Our understanding of the methods used to conduct good scientific research is important for progress in our scientific understanding but also impacts our daily lives. Understanding good scientific methodology allows us to not only conduct experiments, but it helps us to analyze research conducted by others. For example, it helps us to determine whether research studies reported in the news are reliable. Research knowledge also helps us to discriminate among different medical treatments when it comes to making personal health decisions.
Scientific research methods include several steps, which may differ depending upon the topic to be addressed by a study. Standard scientific methods typically include: definition of the research problem, conducting background research, formulation of hypotheses, designing and conducting experiments, analysis of results, formulation of conclusions, and communication of research results to the public.
Central to our acquisition of scientific knowledge is the concept of the experiment. Researchers do experiments to answer questions about the world around us. The following are examples of simple research questions in human physiology:
- Does changing the respiratory rate affect heart rate?
- Does caffeine consumption affect blood glucose levels?
- Does body temperature affect blood oxygen levels?
In order to answer these questions, researchers begin by formulating testable hypotheses. A hypothesis is a tentative statement describing the relationship between the variables in an experiment. Research hypotheses are written as if/then statements that include dependent and independent variables.
A variable is any factor that can change, affecting the experimental results. The dependent variable is the variable in the experiment that is measured by the researcher. The independent variable is the variable that is manipulated by the researcher in order to exert an effect on the dependent variable. In the first example research question, the heart rate is the dependent variable, and the respiratory rate is the independent variable. The researcher will use an experimental method (for example deep breathing) to manipulate a subject’s respiratory rate to measure whether any changes occur in the heart rate.
Dependent variable: the variable that is measured as the output of an experiment (the result)
Independent variable: a variable that is manipulated by the researcher
A hypothesis is a “tentative statement that proposes a possible explanation to some phenomenon or event.” Hypotheses written for the purpose of conducting experiments must be testable. Formalized hypotheses use an if/then format that helps to assure that all important aspects of the hypothesis are intact, including the independent and dependent variables. Additionally, a good research hypothesis has three parts: an explanation of a phenomenon to be tested, a method, and a prediction. A research hypothesis must be written before an experiment is conducted.
Imagine students working on a physiology project involving muscle contraction and temperature. The students observe that cold hands do not function as well at performing certain tasks requiring manual dexterity than do warm hands. The students decide to test grip strength under different temperature conditions using a handgrip dynamometer, which measures the strength of contraction of hand and forearm muscles.
The following are examples of bad and good research hypotheses for this experiment:
- My grip strength will be stronger with warm hands than with cold hands.
This example is not a research hypothesis because it only includes a prediction. A prediction by itself is never a formalized hypothesis.
- If I test grip strength with a handgrip dynamometer, then my grip strength will be stronger with warm hands than with cold hands.
This example is not a research hypothesis because it only includes a method (a test) and a prediction. It does not include any explanation of the phenomenon to be tested.
- If low temperatures suppress muscle contraction, and I test grip strength at different temperatures with a handgrip dynamometer, then my grip strength will be stronger with warm hands than with cold hands.
This is an example of a correct research hypothesis. Note the three parts: “if low temperatures suppress muscle contraction” (a possible explanation of the phenomenon to be tested), “and I test grip strength at different temperatures with a handgrip dynamometer” (the method used for the test), and “then my grip strength will be stronger with warm hands than with cold hands” (the prediction).
This writing sample is also an example of a formalized hypothesis due to the use of the if/then format. In this hypothesis, the independent variable is muscle temperature, and the dependent variable is muscle contraction strength.
Background: The ad for a creatine supplement claims ingesting 10g of creatine once a day for four weeks results in measurable increases in muscle mass. A student decided to test the claim in 10 subjects by measuring the circumference of the upper arm, around the belly of the biceps muscles, before and after treatment. The subjects were not allowed to take part in weight or resistance training during the testing period.
Write a hypothesis as an if/then statement for this experiment:
What is the dependent variable?
What is the independent variable?
Designing Experiments involving Humans
Well-designed experiments must minimize the effects of extraneous environmental and physiological factors, in order to make sure changes recorded in the dependent variable are actually the result of manipulating the independent variable. Experimental controls establish a baseline for the experiment. When conducting human subject experiments in physiology, the control might consist of a separate group of people, the control group, who are not exposed to any manipulation of the independent variable, or it might be the same group of subjects tested before (and then after) altering the independent variable.
