White Squirrel Example

Lab 1 White Squirrel Example: Hypothesis Development & Experimental Design
Hypothesis Creation
Observation + question: I seem to see a bunch of white squirrels on campus. I wonder if there are more here than in other areas?

Research to find potential mechanism: This article says it is because white squirrels (in particular) are being fed. Is that why?
Incomplete Hypothesis: There are more white squirrels here than in other places because they are being fed.
Correct Hypothesis: White squirrel (S. carolinensis) population abundance at WKU is greater than in other similar areas, like EKU, because they are being fed.
- Null Hypothesis: There is no difference in white squirrel abundance at WKU vs. EKU.
- Prediction: If I can compare the density of white squirrels on the WKU campus to that of EKU, there would be significantly more on the WKU campus.

Experimental Design

Dependent variable: The number of white squirrels
Independent variable: Location (WKU vs. EKU)
Confounding variables: Days spent counting white squirrels, season and weather of sampling days, sample scheme, sampling effort, etc.
Assumptions: The forgage (e.g., nuts), trees, and predator populations would be roughly similar. We should also sample these variables, just to be certain. Otherwise we could not rule them out as reasons more white squirrels are located on WKU's campus.

STOP HERE FOR LAB 1

Lab 2 White Squirrel Example (continued): Analysis

Analysis. On 10 similar days, 10 students systematically searched each campus trying to spot at least one white squirrel (we won't get into the details of population sampling right now). They marked if a sighting was made by anyone on each sample run (see data table).

A chi-square analysis will tell us if the difference is significant and potentially meaningful.
We will get into the details of this test in lab, but significance is an important concept we should go over first!

When we complete a statistical test, we are actually testing our null hypothesis. We seek to reject the null (i.e., show that there IS NOT "no difference). Which conversely supports the idea that there is a significant difference.

By convention, we seek to have at least 95% confidence in the assertions we make regarding our data and conclusions.
To do this we set what we call an alpha value (typically 0.05) and assert that any "p-value" below that number means we can reject our null hypothesis.
What this actually means is that 5% of time we could make an error: 5% of the time we will reject a null hypothesis when we should not (a "type I" error), or 5% of the time we will fail to reject one that we should (a "type II" error).

Each type of statistical test provides a "test statistic" and its value dictates the accompanying p-value. Therefore, for most tests, if our p-value is < 0.05, we can reject the null hypothesis and claim significant (and potentially relevant/ biologically meaningful) findings. A chi-square analyses of these data yields a chi-square = 5.4945 and a p-value = 0.019.

These data show there is a significant difference between WKU and EKU, with a p < 0.05, and we can therefore reject our null hypothesis.
We would be very, very unlikely to see these results from chance alone...Maybe it is the food?

Degrees of Freedom (df) are also used to established significance. Degrees of freedom is is the number of independent pieces of information that went into calculating the statistic. For this example, we are measuring in two environments, and df = the number of environments - 1. So here, df = 1.

Extend your knowledge

Log your white squirrel sightings!

This is a real citizen science project logging white squirrel sightings across the US!. Read more about it HERE.