2.1: Lab 2 Background

Learning Objectives

• Apply the scientific process.
• Analyze and interpret scientific data and communicate findings.
• Graph original experimental data.

Introduction: The Scientific Process

Science is not a list of facts, but rather a process for learning something new about the natural world. In its ideal form, science is the objective pursuit of discovery. However, it is important to keep in mind that science is done by people, and we all have inherent and implicit biases. As a scientist, we must try to become aware of our biases so our collection and interpretation of evidence is as objective as possible. Luckily, there’s a solution!

The scientific process is an effective and efficient means of inquiry. It is also called the scientific method, but the word process better implies its iterative nature. Figure 1 demonstrates how the scientific process is iterative by depicting it as a cycle.

Figure 1. The steps of the scientific process (illustration from The Globe Program).

An observation is a reproducible description, measurement, or record of an object or phenomenon. No observation is too “obvious” or simple to make. We never know what could lead to an interesting study later on!

A scientific question asks something about one or more of our observations. We can start broad, but at some point, we want to hone in on what exactly we want to know about. It may seem simple, but this will actually dictate the rest of our study! We practiced writing comparative questions during lab 1 as part of the scientific process.

A hypothesis is what we think the answer to our question is. A hypothesis is often called an educated guess. In this context, educated means it is based on well-accepted facts (repeatedly confirmed observations) and natural laws (generalizations about how nature works). Like an observation, there are a few important properties of a hypothesis:

• Testable. There must be some kind of test (see below) someone can design to determine whether the hypothesis is true.
• Refutable (also called falsifiable). It must be possible to conclude that the hypothesis is false.
• Tentative. A hypothesis is never proven! The word “prove” is effectively never used in biological publications. Instead, we can support or refute a hypothesis. This is because hypotheses change as more information is gained.

There are two kinds of hypotheses:

• A null hypothesis (H0) guesses that there is no difference or relationship between variables. This is really the hypothesis that we are testing. We want to know if we can refute (or reject) H0.
• An alternative hypothesis (HA) is just the opposite – it guesses that there is a difference or relationship between variables.

These two hypotheses go hand-in-hand. If we reject H₀, then, by definition, we support H_A. But if we reject H_A, then we support H₀ (or fail to support H_A). The section on analysis and statistics below explains how to know whether we can support or refute H₀.

A prediction is a statement that tells us what to measure in order to test our hypothesis. These often come in the form of an “If… then…” statement. We can use the template, “If [our hypothesis is true], then [we expect to find ___].” The part following “then” is what we are measuring.

A test is a way to collect data (measurements) that will support or refute our hypothesis. There are generally two ways to carry out a test.

Observational studies are ways to collect data, such as counting organisms or measuring sizes, to make comparisons without changing anything for the subjects. We simply compare two or more variables (a specific quantity, circumstance, or factor) that may or may not be different. These can be useful when experiments are impractical or unethical. But the conclusions that can be determined are limited in that we cannot conclude causation.

• “Correlation does not mean causation.” A correlation describes a relationship between two variables. For example, ice cream sales increase during the same time of year that shark attacks increase. Do ice cream sales cause shark attacks? Of course not! This is called a spurious correlation. There is another variable that is causing ice cream sales and shark attacks to be correlated – can you think of what that might be?

Controlled experiments are considered the “gold standard” of scientific tests. We purposefully change one variable, keep everything else the same, then measure the effect of that changed variable. There are a few variables to consider:

• The dependent variable, also called the outcome variable, is what we are measuring to determine whether it changes over the course of the study, or is different between the groups. The outcome of our experiment depends on the values of this variable.
• The independent variable is what we are purposefully changing. Typically, we want to limit ourselves to having just one independent variable, but sometimes it is worthwhile or necessary to have more. The researchers predict that by changing the independent variable, there will be a change to the dependent variable measurements.
  • We should also consider variation (natural differences in samples). Whether an observation study or controlled experiment, the dependent variable is always prone to variation. My results may be a little different from your results. Comparing the averages from many different samples takes variation into consideration.
• Control variables/groups are kept “normal.” That means that there is no treatment done to the control variable, i.e., there is nothing done to it. This acts as a variable against which to compare the results of the dependent variable. In some labs throughout the semester, we will set up different controls for comparison.
• Standardized variables are conditions that are kept constant and the same for all groups or situations throughout the experiment.

