Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: st: RE: significance of mean and median


From   "Martin Weiss" <[email protected]>
To   <[email protected]>
Subject   RE: st: RE: significance of mean and median
Date   Mon, 24 Nov 2008 15:15:42 +0100

Line for the server...

You probably wanted to say that "the CI did not overlap zero". Regarding the
proper interpretation of statistical tests, you should consult any
undergraduate statistics textbook. The 5% level means that your rejection of
H0 will be wrong in 5% of the times you conduct this test. Note the
parallels between -mean- and -ttest-

*******
sysuse auto, clear
replace mpg=mpg-20
ttest mpg=0
mean mpg
******

-ttest- output also gives you one-sided tests. Your question regarding
whether you can conclude that the mean is negative is answered by the
leftmost output of -ttest-. 

Regarding your -signtest-, do not try to reinvent the wheel. Stata provides
a whole submenu of regression diagnostics which can be employed to
discriminate between two competing models...


HTH
Martin


-----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of Bastian Steingros
Sent: Monday, November 24, 2008 3:02 PM
To: [email protected]
Subject: Re: st: RE: significance of mean and median

Hi Martin,

Using *mean* for several variables, stata has provided result like
mean_var_x=-0.056 and the upper end of the CI did not overlap zero. Given a
95% CI, does the result indicate that the mean of var_x
is negative at the 5% level (1% when used a 99% CI)?

Is signtest not appropriate to compare the median inaccuracies of model 1 +2
or more generally to the medians of two variables?

b



-------- Original-Nachricht --------
> Datum: Mon, 24 Nov 2008 13:46:42 +0100
> Von: "Martin Weiss" <[email protected]>
> An: [email protected]
> Betreff: st: RE: significance of mean and median

> 
> Line for the server...
> 
> On the second part, without judging the appropriateness of your procedure,
> "(GDP_predicted - GDP)" looks like a straightforward residual to me,
> available from -predict-, which would generate a new variable, ready to be
> divided by "GDP" with -replace-. Afterwards -ttest- compares means and
> -median- compares medians...
> 
> HTH
> Martin
> 
> 
> -----Original Message-----
> From: [email protected]
> [mailto:[email protected]] On Behalf Of Bastian
> Steingros
> Sent: Monday, November 24, 2008 12:56 PM
> To: [email protected]
> Subject: st: significance of mean and median
> 
> Hello everybody,
> 
> I have this problem:
> 
> After creating the descriptive statistics of my sample I want to show that
> the mean of a certain variable is statistically significant.
> For example the mean of var1 is 0,45. Now I want to show that the positive
> sign is significant.
> Is ttest the appropriate way to do so?
> 
> Moreover, in another sample I want to predict the GDP using two different
> regression models.
> Finally, I want to compare the predicted values with the real GDP, that
> is,
> I will compare the mean and the median of the inaccuracy. 
> Inaccuracy is defined as (GDP_predicted - GDP)/GDP
> 
> Is there a way to show that the median (mean) inaccuracies of the two
> models
> are statistically different from each other?
> 
> may be *signtest*???
> 
> thanks
> 
> bastian s
> -- 
> Sensationsangebot nur bis 30.11: GMX FreeDSL - Telefonanschluss + DSL 
> für nur 16,37 Euro/mtl.!* http://dsl.gmx.de/?ac=OM.AD.PD003K11308T4569a
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/statalist/faq
> *   http://www.ats.ucla.edu/stat/stata/
> 
> 
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/statalist/faq
> *   http://www.ats.ucla.edu/stat/stata/

-- 
Sensationsangebot nur bis 30.11: GMX FreeDSL - Telefonanschluss + DSL 
für nur 16,37 Euro/mtl.!* http://dsl.gmx.de/?ac=OM.AD.PD003K11308T4569a
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/


*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/



© Copyright 1996–2024 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index