Experimental Design

In this course, we will focus most of our time on controlled experiments as a way to test hypotheses. You need to understand how to properly set up and identify parts of a controlled experiment because you will use these skills for labs in this course.

Let’s set up a scenario that we can use as an example for this lesson. We will imagine that Grandma lives next to a newly built soap factory. The soap factory has a pipe that puts out excess soapy water into the stream near your grandmothers prize-winning tomato garden. Grandma is not happy and is worried the exposure to the soap will affect tomato growth negatively. As a junior scientist, you decide to help her get some answers by conducting an experiment!

Let’s learn some more about how to set up our test for this hypothesis:


What Is a Controlled Experiment?

A controlled experiment is a test for a hypothesis. The factors in the experiment that change are called variables. It is best to set up an experiment in which only one variable is being tested. All other variables should be unchanging, or controlled. This is why we call it a controlled experiment.

What is the one variable we are testing in Grandma’s Tomato Experiment? (ANSWER: the presence of soap.)


There are two variables in every controlled experiment:

  • The variable that is deliberately changed is called the independent (or manipulated) variable. It answers the question: What is changed?
  • The variable that changes in response to the manipulated variable is called the dependent ( or responding) variable. It answers the question: What is measured?

View the variables in Grandma’s Soap Experiment in the image below:



There are two or more groups in every controlled experiment:

  • A controlled experiment will usually have an experimental group and a control group.
  • The experimental group is the manipulated group to which the independent variable is applied. Some controlled experiments may use more than one experimental group.
  • The control group is exposed to all the same conditions as the experimental group EXCEPT for the independent variable. It serves as a point of comparison to the experimental groups.
  • The conditions that are maintained identical for both groups are constants, or may be referred to as controlled variables. They answer the question: What do I keep the same in all groups?

View the groups in Grandma’s Soap Experiment in the image below:


Which of the following are constants, or controlled variables? Click the triangle before each question to check your answer:

    • The species of tomato is the same in both groups.

      This is a constant or controlled variable.

    • The control group was watered at night and the experimental group was watered in the morning.

      both groups would need to be watered at the same time to control this variable.

    • 500mL of water and 500mL of soap solution were applied to the plants each day.

      This is a controlled variable. Both are watered with the same amount.

    • Both will be planted in the same location.

      This is a controlled variable.

    • One group will be watered with water, the other will be watered with soapy water.

      This is the description of the independent variable being applied.



Sometimes you may use more than one experimental group. Let’s imagine we want to see at what concentration the soap affects plant height. We might adjust the experiment as follows to include 3 treatment groups, or experimental groups. View the control group and three experimental groups in Grandma’s Soap Experiment in the image below:


Important Considerations

When you design an experiment, there are factors that could affect your results other than your independent variable. A good scientist plans in advance to avoid these pitfalls:

-Confounding variables

Often, there may be variables that unintentionally affect your experiment. You want to eliminate these confounding variables because they could influence your results. Remember, we want to set up the experiment so that we are only measuring the effect of the independent variable. Which of these might be a confounding variable? Click the triangle before each question to check your answer:

  • The water used to make the soap solutions was very hot.

    This is a confounding variable. The temperature might hurt the plant/growth if it is too hot.


  • The soap used in the experiment was the same soap sold at the factory.

    This is not a confounding variable, it is a *controlled* variable. We would want to use the same soap as the factory releases.


  • You used different species (or types) of tomato plants in each group.

    This is a confounding variable. You would want to control for the species of tomato: perhaps different species grow at different rates?


-Low Sample Size

The sample size is the number of test subjects (usually organisms) in each group. Having a larger sample size reduces the chance that a few individuals’ odd measurements would throw off our findings. We do not want a low sample size (such as one or two individuals) because it would give us inaccurate results that do not represent others of that species/population.

Why not use one or two test subjects to save time and money? We need to consider that some individuals (like Grandma’s tomatoes), by chance or by genetics, may die or show differences because of their DNA. To explore more about sample sizes and statistics (how to detect differences mathematically) navigate to the references in the blue sidebar.


Bias is when the scientist unintentionally skews the data based to their opinions or beliefs. For example, if you are measuring tomato plants and expect for one group to be larger, you may have a tendency to round up when measuring.

-Placebos as Controls

In medical studies where a new drug is tested, scientists use a placebo for the control group. The control group would be asked to take a pill (or treatment) that looks/tastes just like the real thing, but would not be told whether they have the actual drug or not. The experimental group would receive the real drug. This is a blind study, where the test subjects do not know their assigned group.

A double-blind study is used to eliminate the scientists bias, too. The scientist would not know which pill the subjects received until after s/he has collected their results.

Now that we’ve learned about some possible pitfalls when designing experiments, you should have gained some healthy skepticism about experiments in the news. Most scientific studies are sound because they are put through the process of peer review, but we do need to interpret them with an educated perspective.