Are you a statistics student? Mad scientist? Middle-school science fair competitor?
You came to the right place.
It’s imperative to know about experimental units when delving into experimental design.
If that last sentence made your head spin, you’re not alone.
You’re probably wondering what an experimental unit is. It sounds super complicated. Especially when you have an experiment to design and conduct, and you’re running low on time.
Even if you’ve got all the time in the world, experimental units and experimental design can be difficult concepts to grasp. Let’s get down to the basics and discover what exactly these units are, and how you must use them within your experiment.
Defining and Classifying Experimental Units
To understand experimental design, you have to understand how experimental units fit in. Experimental units are factors within an experiment which you apply treatment to in order to observe results.
When discussing experimental design, it’s imperative to know which units are your experimental units. For example, say you are designing an experiment wherein you will test out a diet program on four people.
You are conducting the experiment to determine the effect of the diet on blood sugar levels. In this case, your experimental unit would be each of the four individuals being tested. That’s because they are the factors to which you apply the treatment (the diet) to in order to measure the results (the blood sugar).
Experimental Units in the Context of Experimental Design
What is experimental design?
To understand this concept, you have to comprehend the importance of statistics and the experiments which go into creating them.
Data from science, agriculture, medicine and tons of other fields is collected to make up statistics. Statistics are simply the information that is collected from surveys and experiments. This information serves many different purposes, from diagnosing illness to making market projections.
In order to get that data, someone has to conduct experiments. That’s where the experimental design definition comes into play. According to Britannica, experimental design concerns the design and analysis of experiments to collect information.
Basically, experimental design is the art of designing an experiment so that the results are as non-biased and realistic as possible.
Still with us?
Now, how do experimental units fit in into all this? Penn State University defines the experimental unit as the factor in the experiment which receives controlled treatment. The experimental unit is the one from which scientists take measurements and draw conclusions.
How Experimental Design Affects Experimental Units
There are two main types of experimental design, each one utilizing experimental units in different ways.
Randomized Experimental Design
In a scientific experiment, one of the most common practices is to design experiments using randomization. The use of randomization helps to eliminate bias in an experiment.
In a completely randomized experimental design, the experimental units are assigned to testing groups and treatments completely at random and without any sort of bias.
Randomized Experimental Design Examples:
Say you are conducting a study to determine how different amounts of medicine affect a disease in different people. In order to conduct a completely randomized experiment, you must assign the experimental units (that would be the people being tested) into testing groups completely randomly.
Each experimental unit would be set into a control group or treatment group regardless of age, race, previous conditions, etc. You might just draw names out of a pool to set the groups up.
Randomized Block Design
Sometimes, a completely randomized experiment is a little bit too random. In this case you may implement a block design.
In this type of experiment, you divide up the experimental units according to specific traits which may affect the outcome of the experiment. By implementing a randomized block design, you are able to control more tightly your experiment.
Randomized Block Design Examples
If you were to conduct the abovementioned experiment, you may find factors in your experimental units (the people being tested) such as gender may affect the outcome of the experiment.
In this case, you may assign the experimental units into separate “blocks”, ensuring each block has an equal number of men and women.
In this way you would be able to ensure that the results you were getting within each group in relation to the medication were in fact based on that medication, and not on the outside variable of gender.
Local Control in Experimental Design
Because of factors like gender, age and preexisting conditions, it is important to determine beforehand if your experiment should be completely random or be based around a block design.
Making your experiment more efficient and less biased is known as local control. This is done by choosing a block design over a completely randomized experiment.
Use Experimental Units to Solidify Any Experiment
Now that you understand experimental units and a bit of experimental design, you can get going on creating your experiment. Remember, you are the scientist here. So apply these principles well, and get to experimenting.