these things tend to be cyclical and it is rather irrelevant what party is in the White House. It is quite likely that the graph would look the same if the roles were reversed.
thank you, Lee. The point is that we are in the midst of an economic cycle caused not by greedy fat-cats, but ultimately by politically motivated decisions in Congress. Regardless of who is in power, economic cycles will happen, and there’s not a dern thing we can do about it.
No, it was mostly the banks to blame.
It was *their* poor understanding of statistics getting them in trouble.
Congress is only responsible for letting it happen, (and probably not doing enough to stop it happening again).
Theres always going to be circles in anything, but thats an observation of a pattern, and not a cause of anything by itself.
This thread just proves that data doesn’t make any difference to some. People will believe what their ideology or religion (is there a difference?) tells them to believe. Facts and data are mistrusted and misused. Followers of one network – Fox News – are living in a cult-like alternate reality and told to mistrust facts from other sources. Its a sad irony that flowingdata is part of this double-edged sword.
Since FOX is the only station that reports the other side? I think we all have our views that will not waiver. Like the poster said above, it would likely look the same regardless of who is in office and that these things occur in cycles.
FOX is the only station that presents its self as the single good source of facts, and all other views as conspiratorial.
Informed citizens get information from a broad array of viewpoints, and forms their own opinions — anyone who tells you “only listen to me, the rest of them are lying!” is suspect.
So, gnomic, do you include yourself in the “People” (seems all-inclusive to me) that “believe what their ideology or religion… tells them to believe?”
What’s your ideology/religion and what facts and data are you misusing? Maybe the generalization about Fox News “followers” living in a cult-like alternate reality?
The truth is that this evaluation should be done over the percentage of US unemployment, not over the job loss. Here we see a descent of the unemployment rate, which only means that it is stabilizing. Only NOV’10 represents a real decrease in unemployment…
Nathans title for this post is “Road to recovery – Is the Recovery Act working?” poses a question that requires more graphs to put this data into perspective.
I’d love to see similar graphs for job acquisitions, employment, and unemployment. A quad of these graphs would allow viewers a better understanding of the issue. Even that would not be a complete picture.
Unfortunately, I see graphs like this at the core of the validity issue with graphs and statistics in general. This graph represents the data that the Obama administration has chosen to present from the CBO. It is but a sliver of the issue but serves the political need to present a positive view regarding the current administrationâ€™s actions versus the previous administration.
On a personal note, I am an active Obama supporter. Iâ€™m interested in seeing relevant accurate data, which this graph contains, being used to provide the transparency we have been promised. I think this graph falls short of that goal and serves more as a tool to shape perceptions. How often do we see graphs designed for such purpose? How seldom do we have access to the whole picture?
The CBO report is here:
A CBO transcript is here:
The visualization is a good one, yet out of context.
Thanks for the post and great responses so far!
Does this graph misrepresent the data? No. It does exactly what it’s label says. There are other graphs availible that help tell the entire picture. The raw data is availible. People have identified the limitations and problems with the data set.
The furor over this graphic is that it says something that a large number of people don’t want to believe, either because of their core beliefs or because it clashes with (mis) information from other sources that they trust.
So they attack the data, the message, the graphic and seak to discredit it.
Me? I think this is simply a datapoint with a positive trend and there are issues with data quality, but overall this is accurate. It makes its point simply and is understandable.
And that’s what is making people uncomfortable. This could have been about any polorizing topic and we would be having the same discussion.
The more interesting concept here is the relation of the information to the viewer and the cognitive dissondance that it is causing.
“Does this graph misrepresent the data? No. It does exactly what itâ€™s label says. ”
Oh pish posh. This is cherry-picking at its most obvious. As others have pointed out, a clear picture from graphs requires, at the least, three to four other graphs (read up and down thread).
What is the data source for this? These numbers are very different from what I found at Bureau of Labor Statistics. http://bit.ly/dzfDyc
Oops. I pasted in the wrong url. This is what I meant to link. http://bit.ly/cugcK2
These data are from the Establishment Survey which measures the level of employment, not the rate of unemployment.
This graph shows nothing worthwhile. We can’t observe just how bad things would be had Republicans remained in charge. All this shows it takes a long time to clean out their messes.
No doubt the economy in Obama’s first year was less bad than Bush’s final year. But the only thing that this chart shows is natural economic cycles. Funny that the current administration claims that its stimulus bill would not have a positive impact on the US economy until mid-late Spring 2009. Yet they post this graphic that shows things were already starting to improve (albeit slowly) before any measures were taken.
“But the only thing that this chart shows is natural economic cycles.”
