If a decision maker knows what causes important outcomes like sales, stock price, and employee satisfaction, then he or she can shape firm decisions in a positive way. Causal inferences are very powerful because they lead to greater control. Causal research seeks to identify cause and effect relationships. When something causes an effect, it means it brings it about or makes it happen. The effect is the outcome. Rain causes grass to get wet. Rain is the cause and wet grass is the effect.
The different types of research discussed here are often building blocks—exploratory research builds the foundation for descriptive research, which usually establishes the basis for causal research. Thus, before causal studies are undertaken, researchers typically have a good understanding of the phenomena being studied. Because of this, the researcher can make an educated prediction about the cause-and-effect relationships that will be tested. Although greater knowledge of the situation is a good thing, it doesn’t come without a price. Causal research designs can take a long time to implement. Also, they often involve intricate designs that can be very expensive. Even though managers may often want the assurance that causal inferences can bring, they are not always willing to spend the time and money it takes to get them.
CAUSALITY
Ideally, managers want to know how a change in one event will change another event of interest. As an example, how will implementing a new employee training program change job performance?
Causal
research attempts to establish that when we do one thing, another thing will
follow. A causal inference is just such a conclusion. While we use the term
“cause” frequently in our everyday language, scientifically establishing
something as a cause is not so easy. A causal inference can only be supported
when very specific evidence exists. Three critical pieces of causal evidence
are:
i.
Temporal Sequence
ii.
Concomitant Variance
iii.
Nonspurious Association
Temporal
Sequence
Temporal
sequence deals with the time order of events. In other words, having an appropriate
causal order of events, or temporal sequence, is one criterion for causality.
Simply put, the cause must occur before the effect.
It
would be difficult for a restaurant manager to blame a decrease in sales on a
new chef if the drop in sales occurred before the new chef arrived. If a change
in the CEO causes a change in stock prices, the CEO change must occur before
the change in stock values.
Concomitant
Variation
Concomitant
variation occurs when two events “covary” or “correlate,” meaning they vary
systematically. In causal terms, concomitant variation means that when a change
in the cause occurs, a change in the outcome also is observed. A correlation
coefficient, which we discuss in a later chapter, is often used to represent concomitant
variation. Causality cannot possibly exist when there is no systematic
variation between the variables. For example, if a retail store never changes
its employees’ vacation policy, then the vacation policy cannot possibly be
responsible for a change in employee satisfaction. There is no correlation
between the two events. On the other hand, if two events vary together, one
event may be causing the other.
Nonspurious
Association
Nonspurious
association means any covariation between a cause and an effect is true, rather
than due to some other variable. A spurious association is one that is not
true. Often, a causal inference cannot be made even though the other two
conditions exist because both the cause and effect have some common cause; that
is, both may be influenced by a third variable. For instance, there is a
strong, positive correlation between ice cream purchases and murder rates—as
ice cream purchases increase, so do murder rates. When ice cream sales decline,
murder rates also drop. Do people become murderers after eating ice cream?
Should we outlaw the sale of ice cream? This would be silly because the
concomitant variation observed between ice cream consumption and murder rates
is spurious. A third variable is actually important here. People purchase more
ice cream when the weather is hot. People are also more active and likely to
commit a violent crime when it is hot. The weather, being associated with both
may actually cause both.
Differences
Between descriptive and causal research design
Data
collected through experimental research can provide much stronger evidence of
cause and effect than can data collected through descriptive research. This
does not necessarily mean that analysis of descriptive research data cannot
suggest possible causal linkages among variables, especially when the effects
of uncontrolled variables are filtered through certain analysis techniques available
for that purpose.
Viewing
descriptive versus experimental research is not a clear-cut dichotomy.
Conclusive projects vary from “purely descriptive with no control” at one
extreme to “purely experimental with strict control and manipulation” at the
other extreme.
Conducting
Causal or Experimental Research
Causal or Experimental research is intended to generate the type of evidence necessary for confidently making causal inferences about relationships among variables.
To
make causal inferences with confidence, then, we must manipulate the causal
variable and effectively control the other variables.
No comments:
Post a Comment