Ttest results, rounding and missing values

[Ozer Erdem]

#17607

Hi Dr. Shah,

I am trying to understand why some of the values are not showing in my ttest results.
Here is the code that I use

asdocx ttest  hnr if Sentiment >0, by(Sentiment) ///
stat(obs mean se sd dif t p) c2_tzok c2_dec(0) /// 
c3_tzok c3_dec(0) c4_tzok c4_dec(10) c5_tzok ///
c5_dec(10) c6_tzok c6_dec(10) c7_tzok c7_dec(10) ///
c8_tzok c8_dec(10) c9_tzok c9_dec(10)

Here is an example:

I expect to see the following values under the mean and stdev in the t test results. I am pasting the results from part of ttest table for two variables.

Var -  hnr
Two-sample t test with equal variances
------------------------------------------------------------------------------
   Group |     Obs        Mean    Std. err.   Std. dev.   [95% conf. interval]
---------+--------------------------------------------------------------------
     Pos | 3557765   -.0000301    2.89e-07    .0005454   -.0000306   -.0000295
     Neg | 608,190    -.000026    6.63e-07    .0005172   -.0000273   -.0000247

Var -  jitter

Two-sample t test with equal variances
------------------------------------------------------------------------------
   Group |     Obs        Mean    Std. err.   Std. dev.   [95% conf. interval]
---------+--------------------------------------------------------------------
     Pos | 3557765    .0257654    3.36e-06    .0063411    .0257588     .025772
     Neg | 608,190    .0255683    8.12e-06    .0063336    .0255524    .0255842

However, the results do not have those. It looks like if the values is very small, it does not get copied. I run for several variables and I get the same results for small values.

T-Test of differences for all AudioFeatures

	obs1	obs2	Mean1	Mean2	St_Err	St_Dev1	St_Dev2	dif	t_value	p_value
hnr	3557765	608190	0.0000000000	0.0000000000	0.0000000000	0.0005000000	0.0005000000	0.0000000000	-5.4	0
Jitter loc	3557765	608190	0.0260000000	0.0255000000	0.0000000000	0.0065000000	0.0065000000	0.0000000000	22.4	0
Notes:

From the results, it looks like, I am not getting mean, sterror and diff values for the hnr and stdev values seems rounded.
I am getting same results for the second variable, except I am getting rounded mean values.

Thank you.

[Attaullah Shah]

#17608

Ozer Erdem
I have updated asdocx. It now fixes the decimal points problem in the ttest results. You should now be able to see accurate decimal points in your results.

Additionally, for very small values, using the c#_disfmt(%9.4e) format is a better way to report values in exponential format. For example, instead of seeing 0.00000009 , you would see 9.00e-08. This helps in making the results easier to read and understand.

asdocx_update 

Here is an example:

 * Create a dummy dataset
clear
input float group double score
1 .0000012
1 .000015
1 .0000214
2 .011
2 .039
2 .0610
end

* Report 7 decimal places for the column 4
asdocx ttest score,  replace  c4_dec(7) by(group) stat(obs mean t)
                                 Table: Results
  0 |1                             2          3              4          5           6 
----+------------------------------------------------------------------------------------------
  1 |                          obs1       obs2          Mean1      Mean2      t_value 
----+------------------------------------------------------------------------------------------
  2 |score by group: 1 2           3          3      0.0000125      0.037      -2.556 
-----------------------------------------------------------------------------------------------

* Report exponential format for column 4
asdocx resetfmt

asdocx ttest score,  replace  tzok c4_disfmt(%9.4e) by(group ) stat(obs mean t)

Two-sample t test with equal variances
------------------------------------------------------------------------------
   Group |     Obs        Mean    Std. Err.   Std. Dev.   [95% Conf. Interval]
---------+--------------------------------------------------------------------
       1 |       3    .0000125    5.96e-06    .0000103   -.0000131    .0000382
       2 |       3        .037    .0144684    .0250599   -.0252523    .0992523
---------+--------------------------------------------------------------------
combined |       6    .0185063     .010501     .025722   -.0084873    .0454999
---------+--------------------------------------------------------------------
    diff |           -.0369875    .0144684               -.0771581    .0031831
------------------------------------------------------------------------------
    diff = mean(1) - mean(2)                                      t =  -2.5564
Ho: diff = 0                                     degrees of freedom =        4

    Ha: diff < 0                 Ha: diff != 0                 Ha: diff > 0
 Pr(T < t) = 0.0314         Pr(|T| > |t|) = 0.0629          Pr(T > t) = 0.9686


   
                                   Table: Results
  0 |1                             2          3             4          5           6 
----+----------------------------------------------------------------------------------------
  1 |                          obs1       obs2         Mean1      Mean2      t_value 
----+----------------------------------------------------------------------------------------
  2 |score by group: 1 2       3.000      3.000      1.25e-05      0.037      -2.556 
---------------------------------------------------------------------------------------------
Notes:

[Ozer Erdem]

#17629

Very helpful.
Thank you Dr. Shah.

Ozer

[Author]

Posts