In: Statistics and Probability
1. A poultry company undertook an experiment to determine the
effect of certain treatments on chicken growth. A sample of 24
young chicks were given one of three possible treatments to aid
their growth. In order of cost from least to most costly, the three
treatments were as follows: a cheap food; an expensive food; the
expensive food plus an injection of a growth serum. After a fixed
period of time, the application of the treatments was halted and
the chickens were weighed. The response was their increase in
weight (in pounds) over the time period of the experiment. Because
of space considerations, the experiment had to be carried out in
four different laboratories. These laboratories were randomly
selected and labeled as “North”, “South”, “East” and “West”. Each
laboratory employed a different technician to carry out the
experiment. Six chickens were assigned to each laboratory, and the
three treatments were randomly assigned to the chickens within each
laboratory (two chickens having each treatment within each
laboratory). In your mini-report, explain why it makes sense to
design and analyze this as a block design. Using the sample data,
formally test whether it turned out to be sensible to use blocks,
providing numerical justification (test statistic and P-value) for
your conclusion. Also formally assess the main research question of
interest: whether the three treatments yield significantly
different mean growth, again providing numerical justification
(test statistic and P-value) for your conclusion. If you find a
difference across the treatments, investigate exactly which
treatments differ significantly from each other. Provide
recommendations to the poultry company about their best practices
based on your findings.
SAS code:
/* Problem 1 */
DATA one;
INPUT lab $ treatment :$12. growth;
cards;
East Cheap 4.62
East Cheap 4.93
East Expens 6.25
East Expens 5.97
East Expens_Serum 6.41
East Expens_Serum 6.54
West Cheap 5.36
West Cheap 5.49
West Expens 6.53
West Expens 6.62
West Expens_Serum 6.81
West Expens_Serum 6.63
North Cheap 4.65
North Cheap 5.32
North Expens 5.96
North Expens 6.12
North Expens_Serum 6.33
North Expens_Serum 6.44
South Cheap 5.76
South Cheap 6.32
South Expens 6.47
South Expens 6.68
South Expens_Serum 6.71
South Expens_Serum 6.78
;
run;
Output:
Where alpha(level of significance) = 0.05.
From Anova table:
The lab F-statistics = 14.31 and P-value = 0.0003, it is means in lab have significant difference.
The treatment F-statistics = 79.54 and P-value = <0.0001, it is mean in treatment have significant difference.
But In Interaction lab*treatment F = 2.22 and P-value = 0.1131, it is mean there is no significant difference.
Multiple Comparison for Lab:
Which means covered by bar it means there is no significant difference.
Now see who are significant:
There is a significant difference between South and North, South and East.
There is a siginificant difference between West and North, West and East.
Multiple Comparison for Treatment:
Which means covered by bar it means there is no significant difference.
Now see who are significant:
There is a significant difference between Expens_Serum and Cheap.
There is a siginificant difference between Expens and Cheap.