Professional Documents
Culture Documents
Nicolette Casselli
3/22/2011
Introduction:
population in Flanigan, Nevada and Beaty Butte, Oregon. In the experiment, they took the
adult male horses, figured out the dominant male in the herd, and sterilized it by giving it a
vasectomy. There were groups that had sterilized males, and groups that did not. In the
following two years, they recorded the number of foals born in those herds to determine if the
sterilizations worked.
To help verify whether this sterilization treatment worked, I ran a series of tests on the
data set. I first had to establish whether or not the populations were normal, or close to
normal. I then took my results of normality and continued with testing. The F-test that was
used helped determine which T-test to look at and if the variance within the populations were
enough to test. The T-tests helped in finding whether the means of these populations were the
same, and further showed which populations was larger or smaller for the appropriate
situation. The tests could help conclude the effectiveness of the sterilization treatments on
these horse populations. Below are the results and methods used and further analysis can be
Methodology:
In order to determine whether or not this experiment worked, there were several other
questions that needed to be answered. By using SAS analyst, these questions were answered.
First, I took the variables of location and adults to see if the location of the adult populations
had any effect on the populations within the horses. Many outside elements could affect
population, not just the fact that these horses had been sterilized. This is why the question of
location could be important. After determining population normality with the graphs and the
Proc Univariate, I conducted an F-test and a T-test on the information. Another question that
could be raised using the adult variable could be if the treatments had any effect on the horse
populations. Therefore, I ran tests for the adult vs. treatment variables. The treatments were
labeled “0” for no treatment and “1” for those that got treated. A test for normality was done
and displayed by the graphs and Proc Univariate. Afterwards, a two sampled F-test was
conducted as well as several two sample T-tests to figure out whether the treatments had any
Then, one must look at the foal populations to determine whether or not the
sterilizations worked. The two variables I tested against the foal population numbers included
location and treatment. For foals vs. location, graphs and Proc Univariate analyzed the
normality of the population. Then two sampled F-test and T-tests were done to answer the
question of location’s affect on the foal populations. After that, treatment vs. foal variables
were considered. Again, to test for normality the graphs and Proc Univariate tests were
applied. Normality is important in these experiments because it has effects on the tests that
can later be run on this data set of wild horses. Then, two sampled F-tests and T-tests were
done to figure out the effect of treatments on the foal populations within the two years of the
experiment.
Results/Discussion:
First, I looked at the adult populations and compared them with the locations they were
in and whether or not they were treated in those populations. In the case of adults vs.
locations, It was shown through the F-test that the variances for these variables were equal. I
further tested this using the T-test, to make sure that this conclusion was accurate. The results
were the same for this test as well. I concluded from this that the location of these horses had
little to do with the populations within the adult horses. Next, I took the adults vs. treatment
variables. I ran an F-test on this information and found that the variances for were not the
same. To further test this I ran the T-test and found the same results. Then, I tested which
population was larger than the other. To do this I first took the population with treatment 0
and asked if it was larger than treatment 1. The results were that the treatment 0 populations
were larger than the treatment 1 populations. I made sure this was true by asking the opposite
question and found the same answer. This shows that the population of adult horses had
larger numbers to begin with for those that were not treated. Determining this helps figure out
whether or not the sterilization of these horses worked because if there are not equal, or close
to equal, numbers of horses in these populations being tested, there could be problems with
the results of this experiment. You can view the actual results of these tests in the sections
packet 1 for adults vs. location and packet 2 for adults vs. treatment variables.
To test whether sterilization worked in the following years, I tested the foal populations
against the location and the treatments of these populations. First, I took the foals and tested
them against the locations. The F-test concluded that these variances were the same within
the populations of foals for each location. To make sure this was accurate; I ran a T-test and
found the same result. After this test determined that the location had little to do with the
populations of foals, I ran the tests for the treatments. First, I ran the F-test to see if the
populations were equal for both treatments. I found that they probably were, so I ran the T-
test accordingly. In the T-test I found that the populations of foals were not the same for each
treatment. To determine which population was larger than the other, I ran another T-test that
concluded that the populations for treatment 0 were larger than populations for treatment 1.
