Thescientificmethodallowsaresearchertosystematicallyidentifyaproblem.docx

The scientific method allows a researcher to systematically identify a problem, design and conduct a study, examine the data and report the results. Scientists such as chemists and doctors, along with psychologists, use experiments to study their problems. In an experiment, the researcher has almost complete control over the elements involved in the experiment. Because of this control, the researcher can decide cause and effect, depending on whether or not the results are statistically significant. 
In an experiment, the research first formulates a hypothesis, or statement about the variables the researcher is studying. For example, a researcher wants to see if Germ X causes Disease Y. The researcher hypothesizes that exposure to X will cause Y. The researcher identifies the variables of importance. The independent variable is the variable that the researcher will manipulate (i.e. the variable that affects the behavior in question). In this study, the IV is Germ X. The dependent variable is the result that is measured or observed. In this study, the DV is Disease Y.  The researcher must also carefully define what is meant be each variable so that other researchers can replicate or repeat the study. Defining the variables very specifically is called operationalization. We get what are known as operational definitions. You may not agree with how a researcher defines a particular variable. The key, however, is that the researcher has determined ahead of time how each variable will be defined.
Let’s begin with independent and dependent variables.  The dependent variable is the outcome. Dependent variables are often 
scores
 on tests or 
surveys
. Sometimes DVs are the number of times a participant completes an activity. Dependent variables (outcomes) depend on the independent variable. The independent variable or IV is what the researcher manipulates. Manipulation is not always truly done as sometime we use independent variables such as gender or age. In these quasi-experimental (remember that a true experiment requires the researcher to manipulate the IV) studies we are looking at the effect of the IV on the DV (as in all studies). For example, does a person’s gender affect their 
SAT scores
? Gender is the IV and SAT score is the DV.
Read more about 
variables.

Sometimes, however, the IV is not the only thing to affect the DV. Confounding variables are other variables in a study that might affect the outcome of the study. Since researchers are not always able to have complete control over everything in a participant’s world, it is always possible that something other than the IV variable is responsible for any changes that occur in the DV. For instance, in our gender and SAT example, let’s suppose that males get higher SAT scores. Is gender the only thing that “causes” or contributes to higher SAT scores? Probably not. Males tend to take more higher-level math classes and this may be part of why they get higher SAT scores. Since we didn’t study math classes as an IV, it is a confounding variable. In other words, it may be a plausible explanation for what we find.
To conduct the study the researcher must identify the population of interest. A population for a study does not mean everyone in the world; it means the group of people in whom the researcher is interested. In this study, we will say that our population is all teenagers since this disease affects only teenagers. Since we can almost never study our entire population, we sample that population and study a small group that should represent the population. Thus our sample should look like the population. In this case we need to have male, females, a range of ages, ethnicities, and races.
Characteristics used to get a representative sample are the elements that are most likely to be related to something physical. Our sample should be also randomly selected so that we eliminate the potential of volunteer bias in selection. A random sample means that every member of a population has an equal chance of being selected to be in the sample. One way to do this is through the use of a random number table.
Now that we have our sample (the number in the sample is determined by the size of the population) we are ready to proceed. To do an experiment, we will divide the sample (participants) into two groups. We will use random assignment to the groups in order to avoid further bias. Random assignment means that everyone in the sample has an equal chance of being in each group. It is random in terms of the group to which they are assigned. This guards against the influence of participant characteristics.

“To the untrained eye, randomness appears as regularity or the tendency to cluster.” (W. Feller)

You can see that “randomness” may not be accurate enough to obtain a random sample and chance alone can cause results that appear to follow a regular pattern and so we need to replicate even carefully-controlled studies.
Once we have our sample, we need to divide the sample intro two groups that we can compare. Ideally we will randomly assign participants to groups using the same procedures and rationale as when we did random 
selection
. (You should note that in the real-world of research, we often use intact groups.)  For ethical and pragmatic reasons we cannot always assign participants to groups. We can lessen the potential of bias by using random assignment, though, in situations where we were not able to randomly select our participants. The key contribution of randomness is to lessen any potential bias.
One group is called our control group; the other is our experimental group. The experimental group will receive the treatment  –  in this case, they will be exposed to Germ X. The control group does not receive any treatment. After a specified time period, we will see how many in the experimental group contracted Disease Y versus how many in the control group contract the disease. Using statistical calculations, we will determine statistical significance. When you read a study we look for significance levels of at least .05 and preferably .01. If we find those levels, we can conclude that Germ X causes Disease Y.
Because this was an experiment done in a laboratory in which we controlled the environment and other potentially confounding variables, we can say with pretty good certainty that X causes Y. Without these controls, it is difficult to determine cause and effect.
There is the potential of causing harm to the participants which is a violation of the American Psychological Association’s Code of Ethics for Research with Human Subjects. A major consideration for all researchers is conforming to this code and conducting ethical research. Ethical research means that we get informed consent from participants in all research, we limit deceit in research, we debrief all participants and we do not harm them.
Read the APA 
Code of Ethics
.

As you can see, it can be difficult to conduct ethical research on many psychological topics. Thus we conduct what we call quasi-experiments. For instance, we don’t use random selection sometimes because of informed consent issues.
You can also see that is can be difficult to conduct an experiment on many topics of interest to psychologists because we cannot control the environment and there are many possible explanations. Because of this, much of psychological research is correlational. It does not determine cause and effect but shows us the relationship between two variables. We still identify IVs and DVs, control and experimental groups but they are not in the pure form that they would be in an experiment. For that reason we often refer to such research as a study rather than an experiment. We apply as much scientific rigor as we can within our ethical and practical limits.
With our interest in Germ X and Disease Y, we would do pretty much everything that I have mentioned except that we would not actually expose our control group to the germ. Instead, we would determine if they had already been exposed. We would correlate “exposure” and disease. Our statistical calculations would allow us to say that there appears to be a relationship between exposure and contracting the disease. But we CANNOT say that X causes Y since there are other possible explanations which we cannot rule out.
Some correlational research does not set up quasi-experimental groups but relies on intact or pre-existing groups. In survey research, the researcher questions large groups of people about behavior and demographic characteristics and then draws correlations between variables of interest.
In naturalistic observation, we go to the place of interest. For instance, if we want to study the behavior of teachers, we observe them in their classrooms. We can gather descriptive information but we cannot determine cause and effect. We may do a form of naturalistic observation, a case study, in which we study an individual. Clinical psychologists frequently use case studies. Interviews are another type of correlational study.
Information on 
correlational designs
 can be found at this free textbook.

Some of these correlational approaches yield important information that is helpful in predicting and understanding behavior but their biggest limit is in determining causality. It is here that psychology differs from other sciences. Because it is difficult, because of pragmatic reasons and ethical reasons, to conduct true experiments and because we study humans and there are always confounding variables (or other potential explanations) it is difficult to establish absolute cause and effect in many psychological studies. Humans are complex and that complexity can make them difficult to study.

Table of Contents

 Page 3 of 7 

·

Objectives

·

Thought Questions

·

Research

·

PowerPoint

·

Research Methods Video

·

Related Websites

·

DB 2 – Chapter One