Is there some principle of “natural economic cycles” that indicates the rate of increasing job loss going into a downturn should approximately equal the rate of decreasing job loss coming out of the downturn?
You could assert the same thing if the Obama section of the graph slowed job losses twice as fast, or half as fast, and certainly both can’t be representative of “what would have happened naturally, without intervention”.
“stimulus bill would not have a positive impact on the US economy until mid-late Spring 2009”
All this graph shows is a plateau (unemployment acceleration has essentially stopped).
Yes! For the most part that is the case.
Check out the historic data http://data.bls.gov/PDQ/servlet/SurveyOutputServlet?data_tool=latest_numbers&series_id=CES0000000001&output_view=net_1mth
@Alex – it reminds me of that skit that conan o’brien used to do with celebrity pictures. start with just the body and then zoom out to see everything else that’s going on for a completely different story.
I think it is funny that we are using the word “natural” with the economy, as if it is some sort of organic structure that existed and humans (europeans?) just discovered instead of something that was created over time and that we change anytime we screw it up. Labeling recessions as “natural” just makes downturns and greediness as acceptable when I don’t think they should be.
This graph shows the jobless RATE, not the % of people unemployed. The percentage of unemployed will continue to grow so long as we have negative bars. This graph demonstrates that virtually no new jobs have been created (except seasonal temp help in Nov 2009).
“his graph demonstrates that virtually no new jobs have been created”
This is only true if you consider two people fired and three people hired “one job created”; this is a definition that is inconsistent with the political language typically used to describe the stimulus.
People are going to get canned during economic downturn no matter what. “Creating” jobs usually means that people get hired, which helps offset some of the people who were fired. This is not a net job gain, but that doesn’t translate to “no jobs created”.
Obama: “Look how well I manipulate data!” « The Sagamore Journal
This is from the same group of people that accused the Bush team of politicizing everything unnecessarily?
I’m pretty sure that Bush never released a simplistic color-coded graph showing the dot-com bubble bursting in Clinton’s final year, even though Bush got blamed for that recession endlessly.
As for this graph–it’s lovely propaganda. It’s piss-poor information presentation.
…and as for Nick’s question (and the administration’s assertion): the “recovery” shown in this chart was mostly complete before even 5-10% of the stimulus money was distributed.
Rule #1 of shady statistics: correlation is not causation.
This graph is obviously drawn by someone sympathetic to the administration —- A BETTER ILLUSTRATION wouldbe to look at it as the first half of a declining “bell curve” starting high on the left, and as time progresses, the bell curve comes down at a slightly less angle until it eventually bottoms out and is going straight across with no verticle change! — That is the true picture — not “down and then up” as this graph wants to illustrate!!!
What this graph needs to give it AUTHORITY and IMPACT is a BIG EMBEDDED LOGO!
If it had a BIG EMBEDDED LOGO, which direction should we skew the drop shadow? Left or Right?
I’ll add that as much as this graph makes a valid (if limited) point, the zero boundary is also misleading. It takes the creation of something like 170,000 jobs per month just to keep up with the expanding population. That — not zero jobs created — is our equilibrium point.
And to even ding the unemployment rate by one percent it takes something like 300,000 jobs (these numbers are probably slightly out of date).
If you popped that zero boundary up to 170,000 we’d still look very deep in the hole.
I agree with everyone else who states that it should be about unemployment. And not just unemployeds with unemployment benefits. I recently heard a story that there’s a huge bubble of people who are about to lose their unemployment benefits (even though they’ve already extended the benefits). This is huge. Who cares about job loss when no one can find a job.
Road to Recovery â€“ Is the Recovery Act working? | FlowingData « f.majakovskij
By a convenient :) coincidence http://data.bls.gov/PDQ/servle…..w=net_1mth is currently showing the following:
Wednesday, February 17, 2010 6:31 PM
The database from the Labor Force Statistics from the Current Population Survey program is currently unavailable. The data will be available as soon as possible. (When this database is not available, it is usually because we are in the process of updating it.)
If you need to contact someone about the program or its data, please send a message to the CPS data questions e-mail address below or call the phone number below.
Thank you for your interest in BLS data, and we apologize for the inconvenience.
Erm, your link broke – care to explain what you’re looking at?
That’s not the survey that these data represent.
The only thing this shows is that increase in joblessness has slowed somewhat. That is it. It does not show new jobs created, the rate of unemployment, etc.
So, The 0bamanation posting this as positive will fool those on the distaff side of the bell curve but not anyone with a scintilla of common sense or intelligence.
There are – Lies, Damn Lies and Statistics.
