Wednesday, August 29, 2007

The rich get richer ... you know the drill

The NYTimes has an editorial today that's sort of a "duh" moment (my emphasis):
The Census Bureau reported yesterday that median household income rose 0.7 percent last year — it’s second annual increase in a row— to $48,201. The share of households living in poverty fell to 12.3 percent from 12.6 percent in 2005. This seems like welcome news, but a deeper look at the belated improvement in these numbers — more than five years after the end of the last recession — underscores how the gains from economic growth have failed to benefit most of the population.

The median household income last year was still about $1,000 less than in 2000, before the onset of the last recession. In 2006, 36.5 million Americans were living in poverty — 5 million more than six years before, when the poverty rate fell to 11.3 percent.

...

The gains against poverty last year were remarkably narrow. The poverty rate declined among the elderly, but it remained unchanged for people under 65. Analyzed by race, only Hispanics saw poverty decline on average while other groups experienced no gains.

The fortunes of middle-class, working Americans also appear less upbeat on closer consideration of the data. Indeed, earnings of men and women working full time actually fell more than 1 percent last year.

This suggests that when household incomes rose, it was because more members of the household went to work, not because anybody got a bigger paycheck. The median income of working-age households, those headed by somebody younger than 65, remained more than 2 percent lower than in 2001, the year of the recession.

Over all, the new data on incomes and poverty mesh consistently with the pattern of the last five years, in which the spoils of the nation’s economic growth have flowed almost exclusively to the wealthy and the extremely wealthy, leaving little for everybody else.

Standard measures of inequality did not increase last year, according to the new census data. But over a longer period, the trend becomes crystal clear: the only group for which earnings in 2006 exceeded those of 2000 were the households in the top five percent of the earnings distribution. For everybody else, they were lower.

This stilted distribution of rewards underscores how economic growth alone has been insufficient to provide better living standards for most American families. What are needed are policies to help spread benefits broadly — be it more progressive taxation, or policies to strengthen public education and increase access to affordable health care.

Unfortunately, these policies are unlikely to come from the current White House. This administration prefers tax cuts for the lucky ones in the top five percent.

But we already knew that, didn't we - that is, those of us down here in lower 95%...

Labels: ,

8 Comments:

At 1:32 PM, August 29, 2007 Blogger Barry Leiba had this to say...

«[...]median household income rose 0.7 percent last year — it’s[sic] second annual increase in a row— to $48,201.»

(Yes, the original NYT source said "it's"; arrrrrrrrrrrrrrgh!)

I've always questioned whether "median income" is a sensible one to report. "Average" isn't either, clearly. To see why neither of these works, suppose we have the following set of incomes:
A: $20,000
B: $20,000
C: $20,000
D: $40,000
E: $500,000

The median is $20,000, and the average is $300,000. Median is far better than average (mean) here, but neither reasonably reflects the situation.

With the assumption that actual incomes are pretty smoothly distributed, median isn't too bad to use for this sort of thing, but it still strikes me as being of limited use. What I'd rather see is a set of ranges, with the median perhaps used as a focal point:
10% have household income between [a] and [b].
10% have household income between [b] and [c].
etc....

The median might be about $48K, but it might be the case that only about .1% of the households are close to that... or it might be that 25% are close to that. From the median, alone, I have no idea.

 
At 2:25 PM, August 29, 2007 Blogger fev had this to say...

Should we start a campaign to get the NYT to report standard deviations? That might help.

 
At 3:16 PM, August 29, 2007 Blogger Electric Dragon had this to say...

If they're reporting the median, the inter-quartile range might be more appropriate.

I can't find that data on the census website but what it does list are certain percentiles, eg.:
10th percentile - $12,000
20th percentile - $20,035
80th percentile - $97,032
90th percentile - $133,000

Source - Table A3

 
At 5:27 PM, August 29, 2007 Blogger Barry Leiba had this to say...

Yes, what E.D. says is exactly what I was suggesting.

 
At 6:45 PM, August 29, 2007 Blogger The Ridger, FCD had this to say...

I agree with you about what kind of numbers should be reported, but then again, this was an editorial. We shouldn't be going there for the facts, all the facts, and nothing but the facts.

There's always a problem with people and numbers - we don't get taught how to deal with them and what we do get taught we (mostly) forget. Like that study that everyone (including some math types) was howling about, the one where guys claimed an average of 7 partners while women claimed 3. Where are all the extra partners? people cried. Are they all hookers? Are the men lying - exaggerating? Are the women - under-reporting? Are they both?

Turns out, of course, that the study clearly said those were the MEDIAN numbers, which means all you need are many women reporting 1-2, and even just one guy reporting 20, and you get that sort of thing.

 
At 8:57 AM, August 30, 2007 Blogger fev had this to say...

The sex-partners thing -- mean vs. median -- really gets to the core of what ails reporting about science and math. There _really isn't_ a debate there: the numbers are perfectly plausible, no matter what people say. But journalism tends to confuse policy debate, in which you're supposed to admit and give a certain minimum weight to all sides, with hypothesis testing, in which you set a significance level and eventually reject Young Earth creation, the harmlessness of tobacco and the like as (mind a naughty word on your blog, Ridger?) bullshit.

I'm still working on this one.

 
At 10:11 AM, August 30, 2007 Blogger Barry Leiba had this to say...

True, that one's an editorial... but the non-editorial "just the facts" news items say the same things.

«Turns out, of course, that the study clearly said those were the MEDIAN numbers, which means all you need are many women reporting 1-2, and even just one guy reporting 20, and you get that sort of thing.»

There, so that supports my point: you're a well-informed, intelligent person, and you just got it wrong. What you say is true of the MEAN, not the MEDIAN. Changing the value at the top to be off the scale will affect the mean, but won't change the median.

Now, suppose there are 9 men and 9 women, and they respond thus:
W: 2 2 2 2 2 2 2 2 2
M: 0 0 0 0 0 3 3 3 9
The MEAN is 2 for women and 2 for men. The MEDIAN is 2 for women and... 0 for men.

But when you look at the raw numbers, what would your analysis of relative promiscuity be? The numbers work out either way, but how you view them is very dependent on how they're presented, and whether you understand the statistics behind the presentation (as you say).

 
At 2:18 PM, September 08, 2007 Blogger Barry Leiba had this to say...

(Stupid error in my first comment, way above: a friend pointed out that the average is $120,000, not $300,000. It doesn't change the point, but I should correct the error.)

 

Post a Comment

Subscribe to Post Comments [Atom]

     <-- Older Post                     ^ Home                    Newer Post -->