This means that those populations of wild horses that were not treated had a higher number of
foals being born in the two years that the experiments were being run. The actual numbers for
these results can be viewed in the appendix labeled packet 3 for foals vs. location and packet 4
The foal population results comply with the adult population results. This shows that
the data is consistent in saying that the populations for treatment 0 had larger numbers in the
beginning and the end of the experiment. When looking back at the data sets of wild horse
populations, there is definitely more horses in the non-treated populations vs. the treated
populations. This can lead to the conclusion that the results are not as accurate as they could
be because there is a need for more samples in the treated populations. The results of these
tests suggest that the sterilization treatment did not have much effect on the horse populations
Conclusions:
According to the tests done for both foal an adult populations, the treatment of
sterilization in this experiment did not work. However, when looking at the normality of these
data sets, we can see that they do not completely follow the normal population curve
completely. This could be due to the fact that there were much greater numbers of horses that
were untreated rather than treated. This leads to the conclusion that there are more samples
of treated horses needed for this experiment to make a more accurate decision on whether or
not treatments help control the populations of wild horses in Nevada and Oregon. With more
samples, there could be a more accurate conclusion to this experiment. This may take longer
to physically do, but if accurate results are desired, time must be the sacrifice. However,
getting a larger population to sterilize would be more difficult, because the process of giving
- Proc Univariate
- Box-Whisker Plot
- Histogram
- Scatter Plot
- F-test
- T-test
- Proc Univariate
- Box-Whisker Plot
- Histogram
- Scatter Plot
- F-test
- T-test 1, 2, 3
- Proc Univariate
- Box-Whisker Plot
- Histogram
- Scatter Plot
- F-test
- T-test
Packet 4: Foals vs. Treatments
- Proc Univariate
- Box-Whisker Plot
- Histogram
- Scatter Plot
- F-test
- T-tests 1 and 2
Nicolette Casselli
The UNIVARIATE Procedure
Variable: Adults
Moments
N 38 Sum Weights 38
Mean 110.236842 Sum Observations 4189
Std Deviation 71.1809397 Variance 5066.72617
Skewness 0.41347328 Kurtosis -1.2768588
Uncorrected SS 649251 Corrected SS 187468.868
Coeff Variation 64.5709169 Std Error Mean 11.5470732
Location Variability
Quantiles (Definition 5)
Quantile Estimate
Nicolette Casselli
The UNIVARIATE Procedure
Variable: Adults
Extreme Observations
----Lowest---- ----Highest---
Value Obs Value Obs
25 22 197 8
30 33 226 7
31 27 232 1
35 26 240 15
36 34 243 14
Nicolette Casselli
The UNIVARIATE Procedure
Variable: Treatment
Moments
N 38 Sum Weights 38
Mean 0.5 Sum Observations 19
Std Deviation 0.50671171 Variance 0.25675676
Skewness 0 Kurtosis -2.1142857
Uncorrected SS 19 Corrected SS 9.5
Coeff Variation 101.342342 Std Error Mean 0.08219949
Location Variability
NOTE: The mode displayed is the smallest of 2 modes with a count of 19.
Quantiles (Definition 5)
Quantile Estimate
Nicolette Casselli
The UNIVARIATE Procedure
Variable: Treatment
Extreme Observations
----Lowest---- ----Highest---
0 19 1 34
0 18 1 35
0 17 1 36
0 16 1 37
0 15 1 38
Nicolette Casselli
Two Sample t-test for the Means of Adults within Treatment
- Sample Statistics
- Hypothesis Test
- Sample Statistics
Treatment
Group N Mean Std. Dev. Variance
----------------------------------------------------
0 19 25.78947 6.5708 43.17544
1 19 5 4.9441 24.44444
- Hypothesis Test
- Degrees of Freedom -
F Numer. Denom. Pr > F
----------------------------------------------
1.77 18 18 0.2371