This is a Statistic. That is all.
It does not *not* show jobs being created either mind.
Unemployment rates wouldn’t either.
For that you need jobs lost, jobs created but just as importantly the hours associated with them (underemployment *may* be skyrocketing without being visible).
Some friends had a problem with the Y-axis so i flipped the graph.
Check it: http://sagamorejournal.wordpress.com/2010/02/17/bushobama-job-loss-graph-a-redux/
As someone elsewhere has observed, “Given the fact that the Democrats have controlled Congress for the entire period in this chart, shouldn’t all the columns be blue?”
“Given the fact that the Democrats have controlled Congress for the entire period in this chart”
Another thing that bothers me is that “job loss” is noted in negative numbers. They should be positive numbers. As in “we lost 100 jobs.” Not “we lost -100 jobs.” Or it should be labeled “job growth” and use negative numbers.
Krees- Check out the flipped graph here:
The issue is that the BLS survey actually reports monthly ‘net change in US employment’, not monthly ‘job loss’. But explaining net change to John Q Public isn’t an easy task.
Small difference but a potentially confusing one.
Thanks pslarkin. I was mostly noting that it’s poor charting form, not so much that it was confusing me.
The chart has a sharp aesthetic, but the page as a whole comes off more like an apparatchik editorial instead of a piece of well-reasoned analysis.
No way – so I just watched Shawshank Redemption today, which I had not seen in a long time. Truly a good film. And so I googled “who is andy dufresne”, and FlowingData comes up on the first page of search results, with an article not about the film but about how context matters. “If we don’t have all the information, then we can’t tell a complete story.” And I think that applies here. http://flowingdata.com/2008/05/08/why-did-andy-dufresne-escape-from-shawshank/
ha, that old post sure takes me back. carrying influence in all areas.
show one mile of road –
can’t find a trillion that young man
Mmhhh ok data surely represents the reality but there is a bias in the potential perception:
– This graph wants to make you believe that Obama’s administration is doing better than Bush’s administration concerning the labour market.
– This is clearly false, because job loss itself is not correlated with the decisions of those two leaders but rather with the impact of the crisis on a specific timeframe and the decisions of the companies according to this same timeframe.
– What I would like to see is the impact of president Obama’s decisions and policies to drive job CREATION, which are currently unexisting.
This is where a correlation may exist, in my opinion.
There are lies, damned lies, and statistics….
I don’t get the problem here. It shows that joblessness has slowed. That is the point… the ‘rate of change of the rate of change’ of job loss became positive. I see no ‘manipulation of the data’ as some are saying.
The rate of change of the rate of change is just acceleration. When you apply a force to a mass – there is some acceleration.
Whether or not you can attribute the blue portion with Obama’s actions or whether or not it would have occurred naturally is hard to say. But there is no ‘conspiracy’ or ‘manipulation’ of the data as far as I can see and I’ve yet to read a credible argument posted that there is.
I don’t have a problem with the Y axis, but I do have a problem with the X axis.
“Obama Administration” should start in Nov ’08. That when businesses knew that the Obama would be running the nation and many started layoffs at that time.
Now that – is an interesting point.
>>That when businesses knew that the Obama would be running the nation and many started layoffs at that time.<<
Many businesses suffered enormous financial losses at that time, and projected even more losses in the coming months. Businesses couldn't get the loans and capital required to run their operations. The US GDP shrank by $1 Trillion, and trillions more were lost in the markets. Finance trumps politics.
The unemployment rate started to jump after April 2008, well before the November elections. See http://www.google.com/publicdata?ds=usunemployment&met=unemployment_rate&tdim=true&dl=en&hl=en&q=unemployment+rate
To somehow implying that businesses started laying people off simply because Obama was elected president is naive and biased. It doesn't matter who was elected president at that time, we would have remained in the same boat for a while.
Businesses lay off staff for financial reasons. Any business who lays off staff due to poor political projections will lose to their competitor.
Note the title of the graph is job loss. Now note how the figures on the Y axis are entitled “Jobs Lost.” Finally, ask yourself why these numbers are negative, because I honestly doubt that the Obama camp wants to show how many jobs Bush created. Don’t understand? -100,000 jobs lost means 100,000 jobs created. If you want people to take you seriously, label your graph correctly.
First, geez oh man people will claim graphs are flawed if they don’t what they say.
Two, graphs are dangerous because people think they’re all scientific.
Three, science is dangerous because people think it’s all truthy.
Four, people are dangerous because they basically believe whatever they want.
