Sample Results

Overview of steps

Always record raw data in your notebooks before putting it anywhere else. That way, it was less likely to get accidentally edited or deleted. Then, you can enter it into your group's shared Google Sheets, where you can input various statistics such as mean and standard deviation. These statistics have two main purposes: creating graphs with and performing t-tests.

notebook table example

Below is an example of how we would lay out a table in our notebooks. This table was only used for bacteria, although in some cases it is more convenient and efficient to combine multiple different factors into one table. We recommend only writing each new day and plott label as needed for simplicity and readability.

Excel statistics format

Here's a screenshot of an in-progress Excel stats page. Not all of the stats are used in the final paper, but it is helpful to see them to get a better idea of your data and why the results turn out the way that they do.

T-Testing methods

T-tests are used to determine if the data differences are statistically significant. T-test results can be looked at next to the hypothesis to determine if there is any correlation. There are many different programs you can use to test. Some methods that people at E.S.S.R.E. used to do the testing were: manual calculations, built-in Google Sheets functions, the program built into TI-83 calculators, websites, and custom programs. All t-values were compared to a p-value appropriate for the degrees of freedom to an accuracy of 80%.

Note: For the humus levels, a chi square had to be used instead because the data was qualitative.

results

Note: Sometimes, results will align very clearly with the hypothesis. Other times, the pattern is less clear or even non-existent. Sometimes, there is a pattern but it is not the same as previously expected. Here are three examples that show this variety. We decided to use bar graphs to represent the patterns, but feel free to try other graph types for your experiments.

Humus results

Chi-square tests were used to determine if the differences between the humus levels in each plot were significant, and it was found that the differences in all comparisons between plots were not statistically significant.

moisture results

The moisture test yielded interesting results - sites 1 and 3 had virtually the same moisture levels, while site 2 was unusually high. This is likely caused by an uncontrolled variable, like plant density. Either way, the moisture data does not support our hypothesis.

Bacteria Results

Although this data does not align with our hypothesis, there is a correlation between it and the moisture results. It is possible that plot 2 had too much moisture, which prohibited the bacteria from thriving. The bacteria amounts in plots 1 and 3 are not statistically significant, which might be because the moisture level differences between those plots were not significant.

Note: Median and median absolute deviation were used in order to mitigate the effect of an outlier.

Our Conclusion:

Small relative elevation differences do not cause the same significant difference as previous research predicts because factors that would become negligible on larger scales have a more prominent impact on smaller scales
The patterns discussed in the introduction were often determined using larger or steeper sites such as a mountain, while these experiments occurred on the side of a hill
Future research would be more productive on more dramatic sites and would help achieve a greater overall understanding of the effects of elevation on SOM, bacteria, and moisture content

Page updated

Google Sites

Report abuse