We should also describe the quantitative (numerical) or qualitative (non-numerical) nature of variables. Continuous variables are quantitative and include numbers with an infinite range. Discrete variables can be quantitative or qualitative and include categories.

If we set up our experiment properly, we can determine causation! By keeping various conditions standardized and comparing the results, we can determine whether the independent variable caused a change in the dependent variable.

In addition, when communicating results and conclusions from studies, scientists generally want to include whether the findings are important or significant. Thus, proper analysis involves various kinds of statistics, which are usually included in the graph. Two common ways of determining significance are by comparing the means (averages) of two or more groups (for example, by using a t-test) or calculating a correlation to see if two sets of data are related.

Most statistical tests generate a p-value. The “p” stands for probability. We can think of this as the probability/likelihood that something happened by chance. For example, if p = 0.10, we can interpret this as there being a 10% chance that the results could happen randomly, and not because of the independent variable. Generally, we consider the results to be statistically significant if the p-value is less than 0.05 (p < 0.05), which explains that there is a measurable or meaningful difference between the datasets. However, if the p-value is greater than 0.05 (p > 0.05), we conclude that there is no difference or that the difference is likely due to random chance.

It is extremely important to remember our biases here. In particular, we want to be aware of confirmation bias. This happens when we sift through the results to find something that we agreed with before the test took place. For example, if we think male termites are longer than female termites, we might look for the one or two samples that confirms this for us, ignoring the rest of the variation (differences in the data points), even if your statistical test shows that female termites are actually longer! Whenever we make conclusions, we have to support our claims with evidence, the objective results of the tests.

Graphs allow scientists to quickly show patterns in data and make comparisons between variables or groups. There are many versions of graphs that can be used depending on the type of variable measured. Below are two examples:

Bar graphs show the average differences between groups. For bar graphs, one variable is continuous and the other is discrete.

Figure 1. The average body length of female and male termites.

Line graphs show a trend between two variables. One common use of line graphs shows how a variable (such as length) changes over time. Generally, both variables are continuous for a line graph.

Figure 2. The number of times a pollinator (including bees and butterflies) visited a flower over the course of a day.

Whichever kind of graph we make, we need to make sure that we always label the axes! The dependent variable always goes on the y-axis (vertical) because that is the variable we are measuring. The independent variable goes on the x-axis (horizontal). We may also need to use a legend to label independent variables if we use more than one. Below the graph, we also add a caption to describe what the graph is showing the reader. For example, “Figure 1. The average body length of female and male termites.”

Always keep in mind that the scientific process is iterative. Once we get our results, we can ask more questions. Why didn’t our independent variable cause a change in our dependent variable? Does another independent variable cause the same or different change in our dependent variable? So, just as we learn something new, we find ways to start over and make other new discoveries!

Introduction: Termites

The eastern subterranean termite (Reticulitermes flavipes) is common throughout North America. They play an important ecological role by feeding on wood, but are also considered a major pest in the United States. Have you ever seen a tented house? Chances are, the house was being fumigated to eradicate a colony of these termites.

Termites have a complex social structure composed of worker, soldier, and reproductive castes. All workers and soldiers in a colony are actually sisters! Worker termites are the only ones to eat wood. When workers find a food source, they will lay down a chemical trail of pheromones to indicate its location to fellow workers. Termites will follow this trail of pheromones very closely, enabling them to reach the food source efficiently.

As it turns out, some pens closely mimic these pheromones! Today, we will be designing experiments to test whether R. flavipes worker termites follow trails made by certain pens.

Introduction: Experimental Design

While designing our experiments, we need to keep a few things in mind:

• What is the dependent and independent variable?
  • Remember the difference between continuous and discrete variables. We may want to consider measuring one of each.
    • Amount of time, number of times, types of pens, ink color, etc.
• How will we measure the dependent variable?
• How can we stay organized?
  • Create a table to record the data.
  • Assign roles to the group members. Time keepers, counters, recorders, etc.
• Are we controlling for/standardizing all variables except the independent variable?
• Are we testing one individual at a time? If so, is each termite exposed to alternative treatments in random order? Are we controlling for fatigue, habituation, and prior experience?
  • How can we design the experiment to make sure termites in later trials are responding to the ink and not to pheromones laid down by earlier termites?