Hi all,
I am trying to get alternative Hypotheses in a contrast post regresion to get similar ttest results, I wonder if its posible:
Ha:diff < 0
Ha: diff > 0
In my example I show ttest results and contrast to get the same results: The signs for the difference, the test statistic, and the confidence intervals are reversed because
the difference is taken in the opposite direction, also I tried with pwmean test:
Thanks in advance,
Regards
Rodrigo
I am trying to get alternative Hypotheses in a contrast post regresion to get similar ttest results, I wonder if its posible:
Ha:diff < 0
Ha: diff > 0
In my example I show ttest results and contrast to get the same results: The signs for the difference, the test statistic, and the confidence intervals are reversed because
the difference is taken in the opposite direction, also I tried with pwmean test:
Code:
* Example generated by -dataex-. For more info, type help dataex
clear
input byte(fertilizer irrigation) float(yield N03_N)
1 0 25.5471 9
1 0 26.9514 7.5
1 0 30.42061 8
1 0 32.142567 8
1 0 33.476627 9.5
1 0 35.112484 8.5
1 0 35.643112 9
1 0 36.786076 10.5
1 0 37.020485 9.5
1 0 37.382446 10
1 0 37.633434 8
1 0 38.24033 10
1 0 38.47208 9
1 0 39.00444 9.5
1 0 39.03085 8.5
1 0 39.46735 9
1 0 42.28565 8.5
1 0 42.9318 9
1 0 44.10148 9.5
1 0 46.60105 10.5
1 1 35.594612 10
1 1 36.84657 9
1 1 39.16208 8.5
1 1 39.64972 10.5
1 1 41.53548 10.5
1 1 41.65326 10.5
1 1 42.89606 9
1 1 43.16227 10
1 1 43.19808 10.5
1 1 46.89809 10.5
1 1 46.99522 9
1 1 47.37473 10
1 1 48.75241 9
1 1 48.78089 11
1 1 49.3729 10
1 1 49.88928 10.5
1 1 50.06085 10
1 1 50.10778 10
1 1 53.88142 10.5
1 1 60.4341 11
2 0 30.33382 8
2 0 33.649242 5.5
2 0 33.986282 6
2 0 34.199596 6
2 0 34.30233 7.5
2 0 34.330006 5.5
2 0 34.98474 7.5
2 0 35.485172 8
2 0 35.540104 6
2 0 36.922596 6.5
2 0 41.4649 8
2 0 41.47297 8.5
2 0 41.63222 7
2 0 41.70517 9.5
2 0 42.54293 9
2 0 42.77414 8
2 0 43.02679 8
2 0 43.03543 7.5
2 0 46.14563 10.5
2 0 48.36243 11
2 1 41.1847 9
2 1 42.89641 9
2 1 45.26883 8.5
2 1 45.53941 7
2 1 45.77157 10.5
2 1 46.79755 9.5
2 1 49.55231 8.5
2 1 49.87773 10
2 1 50.73264 9
2 1 51.20868 11
2 1 51.83252 10.5
2 1 52.63506 10.5
2 1 52.82097 9
2 1 53.07209 8.5
2 1 55.70432 11
2 1 56.25944 10.5
2 1 56.64001 12
2 1 57.53093 9
2 1 58.14713 11
2 1 60.03719 12.5
3 0 28.385687 12.5
3 0 31.1877 13
3 0 31.540403 13
3 0 32.954784 12.5
3 0 33.825752 13.5
3 0 34.072716 14
3 0 35.216423 14.5
3 0 35.538937 14
3 0 35.55414 13.5
3 0 35.731876 14
3 0 38.014694 15
3 0 38.35717 15
3 0 38.37197 14.5
3 0 38.76787 14.5
3 0 39.05844 14.5
3 0 39.13557 15
3 0 39.53992 14
3 0 40.09854 14.5
3 0 40.11148 15
3 0 41.41248 16
end
label values fertilizer fertlbl
label def fertlbl 1 "10-10-10", modify
label def fertlbl 2 "10-08-22", modify
label def fertlbl 3 "16-04-08", modify
*****************************************************************
*ttest test
. ttest yield, by(irrigation)
Two-sample t test with equal variances
------------------------------------------------------------------------------
Group | Obs Mean Std. err. Std. dev. [95% conf. interval]
---------+--------------------------------------------------------------------
0 | 100 37.91277 .5300607 5.300607 36.86102 38.96453
1 | 100 47.93122 .5630353 5.630353 46.81403 49.0484
---------+--------------------------------------------------------------------
Combined | 200 42.92199 .5242462 7.413961 41.8882 43.95579
---------+--------------------------------------------------------------------
diff | -10.01844 .7732872 -11.54338 -8.493509
------------------------------------------------------------------------------
diff = mean(0) - mean(1) t = -12.9557
H0: diff = 0 Degrees of freedom = 198
Ha: diff < 0 Ha: diff != 0 Ha: diff > 0
Pr(T < t) = 0.0000 Pr(|T| > |t|) = 0.0000 Pr(T > t) = 1.0000
********************************************************************
*post regresion test
qui reg yield i.irrigation
contrast i.irrigation, effect nowald
Contrasts of marginal linear predictions
Margins: asbalanced
------------------------------------------------------------------------------
| Contrast Std. err. t P>|t| [95% conf. interval]
-------------+----------------------------------------------------------------
irrigation |
(1 vs base) | 10.01844 .7732872 12.96 0.000 8.493509 11.54338
------------------------------------------------------------------------------
*******************************************************************************
*pwmean test
. pwmean yield, over(irrigation) effects
Pairwise comparisons of means with equal variances
Over: irrigation
------------------------------------------------------------------------------
| Unadjusted Unadjusted
yield | Contrast Std. err. t P>|t| [95% conf. interval]
-------------+----------------------------------------------------------------
irrigation |
1 vs 0 | 10.01844 .7732872 12.96 0.000 8.493509 11.54338
------------------------------------------------------------------------------
Regards
Rodrigo
Comment