Hi All,
I am doing pwmean testing with Tukey as a posthoc test for Anova and Kruskal-Wallis. In the example below I have weight data in % (sieve8pp) at three various seasons (Season_o2 variable): spring, summer and winter. AOV indicated the significant moisture difference among seaons (P=0.015), thus I did run this code:
pwmean sieve8pp if age=="market", over(Season_o2) mcompare(tukey) pveffects groups
From the output I see that the pairwise comparisons were significant between summer vs spring (P=0.035), and between winter and spring (P=0.022). The weight difference was not significant between winter and summer (P=0.66).
Thus I would mark the seasons this way:
summer 3.66 A
winter 5.11 A
spring 8.83 B
However, Stata provides letter A for summer and winter, and no letter at all for spring.
Should I go ahead and individually mark the variables when the letter is missing, but the posthoc tests' p value is significant?
Thank you for your time and effort!
Sincerely,
Gabriella
I am doing pwmean testing with Tukey as a posthoc test for Anova and Kruskal-Wallis. In the example below I have weight data in % (sieve8pp) at three various seasons (Season_o2 variable): spring, summer and winter. AOV indicated the significant moisture difference among seaons (P=0.015), thus I did run this code:
pwmean sieve8pp if age=="market", over(Season_o2) mcompare(tukey) pveffects groups
From the output I see that the pairwise comparisons were significant between summer vs spring (P=0.035), and between winter and spring (P=0.022). The weight difference was not significant between winter and summer (P=0.66).
Thus I would mark the seasons this way:
summer 3.66 A
winter 5.11 A
spring 8.83 B
However, Stata provides letter A for summer and winter, and no letter at all for spring.
Should I go ahead and individually mark the variables when the letter is missing, but the posthoc tests' p value is significant?
Thank you for your time and effort!
Sincerely,
Gabriella
