Page 41 -

26 CHAPTER 2 Experimental research

of a hypothesis. Important concepts related to hypothesis testing, such as Type I and
Type II errors and their practical implications, are examined in detail.

2.1 TYPES OF BEHAVIORAL RESEARCH
Viewed broadly, all of the methods mentioned above are kinds of empirical inves-
tigation that can be categorized into three groups: descriptive investigations, rela-
tional investigations, and experimental investigations (Rosenthal and Rosnow, 2008).
Descriptive investigations, such as observations, surveys, and focus groups, focus on
constructing an accurate description of what is happening. For example, a researcher
may observe that 8 out of 10 teenagers in a class who frequently play a specific com-
puter game can touch type while only 2 out of 12 teenagers in the same class who do
not play the game can touch type. This raises an interesting observation. But it does
not allow the establishment of a relationship between the two factors: playing the
game and typing. Neither does it enable the researcher to explain why this happens.
Relational investigations enable the researcher to identify relations between mul-
tiple factors. That is, the value of factor X changes as the value of factor Y changes.
For example, the researcher may collect data on the number of hours that the teenag-
ers play the computer game per week and measure their typing speed. The researcher
1
can run a correlation analysis between the number of hours and typing speed. If the
result is significant, it suggests that there is a relationship between typing speed and
the time spent playing the game. The results of relational studies usually carry more
weight than what can be learned through descriptive studies. However, relational
studies can rarely determine the causal relationship between multiple factors (Cooper
and Schindler, 2000; Rosenthal and Rosnow, 2008).
Using the same example, the significant correlation result does not allow the re-
searcher to determine the cause of the observed relationship. It is possible that play-
ing the computer game improves typing speed. It is also possible that teenagers who
type well tend to like the game more and spend more time on it. To complicate mat-
ters even more, the correlation can be due to hidden factors that the researcher has
not considered or studied. For example, it is possible that teenagers who read well
tend to type faster and that teenagers who read well tend to like the game more and
spend more time on it. In this case, playing the computer game has no impact on the
typing speed of the teenagers.
How, then, can the researchers determine the causal effect between two factors?
The answer lies in experimental research (Kirk, 1982; Oehlert, 2000). The research-
ers may recruit teenagers in the same age group and randomly assign the teenagers
to two groups. One group will spend a certain amount of time playing the computer
game every week and the other group will not. After a period of time (e.g., 3 months
or longer), the researchers can measure each teenager's typing speed. If the teenagers
who play the computer game type significantly faster than the teenagers who do not

1
Correlation analysis is a statistical test designed to identify relationships between two or more factors.
Details of correlation analysis are discussed in Chapter 4.