Re: We have the pitching, what about offense?

On Wed, 21 Jan 2009 19:40:00 -0800 (PST), jvampatella@xxxxxxxxx wrote:

On Jan 21, 5:58 pm, McDuck <wallyDELETEMEMcD...@xxxxxxxxxxx> wrote:


<sigh> Here's a quote by an engineer about reverting to the mean:
"As an engineer (well, I got the degree anyway) I've always enjoyed
science and mathematics. One of my favorite expressions - revert to
the mean - has to do with systems or processes that swing outside
expected boundaries. Generally speaking, a series over a long period
of time may experience highs and lows, but usually settles back at the
mean." (from http://workingsmarter.typepad.com/my_weblog/2005/11/revert_to_the_m.html)

The guy is not addressing any kind of statistical issue. And Engineers
are not statisticians. and this guy "may" only have an undergrad
degree in engineering --- does not claim ANY expertise even in his
subject. so why is he an authority? If you doubt what I said, I
suggest you look it up in a statistics book. Lots of them on line, by
the way.


Note the "long period of time" and systems that "swing outside
expected boundaries". We don't know that Pedroia's 2008 year was
"outside expected boundaries". You're a lawyer, not a statistician,
and please don't claim any level of expertise here. I'll take the
word of a mathematician and engineer in this area over yours, thanks.

Fine, I do not claim to be a mathematician, although my undergraduate
degree is in mathematics. I certainly am not an expert on statistics,
although I took a course in it in college and have used/read stats
articles in my work. But, believe me, John, I'm telling you really,
really elementary stuff. I would not pretend to be good enough in
statistics to do any serious statistical work. Indeed, when I have a
serious issue, I go to my sister, who has her PhD in the subject,
wrote a textbook on it, and taught it for many years.


(Also, you apparently do not understand what "sample size" means, not
that it matters much. We are looking at all of Pedroia's ABs in the
majors, not some sample --- stated differently, our sample is 100% of
the cases --- can't get any bigger sample than that.)

LOL. That's a ludicrous argument. Of course our sample size is 100%
of his at-bats, but that doesn't mean it's not a small sample size and
that it "can't get any bigger" than that. Just plain stupid.

John, you apparently do not have ANY idea what sample size means.
Just admit it. It is really dumb of you to call me stupid when you are
making a very elementary error in statistics. Read a book or something
or ask some math person you trust. Anyone, and I mean anyone, who
knows anything on the topic will tell you that you have made a
mistake.

If he
only had five at-bats in the majors, and we looked at all five, that's
100% of the cases, right? Can't get any bigger sample than that?
It's not the percentage of cases we're looking at, it's the raw number
of cases.

Yes, the above is true, which is why I suggested your "sample size"
error was not all that important. The raw number of cases is not the
sample size.


Regression to the mean when you're looking at 10 at-bats is
pointless...it really is.

Correct. Also correct that the stats need to have predictive value for
any claim about regression to the mean to have significance, as I
explained in response to your HR comment. The fact that you stumble
around and get some stuff right, however, does not mean you understand
the stat issue.

And yet that may be 100% of the available
data.

Now you are getting somewhere. The problem is the size of the data
set, not the sample size. Look it up. You should not give me lectures
on statistics when you cannot get even the basic terminology right.

Of course you can still do the math and it probably comes out
correct, but the context of this discussion is trying to predict what
Pedroia will do next year. Five years' worth of at-bats is a hell of
a lot bigger sample size than two years and change. You're really,
really reaching here.

I think you mean bigger data set, not bigger sample size. You got it
right above, and then reverted to your old error. It happens. Don't
get discouraged.

No one has a good way of translating college stats into MLB stats, so
those lower-level stats are nearly worthless for the current purpose
(but good at predicting he's be a good hitter in MLB. The minor league
stats somewhat less worthless, but not very helpful for the matter at
hand, since the translations are really, really rough. No, the only
data of much relevance for predicting whether Pedroia would do better
or worse in 2009 are the MLB stats from 2006-2008.

As I explained in greater detail to Dano, and to you in my Ted
Williams post, we can predict a reversion to the mean with confidence
when the aberration from the mean is very large, like hitting .400.
Still, having an MVP year is obviously a pretty significant deviation
from the mean. Pedroia did not got a single vote for MVP in 2007, so
2008 was in some ways a big step up.

The increase in SLG from 2007 to 2008 was "only" around 50 points,
with an insignificant drop in OBP. Not a compelling case for reversion
to a lower SLG number, although a drop is somewhat more likely than
an increase or no change. That is why I suggested that we should not
be expecting better hitting from Pedroia but should not expect a
significant drop as well, based solely on reversion to the mean. I do
not know enough or have enough data to compute the odds of Pedroia
dropping down --- my rough estimate is that the odds are positive but
not large.

I also noted that there is still a reasonable chance, which I have no
way of quantifying, that Pedroia will improve his hitting in 2009, but
that is somewhat less likely than a small decrease, based on his past
record. Again, I do not have the data to quantify the matter, nor do I
know enough statistics to do it. (One advantage I have over you from
studying statistics in college is that I have some rough idea of what
I actually do not know.)

If you have some way of developing a stat argument for an increase in
hitting from Pedroia for 2009, fine. I don't know how to do it. I do
know that you do not do it by trotting out his college or minor league
stats or by telling us that he will be 26 in July of 2009. As I noted
for Ted, he regressed to the mean from his fantastic 1941 year even
though his 1942 year was terrific by any sensible measure.

If there were stats showing that hitters from age 25 to 26 increased
SLG from x to y and did not reduce OBP, then you'd have the start of a
decent stat argument for an improvement by Pedroia in 2009.
.

Relevant Pages

  • RE: How to link tables via the field names themselves, not their c
    ... you can query the database to do what you want easily. ... There are approximately 30 different aspects (or statistics) which are ... Beacuse there are so many stats per contract, ...
    (microsoft.public.access.queries)
  • RE: How to link tables via the field names themselves, not their c
    ... you can query the database to do what you want easily. ... There are approximately 30 different aspects (or statistics) which are ... Mulitple audits are made per contract. ... Beacuse there are so many stats per contract, ...
    (microsoft.public.access.queries)
  • Re: Just Arod being Arod
    ... they will lead to better or worse statistics. ... Player B has very little heart, guts, and nerve. ... ARod has a hell of a long way to go to match Reggie in Yankee lore ... pitches, it's in his stats. ...
    (alt.sports.baseball.ny-yankees)
  • Re: [Bug 1806] New: disks stats not kept for DM (device mapper) devices
    ... This appears to be due to the fact that the stats are traced via ... > request structs which are not generated until below the device mapper ... > (only the BIO) it has no place to record the start time. ... how I/O statistics are managed. ...
    (Linux-Kernel)
  • Re: Intermittent crackling line problems
    ... Intermittent and inconsistent crackling. ... disappears whenever engineer calls. ... way I would tackle it would be to put an oscillator (tone) on your ... onward stampede for more stats, & higher stats normally the guys will just ...
    (uk.telecom)