Dear SAS Community,
I would greatly appreciate it if you could give me some feedback on the best approach to analyze data skewed to the left.
Context: I would like to know if there are significant differences in the dependent continuous variable firmness (AvgFirm) between two avocado varieties (BL516 vs Hass) after 0, 1, 3, and 6 weeks of storage for each harvest month as you can see it in these graphs. I made these graphs using the mean but I guess that is not a good idea.
When I plot the data using the following code I get the following distributions for each week, where you can see that for week 6 the data is skewed to the left.
proc univariate data=one normal plot; where Variety in('BL516', 'Hass'); var AvgFirm; histogram AvgFirm/normal (color=red); class Wks; run; quit;
The UNIVARIATE Procedure
Wks = 6
Fitted Normal Distribution for AvgFirm
Parameters for Normal Distribution
Parameter
Symbol
Estimate
Mean
Mu
9.49237
Std Dev
Sigma
9.399897
Goodness-of-Fit Tests for Normal Distribution
Test
Statistic
p Value
Kolmogorov-Smirnov
D
0.2351068
Pr > D
<0.010
Cramer-von Mises
W-Sq
3.3803286
Pr > W-Sq
<0.005
Anderson-Darling
A-Sq
17.9411189
Pr > A-Sq
<0.005
Quantiles for Normal Distribution
Percent
Quantile
Observed
Estimated
1.0
1.90000
-12.37506
5.0
2.60000
-5.96908
10.0
3.03000
-2.55408
25.0
4.20000
3.15224
50.0
6.23000
9.49237
75.0
10.10000
15.83250
90.0
22.75000
21.53882
95.0
28.00000
24.95382
99.0
51.20000
31.35980
For this reason I would analyze the data for each week separately using genmod. What dist= and link= function do you think it would be a good approach to model left skewed data effectively?
proc genmod data=one; where Wks=6 ; class Variety Harvest; model AvgFirm=Variety|Harvest/type3 dist= link=; slice Variety*Harvest/sliceby=Harvest/ adjust=simulate(seed=1); run;
I would greatly appreciate it your help!
Thanks
... View more