Re: WHO STOPS HIM?


The way to stop him is to stop making the errors which he
finds.

You should thank him for taking the time to read your stuff.

Most readers of this newsgroup just skip over your "stuff".
Those of us who do so marvel at his patience in dealing with
your "stuff".

The usual way of dealing with "internet personalities" is to
"stop feeding the troll". Unfortunately some may not recognize
the content for what it is so we are thankful for the corrections
even if less than happy with the traffic.



On 2007-12-14 08:15:07 -0400, "Luis A. Afonso" <licas_@xxxxxxxxxxx> said:

Every time I wrote a program intending to evaluate Critical values by simulation there are a precise idiot (Jon Smith) that have something to criticize.



*** Date: Dec 9, 2007 3:52 PM
Author: John Smith
Subject: Re: Heâ??s not John Smith a * burro * heâ??s not?

Luisa,

I did not ask about 10% critical values. I asked about the 1% critical values. Your 1% simulation results are non-monotonic, and that your critical value for 1% when n=50 (12.46) is greater than the asymptotic value (9.21). Explain THIS!! IF you cannot then you are a fraud. John ***Notice that NONE of Lawford's results are non-monotonic in that region. Clearly your code is wrong. John ***

A COMMENT

Everyone knows John Smith IS A COMPULSIVE LIAR.
Besides e knows nothing in Statistics in general and Monte Carlo simulations in particular.
He thinks that to state a liar is sufficient to be believed, a mental dysfunction that should be followed closely by the respective experts.
If he reads the Lawford´s paper he surely DOESN`T FIND ANY TABLE and HE SIMPLY LIES when he states that for 1% (NOT CRITICAL ad he wrote but SIGNIFICANCE LEVEL) the overshot 12.46 is not present.
By the way: because the program calculate SIMULTANEOUSLY for the 5%, 1% and 10% it is impossible to be wrong as John states but is unable to say WHERE AND WHY.
No one believes that the same program can provide GOOD AND WRONG values at he same time.


My values are:


*** Date: Mar 7, 2007 5:55 PM
Author: Luis A. Afonso
Subject: Re: Jarque-Bera test: confidence intervals for normal data

Jarque-Bera test: confidence intervals for normal data, 2

RAW RESULTS

_SIZE_________5%___________1%______
__10______2.54__2.53_____5.72___5.70__
__15______3.30__3.29_____8.26___8.18__
__20______3.81__3.79_____9.73___9.66__
__25______4.16__4.16____10.78__10.74__
__30______4.40__4.41____11.33__11.27__
__35______4.57__4.59____11.67__11.80__
__40______4.74__4.77____11.95__12.01__
__45______4.86__4.88____12.29__12.19__
__50______4.95__4.95____12.36__12.46__

(1 million of samples, size N, 2 times, for each sample size). Remembering that the theoretical, asymptotical, Chi-squared, 2 DF, are 5.991 (5%) and 9.210 (1%) this calculation´s utility seems unquestionable.***


It follows a Table can be read from Thorsten Thadewald & Herbert Bunning
___Jarque - Bera Test and the Competitors for Testing Normality - A Power Comparison___________

(Excerpt of Table 5)

_alpha%___10____20_____50_____100____200____500
___1____5.738__9.458__12.331__12.296__11.750__10.601
___5____2.535__3.768___5.004___5.448___5.728___5.825
__10____1.618__2.335___3.192___3.643___4.081___4.324
___1___________________________________inf:___9.210
___5________________________________________5.991
___10_______________________________________4.605

The INNACETTABLE overshoot was well fond by the COLLEAGUES RESEARCHERS.
CLERK John Smith was berrer to him to be quiet.

_________

Luis Amaral Afonso









REM "JB"
CLS
DEFDBL A-Z
INPUT " N = "; n
LOCATE 1, 15
PRINT " CHI2 , 2DF ";
PRINT " 4.605 5.991 9.210"
PRINT " 10% 5% 1% "
DIM x(n)
DIM JB(8004)
pi = 4 * ATN(1): ju = 1
DEF fnJB (S, K, n) = (n / 6) * (S ^ 2 + (K - 3) ^ 2 / 4)
all = 400000
ball = all / 10
FOR ji = 0 TO 8004: JB(ji) = 0: NEXT ji
FOR v = 1 TO all
LOCATE 2, 50
PRINT USING "##########"; all - v
RANDOMIZE TIMER
S1 = 0
uu = 0: uuu = 0: uuuu = 0
FOR i = 1 TO n
a = SQR(-2 * LOG(RND))
x(i) = 1 * a * COS(2 * pi * RND)
S1 = S1 + x(i)
NEXT i
FOR ii = 1 TO n: d = x(ii) - S1 / n
uu = uu + d * d / n
uuu = uuu + d * d * d / n
uuuu = uuuu + d * d * d * d / n
NEXT ii
S = uuu / (uu ^ (3 / 2))
K = uuuu / (uu * uu)
z = fnJB(S, K, n): z = INT(z * 100 + .5)
IF z > 8000 THEN z = 8000
IF z < 0 THEN z = 0
JB(z) = JB(z) + 1
IF INT(v / ball) <> v / ball THEN GOTO 40
vc(0) = .9 * v: vc(1) = .95 * v: vc(2) = .99 * v
FOR cc = 0 TO 2
S = 0: vv = vc(cc)
FOR jj = 0 TO 8000
S = S + JB(jj)
IF S > vv THEN GOTO 20
NEXT jj
20 REM
LOCATE ju + 2, 20 * cc + 1
aa(ju, cc) = jj / 100
PRINT USING " ##.## #.#### "; aa(ju, cc); S / v;
NEXT cc
PRINT USING "############"; v
ju = ju + 1
40 NEXT v: PRINT
END


.