[FoRK] Bar Stool Economics

Jeff Bone jbone at place.org
Sat Oct 2 05:44:10 PDT 2010

On Oct 1, 2010, at 15:29 , Aaron Burt wrote:

>> Or, indeed, is meaningful at all?
> I say it is.  Give me something better or stop squirming.

Fine, I will hold your hand.

>> And third, you might want to consider actually fiddling with the controls
>> on the form in question.  For one thing, it DOES provide net numbers, to
>> any given state for all states for any interval within the timeframe.
> Yep, that's what I used.

Really?  Why all the machinations, then?  Did you just click the "Net Migration" radio control or not?  If you had, you wouldn't have needed to do anything further.

> I'm working from the data and thesis espoused in the blog postings you were
> expounding on and I was arguing against.  

I was looking at the data those blog postings were working with (third link) which they were correctly summarizing and interpreting, albeit in a relatively simple fashion (as they are blog posts, of course, not submissions to some policy journal.)  To be clear, we're looking at this:


>> If you want a meaningful comparison, you might want to look at
>> state-to-state comparisons in absolute numbers, net. <snip>
> That's what I did.

Apparently not.

>> Bottom line, while Texas grew by a net gain of 362,494 returns during the
>> period, California shrank by 723,098.  For example.
> Interesting examples, as Califoria and Texas are the 2 most populous states
> in the US.  

Ah, and therein lurks the lesson.  The ONLY reason comparing the percentage state loss / gain is meaningful IN THAT PARTICULAR CASE is precisely because they are *ROUGHLY* close in size.  But *roughly* is only meaningful in to whatever extent you are willing to consider them *the same size.*

Comparing the percentage loss / gain of California to e.g. that of Rhode Island is *meaningless* for hopefully obvious reasons.

Consider this:  if you added up all the state population percentage gains and losses, what would you get?

If you want a meaningful percentage comparison, you'd need to use *the sum of all the state percentage changes* as your denominator.

Beware percentages when doing this kind of thing.  Make sure you're comparing apples to apples.

> BTW, one of us is reading the inflow/outflow numbers backward.

Yes, indeed you are.  Let me help:

"(For example, suppose you chose Alabama. A value of 100 (positive) in the "Net # Tax Returns Inflow" column for the Alaska row means that 100 more tax returns moved from Alaska into Alabama than vice versa.)"

Got that?  For the target state selected, positive means positive net inflow from the source states in the table to the target state.  While it doesn't state it explicitly, a 2nd grader should recognize that the opposite condition must also hold:  a negative number means negative net inflow (i.e., outflow) from the source states in the table to the target state.

So, for example, let's take California for example.  Let's substitute California for Alabama in the above:

"For example, suppose you chose California. A value of 100 (positive) in the "Net # Tax Returns Inflow" column for the Alaska row means that 100 more tax returns moved from Alaska into California than vice versa."

Got that?  You pick the target state in the drop down and pick some interval --- we'll just leave the timeframe set to the default, that's probably an advanced concept for this exercise.  Then, the table shows you a bunch of positive or negative numbers, one row / number per state listed on the left of the table.  To determine the net flow between Alaska and California, you look up the row that has Alaska in the left column.  You find the number next to it.  Oh, looky looky, it's negative.  In fact it is -5,200.  That means -5200 people moved FROM Alaska TO California, i.e., 5,200 people moved FROM California TO Alaska.  Now just scan down that column.  Look at all those negative numbers.  What a shame. 

> Ah, so my numbers ARE meaningful?  

Nope, not even a little bit.

> * A good reason to re-do an analysis that nobody would pay attention to,
>  since it doesn't support an ideology.

Oh, analysis needs to support *ideologies.*  What an interesting concept, thanks.  I suppose data can also be classified as good or evil.

> They're not Murdoch property.  They are from Planet Murdoch.  They play to
> the Murdoch worldview, using Murdoch rhetoric to support Murdoch ideology.

How nice it must be to live in such a simple place as Planet Burt, where everything is either red or blue, where green (or anything right of, say, yellow, presumably) originates on something called "Planet Murdoch", where negative is positive, where apples and oranges are the same thing, and where nobody is allowed to think for themselves except Aaron --- who apparently can't read or do math but isn't bothered by that fact even a little bit.

Yes, that's sarcasm.  You make me despair for humanity.

$10.50 for not even being able to read the instructions for using a fucking web form properly.  That, and inflation.  (Hey, but DJIA is still over $10k!  Everything MUST be ok!)


More information about the FoRK mailing list