design, and to the degree that other variables have been controlled, we can be more confident in making causal inferences with these designs than we can with nonexperimental research. We discuss the various ways to control these other variables in Chapter 4.
Conceptual Exercise 2A
1 Identify the IVs and DVs for each of the following findings:Reaction time decreased when more practice trials were given.The amount of exercise had an effect on depression ratings.
Moderating Variables
Many cold remedies display warnings that they should never be taken with alcohol. It is often the case that these drugs can cause drowsiness, but this cause-and-effect relationship is increased with the consumption of alcohol. In this example, alcohol is acting as a moderating variable by amplifying the drowsiness effect of the drug. Moderating variables act to influence the relationship between the IV and the DVs. A moderating variable can increase, decrease, or even reverse the relationship between the IV and the DVs. If, as discussed previously, the IV is the cause in the cause-and-effect relationship and the DV is the effect in the cause-and-effect relationship, the moderating variable is a third influence that must be taken into account to clearly describe the cause-and-effect relationship. For example, in his famous studies on obedience, Milgram (1974) found that the actions of a confederate-companion (someone posing as a participant who is actually part of the study) could produce a strong moderating effect. When the confederate-companion agreed to shock the learner, 93% of the true participants continued to administer shocks, but when the confederate disobeyed the order, only 10% of the true participants continued.
In the Knez (2001) study of the effect of light on mood, gender was identified as a moderating variable. Relative to cool light, warm lighting produces a more positive emotional response in women than it does in men. Therefore, the influence that lighting has on mood is moderated by the gender of the participant.
In a recent study conducted in Islamabad, Altaf and Awan (2011) examined how spirituality moderates the relationship between office workload and job satisfaction. By surveying 76 employees, they assessed the relationship between workload, workplace spirituality, and level of job satisfaction. Workplace spirituality is a person’s sense of meaning and purpose. They found that spirituality was positively related to job satisfaction, but they didn’t find the negative relationship between workload and job satisfaction that had been reported in the literature (Khan, 1980, as cited in Altaf & Awan, 2011). They suggest that workplace spirituality moderates the usually negative effect of workload on job satisfaction. Essentially, if a workplace fosters an environment where employees find spiritual meaning, a heavy workload does not reduce job satisfaction. (For more information on moderating variables, see the discussion of factorial designs in Chapter 7.)
The following diagrams illustrate different relationships between IVs and DVs, including how other variables can affect a relationship.
This diagram shows a direct relationship between an IV and a DV.
For example,
IV DV
Condom use Sexually transmitted infections (STIs)
Condom use is directly related to incidence of STIs in sexually active young people.
This diagram illustrates how the relationship between the IV and the DV is changed by another variable. The moderating variable may strengthen, weaken, or nullify the relationship between the IV and the DV.
For example,
IV Moderating variable DV
Condom use Age STI
If the causal relationship between condom use and STIs is stronger in younger people, perhaps because they are having more sex with more partners, and weaker in older people, who tend to be in monogamous relationships, then age moderates the effect of condom use on STI incidence.
This diagram demonstrates when the relationship between the IV and the DV is accounted for by another variable.
For example,
IV Mediating variable DV
Level of education Condom use STI
If highly educated people have fewer STIs than poorly educated people, condom use might be a mediating variable. The relationship between level of education and STI incidence is explained by the greater use of condoms by better-educated people.
Mediating Variables
Sometimes the relationship between cause and effect is directly linked—your baseball strikes a window, and it breaks. However, there are many instances when this relationship is anything but direct—you look at a bright light, and your pupil constricts. Certainly, this is a cause-and-effect relationship, but there are many intervening steps. Suppose you just had an eye examination and the doctor used eye drops to dilate your pupils. What will this do to the cause-and-effect relationship? You leave the office and go into the bright sunlight and . . . nothing, no pupil constriction. Clearly, the eye drops are acting on some mediating variable between the light and the pupil constriction.
Identifying mediating variables may be centrally important to your research or entirely trivial, depending on how the research fits into the particular theory. For much behavioral research, the mediating variables may be unimportant. Instead, the focus is on identifying and describing the environmental cues (cause) that elicit behavior (effect). Contrast this position with cognitive research, where much of the focus is on identifying mediating variables.
In the Knez (2001) study, the identification of a mediating variable was an important point. He was trying to show that the characteristics of light do not directly influence cognitive performance but, rather, that the light influences the participant’s mood and that a change in mood, in turn, affects the participant’s performance.
In another example of a mediating variable, Lassri and Shahar (2012) examined whether childhood maltreatment affects the quality of romantic relationships between adults. They administered a questionnaire to 91 undergraduate students at the University of Negev, Israel, and measured a number of variables, including childhood emotional maltreatment, self-criticism, and quality of romantic relationships. By using a correlational technique called structural equation modeling (which will be discussed later in this chapter), they found that emotional maltreatment in childhood is related to increased incidence of self-criticism and that self-criticism is related to poor romantic relationships. In other words, a childhood of emotional maltreatment makes a person more likely to engage in critical self-statements, which leads to the person experiencing troubled romantic relationships as an adult. Self-criticism is a mediating variable between childhood emotional maltreatment and adult romantic relationships.
The