And Five. The truth is that the delineation is probably unhelpful, except insofar as you can draw a line between when it was a bad idea to be expanding gov’t expenditure and when it was a good idea to be expanding gov’t expenditure. To do that, you’d have to go back a way, probably. When people were competing to buy houses in Baltimore County, it was time to raise the interest rate and tell everyone to go home to their wife and kids and play foosball or something.
Once everyone starts firing everybody, that’s when deficit spending starts to make sense. Hard to see that on this graph, though. Pretty colors, maybe.
Oh, and six. Obama RULLEZZZ. I’ll just leave that there.
at first glance it looks like Obama is doing a wonderful job…but this is where statistics can lie. First, the data only goes back to 2007…Bush was in office much longer, are they showing up to that date, because the data doesn’t favor the Obama team as much? Second, at what cost is job creation? Would it make sense to spend 1 million dollars to save 1 job? I would like to see the data go back further, and have an overlay of the deficit during the same time.
>First, the data only goes back to 2007…Bush was in office much longer, are they showing up to that date, because the data doesn’t favor the Obama team as much?<
As was said earlier in this thread, the graph begins in December 2007, which is when the recession began according to the National Bureau of Economic Research.
OK, I’m an Obama supporter, but…
Since Obama wasn’t inaugerated until Jan. 21st, do you realistically think his administration was responsible for the inflection in the curve, that he had the impact to began slowing the rate of job loss?
These measures are lagging indicators. Maybe, one could argue that the election in Nov. sparked a change, but I really don’t see how one can imply cause and effect here…by using the colors associated with a sitting administration, there is an implied correlation and cause-and-effect relationship. You just cannot make that assumption. I’d suggest that one read John Paulos and his books for further info on how statistics can be used to “prove” a story for anyone’s particular point of view.
If the argument is that the recovery act impacts job losses, then shouldn’t this graph show both variables? The business cycle explanation makes more sense. More importantly, how will debt created by the stimulus spending weigh on future (net) job creation? To be credible, the relationship between prosperity and government spending (debt) needs to be tracked, and over a period longer than just one business cycle.
How would the graph look if it was colored to show House/Senate majorities, rather than who sat in the White House?
Debunking the “Republican Congress Creates Jobs” Chart Or “How To Make Numbers Say Anything You Want”
The job numbers are probably accurate. But causation has not been proved. Can’t we go back in history and see similar patterns of economic recovery with or without huge stimulus packages? The economy is so complex that pointing out one variable as the cause for job growth is just plain silly!
having studied the basic laws of economics for three years I know that this is total B.S. . Any economic recession by its nature will create its own recovery, both due to capital accumulation and a lower cost of investment. The last two presidents(conservative and moderate) both had booms and busts in their terms, because it wasn’t their doing. The President does have economic responsibilities, but that said it isn’t an easy thing to measure as many of his decisions don’t affect us for a long while. e.g. the rule of 70 says that an economy doubling happens every 70/%growth, with the U.S. averaging 2% growth or a doubling every 35 years, that’s something that presidents do affect, but it’s hard to see 35 years into the future.
Kiplinger.com has provided their own graph. See it here:
Their graph only includes private-sector jobs (No temporary bump from the 58,000 Census workers), and they attempt to smooth out the graph over several months, to account for the seasonal increase in construction workers.
This graph also goes back to early 2007, to show some numbers from before the recession.
Let’s not forget that the market has tended to do better under Democratic presidents over decades.
This was an interesting chart. Most people that I showed it to (engineers, accountants) assumed at first glance that the chart showed total unemployment, which didn’t make sense to anyone. Only later did we figure out that it shows net monthly job loss/gain, or the first derivative.
As political propaganda, was this purposely meant to be deceptive? I don’t know. However, we were able to successfully find the data sources and recreate the data.
What Does the Obama Job Chart Really Mean?
As they say, “correlation does not imply causation.” The timing is indeed interesting, but the fact there there was so much uncertainty between the end of Bush’s term and the start of the Obama Presidency, I’m not surprised that ANY government acknowledgement of the problem might have stopped or slowed the job losses.
The fact that the Stimulus spending is spread out over such a long period does not correlate well. Most of the Stimulus money is still to be spent, a year after passage.
Become a member.
Learn to visualize your data.
What you get
I call myself a statistician, because, well, I’m a statistics graduate student. However, the most important things I’ve learned are less formal, but have proven extremely useful when working/playing with data.
We don’t all start our work days at the same time, despite what morning rush hour might have you think.
There are rules—usually for specific chart types meant to be read in a specific way—that you shouldn’t break. When they are, everyone loses. This is that small handful.
The data goes back to 1960 and up to the most current estimates for 2009. Each line represents a country.