# Construct Types of Variables for Science Experiment

Variables are central to any scientific endeavor, because they represent the action of an experiment or observation. In an experiment, we refer to them as dependent, independent, and extraneous, concepts that are learned best through examples.

The independent variable should be prominently displayed in a science fair project that follows the experimental method. Independent simply means that the variable is only being manipulated by the researcher. Imagine an experiment where you want to find out whether giving homework really helps improve students’ math grades. Or maybe you want to find out the highest volume you can turn your stereo before your parents make you turn it down. The amount of homework we give, or the volume of the stereo, is the independent variable.

Dependent variables that are measured should be displayed, as well as a few extraneous variables that are being controlled for. Dependent means that the researcher expects it to change, or not, in response to the independent variable. A parent might be curious to know whether the volume level of their child’s music is a product of their emotional state, or a principal might question whether new math teachers tend to give more or less homework than more experienced ones, In these cases the stereo volume and amount of homework given have become dependent variables. Can you name the new independent variables? Can you name the dependent variables in the preceding paragraph?

The relationship of variables, or things that change, to each other will depend on the hypothesis. The hypothesis often begins as a research question, either practical or based in theory, though it will need to be stated as a prediction if an experiment is intended, and not simply a case study or observation. In an experiment, there are often extraneous variables. These are defined as variables that potentially influence the dependent variable. We have to control for these, or minimize their influence, to make sure that we are truly testing for our independent variable.

Imagine you find that turning the stereo volume to four does not get a reaction yet turning it to five gets a request to turn it down. That was on Monday. On Tuesday you decide to run the experiment again and find you can turn the stereo up to 8. Obviously there are extraneous variables influencing whether your parents consider the music to be too loud. Maybe they were stressed out on Monday and had a lower tolerance for loud music, or maybe they actually like the CD you played on Tuesday but not the one on Monday. At this point you could either reformulate the hypothesis or attempt to control the extraneous variables, which could be the CD chosen or the mood of your parents. Can you think of other extraneous variables influencing a parent’s tolerance for volume levels?

Researchers have some extremely complicated and effective ways to control for all types of extraneous variables, like matching groups and statistical regression, but none is better than establishing a control group and assigning groups randomly. Obviously these techniques are not always possible. In the case of the parents, you know that volume tolerance is unique to them, because your friend next door constantly turns the volume to 11 with no complaints, while your friend at school has been ordered to always use headphones. But what about the school principal?

She administers a middle school, grades 5-8, and she predicts that math teachers with more than five years experience give less homework on average than other teachers. The first thing she decides to do is set up a control group. The participants cannot be randomly assigned, because they will all be teachers at her school, but she can make it somewhat random by choosing five teachers from the roster randomly. What this means is that results from the experiment apply only to her school, they cannot be generalized to all teachers. So she asks the five to report to her the amount of homework they assign over the next week. Next she prints out a roster of all math teachers with more than three years experience and randomly chooses five of them to report their homework assignments. Here we have a control and experimental groups. Can you name the dependent and independent variables? Are there extraneous variables, or alternative explanations, not accounted for? How would you set the experiment up differently, if the hypothesis predicted that new math teachers give less homework?

In some types of experiments, especially research on human behavior, it is extremely difficult or impossible to remove all the influence of extraneous variables. We can still learn from this kind of research, especially with a clever experimenter skilled in asking the right questions to create appropriate dependent variables to measure. Imagine we have ten strawberry plants and predict that giving more than the recommended fertilizer will increase the number of strawberries produced. We set up a control with five plants receiving the recommended fertilizer amount. The other five we decide to give an extra tablespoon of fertilizer. It seems straight-forward, we will measure the effects of our independent variable (the amount of fertilizer given) on the dependent variable (amount of strawberries produced), while controlling for extraneous variables with the control group that receives everything the experimental group receives except the increased fertilizer.

In this experiment, the dependent variable is referred to as a construct. A construct is a term that has to be defined in order to be measured, and all variables are constructs. More fertilizer is a construct defined here as an extra tablespoon. The amount of strawberries is a construct as well. We might take it to mean the number of berries produced, but this raises a question. What if the plants receiving the extra fertilizer produce more berries but of less weight, or what if they produce fewer berries of greater weight. We decide to define more or less berries in terms of weight. Notice, our hypothesis is merely concerned with amount produced and does not include anything on taste or nutrition. We might include a qualitative measure of taste, so we know whether we really want to give them more fertilizer in the future. If we had a lab, we could run nutritional tests to measure the anti-oxidant power of each group.

Let’s return to the first example. Parents asking for the music to be turned down is a construct. Does it mean them coming to your room to ask or waiting until dinner-time to remind you to keep it down? This could be recorded quantitatively on a chart containing different volumes and a turn-it-down and no response columns. But what happens when, for the first time ever, they yell across the house for you to turn it down? Does that count? Researchers often include space for qualitative information in their records to account for just such abnormalities, because a yell could mean that they care less about the volume than an actual visit to your room, or it could suggest the presence of extraneous variables.

Life is composed of variables. They are everywhere and can be considered the action, or verb, of life. In a science experiment we seek to isolate a few variables in order to test their effects on each other. For a science fair project it is especially important to prominently display the variables of your experiment with their proper names. Doing this is rarely easy and requires a lot of thought and practice.