Analysis
After we collected all of our data we started to analyze it. First we used stem
and leaf plots to see if there was correlation among active aluminum levels, pH
levels, and soil texture. Since active aluminum was on an ordinal scale, we made
one stem and leaf plot for each active aluminum level (levels 1 through 5
excluding 3). We excluded level 3 because that represented a medium active
aluminum level, which was not part of our data. After each stem and leaf plot
was created , we found where the numbers peaked for pH and for the soil texture
percentages. We put this data into a graph and saw that there was no correlation
between the peaks of the pH values on the stem and leaf plots and the peaks of
the clay percentages on the stem and leaf plots. However, we knew that there
should have been a correlation so we went back and examined the samples that had
changing active aluminum values from day 1 to day 2. We then examined their pH
values as well and to determine which ones followed the Kentucky study pattern
(where decreasing pH causes increased active aluminum and vice versa). We then
decided to see whether these samples were clustered in a certain area in the
ecosystem. Using the maps of the four microclimates, we discovered that the
samples, which followed the correct pattern, were evenly distributed all over
the ecosystem. This showed us that the location of the samples was not the
reason that they followed the pattern . We then found the clay, sand, and silt
levels for each of the samples which followed the correct pattern and averaged
them together to get one clay percentage, one sand percentage, and one silt
percentage. We did the same for the samples which did not follow the correct
pattern. We used three t-tests to compare the clay percentage from the samples
that followed the correct pattern to those that did not, as well as silt, and
then sand. The t-tests showed that silt did not cause some samples to follow the
correct pattern. We concluded this because the ts was less than the tµ
when using an alpha value of .1
This showed us that we were more 90% confident that silt was not the causative
agent. There was no statistically significant difference of silt between the two
sets of samples. Another t-test (which was used for the clay percentage
comparison) showed us that we were also 90% confident that clay was not the
causative agent either for the same reason as the silt t-test showed. Therefore,
for silt and clay, we are accepting the null hypothesis. The sand t-test,
however, showed us that there was a statistically significant difference in sand
between the two sets of samples because the ts was greater than the tµ
when using a .1 alpha value. This
shows that we were 90% confident that the sand was the causative agent, and
therefore, we rejected the null hypothesis for sand.
From our data and analysis, we can conclude there is a higher percentage
of sand in the soil in the backwoods of Roland Park Country School where the
aluminum levels do not change drastically from day to day when compared to those
that do. After reviewing this fact, we realized that we ended up somewhere
completely different from where we started. We started this experiment thinking
that a much higher percentage of clay would cause pH to lower and therefore
cause aluminum levels to rise. This would therefore cause our data to differ and
not follow the pattern that was set by the Kentucky study. However, our results
show that clay, statistically, has nothing to do with it. However, since sand is
silicon dioxide and absorbs water and lets it pass through very quickly, we have
doubts that the difference in sand in different soil samples is really the cause
of the changes in aluminum in the soil. After realizing this, we went back to
our original data and pulled out the 3 outliers, which were the most drastic,
and the most significant from the rest of our data. The 3 outliers were samples
7, 8, and 9 from day 1. Sample 7 is an outlier because it has a very high
aluminum level over both days while the pH value stays at 7.0. Sample 8 is an
outlier because the aluminum level changes from very low to very high from day 1
to day 2 while the pH value stays at 7.0. Sample 9 is an outlier because the
aluminum level stays the same at very high from day 1 to day 2, however, the pH
value was at 8.4 on day 1 but on day 2 it decreased to 7.3. We then went back
and looked at where each of these 3 samples was taken and our data showed that
sample 7 and 8 were taken in site 4 quad 2, and sample 9 was taken in site 4
quad 3. This leads us to believe that there is something in microclimate 4 that
is causing abnormal results. There are 2 obvious properties that microclimate 4
has which are different from all of the other microclimates. One is that site 4
is a wetland, and the other is that it has a monoculture of jewelweed. Therefore
we can infer that there is a certain aspect of jewelweed or of wetlands that
causes this abnormal difference in pH levels and aluminum levels in the soil.
This would be an excellent future research question. We did not have time to
investigate any other topics than sand, however, it would be very interesting to
know whether it really is the jewelweed or the fact that site 4 is a wetland
which is causing this abnormality.