[PRCo] Re: Indian cultural differences

Schneider Fred fwschneider at comcast.net
Sat Dec 6 16:32:09 EST 2008


I think what John is saying is they are no good because I don't like  
them.
Unemployment estimates are estimates.  Correct.   But the national  
estimates are reliable to within 1/100th and proven so every ten  
years when we have a U. S. dicennial census.   If you don't like to  
hear what they say, sorry about that.   If the national rate moved  
from 2.3 to 2.2, maybe it moved and maybe it didn't.   If it moved  
from 2.3 to 2.1 or to 2.5, it did move and it moved in the direction  
published and with 0.1 of what was shown.   I was amazed in all the  
years I worked in Lancaster, the U. S. census came within 0.1 of the  
estimates for the average of March and April of 1960, 1970, 1980,  
1990 for the county too.   The census had a 100 percent count.    
Because all areas have to by BLS procedure add up to the nation,  
there will be aberrations in area data.   Statistics 101 ... the  
larger the area, the smaller the sample you need for reliability.   A  
one percent U. S. sample will give you 99.9 percent accuracy.   You  
would probably need a 20 percent sample in Lancaster County to get  
the same reliability.   Obviously then, of the area data, New York  
and Los Angeles and Houston probably are the most accurate while  
Wyoming is probably poor in comparison to New York.   After all,  
Wyoming has only as many people as Lancaster County.    In huge areas  
you can use the national sample and make it work by itself.   In  
small areas you need to start with unemployment claims and adjust  
them upward using national rations to accommodate people who have  
exhausted benefits and are no longer filing or people who have become  
discouraged, etc.  Obviously the
greatest data problems will be in Nevada, Montana, Wyoming, North  
Dakota but not in Philadelphia County or Cook County or Los Angeles  
County or the nation as a whole.   I should know John, I worked with  
those data for 30 years.    (Last year's national rate average was  
4.725%.   Assuming an accuracy to within .01, then it is accurate to  
between 4.71 and 4.73.    And we had 5,792,000 on an annual average  
unemployed ... give or take 5800 on an annual average basis.  I think  
that is fair accuracy.

Consumer price index.   Quite reliable.   But it doesn't show what  
prices do.   It shows what consumers are buying.  CONSUMER PRICE  
INDEX.   If consumers cut back on driving because the price of  
gasoline goes up or turn the thermostat down because oil goes up or  
change to chicken because beef goes up in price, then the CPI does  
not reflect the entire magnitude of those changes.  It reflects not  
price changes but what people are spending.   I suspect it also might  
show a down turn for the first time in years because of the magnitude  
of the gasoline and housing price drops.   Yes, I just looked ...  
those two components have dragged it down from September to October.   
The 1% drop in October was the worst since 1946.   The average of  
+0.23 so far this year is very low.  Another bad month could make it  
the lowest ever.  A couple of very bad months could make it almost  
zero making those people with CPI figured into union contracts very  
unhappy.   Those of us on social insecurity are not going to get much  
if anything in 2009.   If it doesn't change at all between now and  
the end of the year, it will go up about $30 a month ... but we know  
gasoline is continuing to drop.   Today is was $1.77 around the  
corner.   I can imagine social security almost flat from 2008 to  
2009.   (The flaw is the people on social security don't drive as  
much as people working; they are penalized.)


http://www.bls.gov/news.release/cpi.t01.htm

If you take this link and plug in 1947 and push go, you can get the  
monthly CPI changes since 1947.

http://data.bls.gov/PDQ/servlet/SurveyOutputServlet

If you wish to know what prices are doing, then look at wholesale  
price index.

In general, our federal statistical data are very good because the  
Bureau of Labor Statistics works independent of the politicians to  
keep it good.   The politicians may put their spin on it later and  
say the other guy was responsible for making it bad, but the data are  
decent most of the time.   That does not say, as in statistics 101,  
that you cannot have one time in ten that you have fluke data.    
National averages tend to iron out the month when you had a fluke.    
Remember that if the national unemployment rate falls within 0.1 nine  
months out of ten, then the nation rate will be within .01.



The problem with highway fatality numbers will be getting identical  
data from different countries.    We also have no clue if police are  
honest in reporting numbers.

Transit ridership numbers are bad because the agencies adjust them to  
get more money for their own use.   Some are very good like PATCO and  
Houston Metro.  Some, like SEPTA are incapable of being proven.   I  
think Baltimore needs a good powder keg to blow their numbers off the  
planet.   I fail to see how they can haul 85,000 a day on the  
subway ... look at the youtube pictures of the subway and all you see  
are empty trains.   And the light rail claims to haul about 30,000 a  
day and I see empty trains even on weekdays ... their numbers presume  
100+ people on every car in both directions at each end of the  
line.   No damn way.   I would bet some of those numbers are off by  
50 percent.

But don't assume all statistics are bad.




On Dec 6, 2008, at 3:06 PM, John Swindler wrote:

>
>
> A lot of statistics aren't worth the paper they are printed on,  
> such as consumer price index, unemployment,  highway fatalities and  
> transit ridership numbers.  All are nothing more then estimates  
> with a lot of assumptions.  But they can provide some means of  
> arriving at ballpark estimates.
>
> And this is nothing compared to some financial reports. (which also  
> aren't worth the paper they are printed on)  But best not go there.
>
> My favorite reference is an internal memo from a certain transit  
> authority that talks about a "heuristically derived formula" for  
> their ridership statistics.  And yes, I had to look that word up.
>
> John
>> Date: Sat, 6 Dec 2008 11:48:21 -0800> From: pcc_sr at yahoo.com>  
>> Subject: [PRCo] Re: Indian cultural differences> To: pittsburgh- 
>> railways at dementia.org> > Mr.Schneider;> > > This certainly puts it  
>> in perspective doesn't it.> > It has been stated that the U.S. has  
>> higher work place injuries as well within the last couple decades;  
>> I have not heard any comparisons recently. Relative to injuries  
>> one has to wonder about the criteria for reporting; this could  
>> vary considerably between countries, even locations inside  
>> borders.> > > Phil> > > > > > ----- Original Message ----> > From:  
>> Schneider Fred <fwschneider at comcast.net>> > To: pittsburgh- 
>> railways at dementia.org> > Sent: Saturday, December 6, 2008 11:28:33  
>> AM> > Subject: [PRCo] Re: Indian cultural differences> > > http:// 
>> www.driveandstayalive.com/info%20section/statistics/ 
>> statsmulticountry-percapita-2004.htm> > > > Most recent data I  
>> could find was 2004. "Close enough for govmint > > work."> > > >  
>> Yes, China killed 107,077 people on!
>   their roads in 2004 but that is > > out of a population of 1.3  
> billion people. India lost 90,000 out of > > a billion people. We  
> lost 42,636 out of 293.5 million.> > > > China had 8.26 fatalities  
> per 100,000 people,> > India had 8.33 and> > we had 14.53.> > > >  
> Surprisingly, the worst death rate was> > Russia with almost 25  
> deaths per 100,000 population.> > > > Germany, with a motor vehicle  
> registration rate about the same as > > ours but with people  
> driving about half as many miles per year as we > > do and with a  
> tremendous number of miles of express highways (very > > well  
> designed ones) with no speed limits outside of rural areas, has > >  
> a fatality rate half that of the U. S. A. > > Their rate is 7.09  
> per 100,000.> > > > > >
> _________________________________________________________________
> Send e-mail anywhere. No map, no compass.
> http://windowslive.com/Explore/hotmail? 
> ocid=TXT_TAGLM_WL_hotmail_acq_anywhere_122008
>






More information about the Pittsburgh-railways mailing list