Conducting an Experiment
A science experiment is a procedure designed to test a prediction. Some types of experiments are fairly simple to design. Others may require ingenious problem solving.
Starting With Questions or Problems
A gardener collected seeds from a favorite plant at the end of the summer, stored them indoors for the winter, then planted them the following spring. None of the stored seeds developed into plants, yet uncollected seeds from the original plant germinated in the normal way. The gardener wondered: Why didn't the collected seeds germinate?
An experiment may have its beginning when someone asks a specific question or wants to solve a particular problem. Sometimes the original question leads directly to an experiment, but often researchers must restate the problem before they can design an appropriate experiment. The gardener's question about the seeds, for example, is too broad to be tested by an experiment, because there are so many possible answers. To narrow the topic, the gardener might think about related questions: Were the seeds I collected different from the uncollected seeds? Did I try to germinate them in poor soil or with insufficient light or water? Did storing the seeds indoors ruin them in some way?
Developing a Hypothesis
In science, a question about an object or event is answered by developing a possible explanation called a hypothesis. The hypothesis may be developed after long thought and research, or it may come to a scientist “in a flash.” How a hypothesis is formed doesn't matter; it can be useful as long as it leads to predictions that can be tested.
The gardener decided to focus on the fact that the nongerminating seeds were stored in the warm conditions of a heated house. That led the person to propose this hypothesis: Seeds require a period of low temperatures in order to germinate. The next step is to make a prediction based on the hypothesis, for example: If seeds are stored indoors in cold conditions, they will germinate in the same way as seeds left outdoors during the winter. Notice that the prediction suggests the basic idea for an experiment.
Designing an Experiment
A carefully designed experiment can test a prediction in a reliable way, ruling out other possible explanations. As scientists plan their experimental procedures, they pay particular attention to the factors that must be controlled.
The gardener decided to study three groups of seeds: (1) some that would be left outdoors throughout the winter, (2) some that would be brought indoors and kept at room temperature, and (3) some that would be brought indoors and kept cold.
Controlling Variables
As researchers design an experiment, they identify the variables, factors that can change. Some common variables include mass, volume, time, temperature, light, and the presence or absence of specific materials. An experiment involves three categories of variables. The factor that scientists purposely change is called the manipulated variable. A manipulated variable is also known as an independent variable. The factor that may change because of the manipulated variable and that scientists want to observe is called the responding variable. A responding variable is also known as a dependent variable. Factors that scientists purposely keep the same are called controlled variables. Controlling variables enables researchers to conclude that the changes in the responding variable are due exclusively to changes in the manipulated variable.
What Is a Control Group?
When you read about certain experiments, you may come across references to a control group (or “a control”) and the experimental groups. All the groups in an experiment are treated exactly the same except for the manipulated variable. In the experimental group, the manipulated variable is being changed. The control group is used as a standard of comparison. It may consist of objects that are not changed in any way or objects that are being treated in the usual way. For example, in the gardener's experiment, the seeds left outdoors would be the control group, because they reveal what happens under natural conditions.
For the gardener, the manipulated variable is whether the seeds were exposed to cold conditions. The responding variable is whether or not the seeds germinate. Among the variables that must be controlled are whether the seeds remain dry during storage, when the seeds are planted, the amount of water the seeds receive, and the type of soil used.
Forming Operational Definitions
In an experiment, it is often necessary to define one or more variables explicitly so that any researcher could measure or control the variable in exactly the same way. An operational definition describes how a particular variable is to be measured or how a term is to be defined. (“Operational” means “describing what to do.”)
The gardener, for example, had to decide exactly what the indoor “cold” conditions of the experiment would involve. Since winter temperatures often fell below freezing, the gardener decided that “cold” would mean keeping the seeds in a freezer.
Interpreting Data
The observations and measurements that are made in an experiment are called data. Scientists usually record data in an orderly way. When an experiment is finished, the researcher analyzes the data for trends or patterns, often by doing calculations or making graphs, to determine whether the results support the hypothesis.
For example, after planting the seeds in the spring, the gardener counted the seeds that germinated and found these results: None of the seeds kept at room temperature germinated, 80 percent of the seeds kept in the freezer germinated, and 85 percent of the seeds left outdoors during the winter germinated. The trend was clear: The gardener's prediction appeared to be correct.
To be sure that the results of an experiment are correct, scientists review their data critically, looking for possible sources of error. Here, “error” refers to differences between the observed results and the true values. Experimental error can result from human mistakes or problems with equipment. It can also occur when the small group of objects studied does not accurately represent the whole group. For example, if some of the gardener's seeds had been exposed to a herbicide, the data might not reflect the true seed germination pattern.
Drawing Conclusions
If researchers are confident that their data is reliable, they make a final statement summarizing their results. That statement—called the conclusion of the experiment—indicates whether the data support or refute the hypothesis. The gardener's conclusion was: Some seeds must undergo a period of freezing in order to germinate. A conclusion is considered valid if it is a logical interpretation of reliable data.
Following Up an Experiment
When an experiment has been completed, one or more events often follow. Researchers may repeat the experiment to verify the results. They may publish the experiment so that others can evaluate and replicate their procedures. They may compare their conclusion with the discoveries made by other scientists. And they may raise new questions that lead to new experiments. For example, Are the spores of fungi affected by temperature as these seeds were?
Researching other discoveries about seeds would show that some other types of plants in temperate zones require periods of freezing before they germinate. Biologists infer that this pattern makes it less likely the seeds will germinate before winter, thus increasing the chances that the young plants will survive.