Experimental studies may be in vitro, conducted in highly controlled laboratory conditions (example: in a test tube), or in vivo, conducted in a live organism. Controlled laboratory experiments (also called “bench research,” molecular, or cellular research) allow for a great amount of control over the variables that could affect experimental outcomes because all the components in the experimental system can typically be easily accounted for and measured. In human subject research, studies that use human participants to answer a research question, there is typically much less control over experimental variables due to the natural anatomical, physiological, and environmental variation innate to human populations. These are called external variables and can profoundly affect the outcome of an experiment. For example, two subjects may metabolize a compound differently due to differences in enzymes or two subjects that may react to cardiovascular stress differently due to their sex, age, or fitness level. To account for these external, or uncontrolled, variables in human subjects, experiments often use a within-subjects design (below) where the dependent variable is measured in the same subjects before and after manipulating the independent variable.
In human subjects research, there are two main types of experimental designs: within-subjects design and between-subjects design. In a within-subjects design, the subjects of the study participate under each study condition, including in the control group. In the most simplistic design, the subjects participate in baseline measurements for the control (no treatment) and then participate under experimental conditions. Because the subjects in this kind of study serve as their own control group, variation in the results due to many external variables can be reduced.
An example of a simple within-subjects design can be found in many pharmaceutical studies where a group of participants is given a placebo drug for a defined amount of time, and then the same group is given an experimental drug. Differences in physiological measurements after treatment with the experimental drug are inferred as effects of drug administration.
One disadvantage of this research design is the problem of carryover effects, where the first test adversely influences the other. Two examples of this, with opposite effects, are fatigue and practice. In a complicated experiment, with multiple treatment conditions, the participants may be tired and thoroughly fed up of researchers prying and asking questions and pressuring them into taking tests. This could decrease their performance on the last study.
Alternatively, the practice effect might mean that they are more confident and accomplished after the first condition, simply because the experience has made them more confident about taking tests. As a result, for many experiments, a counterbalance design, where the order of treatments is varied, is preferred, but this is not always possible.
Another type of experimental design is the between-subjects design. In the between-subjects design, there are separate participants for the control and treatment groups, which avoids carryover effects. However, the between-subjects design may make it impossible to maintain homogeneity across the groups: age, gender, and social class are just some of the obvious factors that could result in differences between control and treatment groups, skewing the data.
Within-subjects design: the subjects in the study participate in the control and treatment conditions
Between-subjects design: different groups of subjects participate in the control and treatment conditions
No matter how careful we are in creating an experimental design, no experiment can be perfect. We must assume there is some margin of error in the collected data. There are three general types of errors that can impact the outcome of an experiment:
- Human error: human errors are simple mistakes made by an experimenter. For example, the experimenter didn’t appropriately attach a sensor or read a patient’s blood pressure wrong.
- Systematic error: systematic error is often due to poor experimental design or instrument error (poorly calibrated instrument, etc.). Systematic errors include sampling bias, selection bias, and measurement bias.
- Sampling bias: the participants in the study are not representative of the population at large; thus, the results cannot be generalized outside of the study population. For example, data from a study conducted on only 80- year-old men may not be generalized to everyone else in the human population.
- Selection bias: the assignment of subjects to control and treatment groups was not random, resulting in experimental results highly impacted by external variables. For example, a control group that included only females and a treatment group that contained only males.
- Measurement bias: the experimenters rate subjects differently due to their own expectations of experimental outcomes.
- Random error: by-chance variations in measurements that cannot be controlled. Random errors can be reduced by repeated measurements.
The box below lists some sources of error that are possible in all human subject experiments.
Common factors adversely affecting the outcome of human subject experiments:
- Subjects in the study are not representative of the human population at large: e.g., small sample size is too small to fully account for variation in the population
- Interference due to external variables
- Problems with the reliability or accuracy of instruments: e.g., equipment does not have the precision to detect changes in the dependent variable
- Human error: the researcher makes an erroneous measurement or other error
- http://www.accessexcellence.org/LC/TL/filson/writhypo.html ↵
- Martyn Shuttleworth (May 16, 2009). Within Subject Design. Retrieved Jul 30, 2019 from Explorable.com: https://explorable.com/within-subject-design Creative Commons-License Attribution 4.0 International (CC BY 4.0). ↵