Mottos/Mantras/Take-Home Lessons From Chapter 3
- There is usually more than one way to do something. You can do research to
find out which of those ways is best.
- Just as debate moderators can minimize, maximize, or even reverse negative
feelings between two opponents, moderating variables can increase, decrease, or
reverse a factor's effect.
- Sometimes, without a mediator coming between them, the relationship between
two groups (e.g., a divorcing couple, union and management) no longer exists.
Similarly, if the mediating mechanism is blocked, the cause-effect relationship
between an environmental event and a behavioral reaction no longer exists.
- For almost anything, you can ask "Who," "what," "when," "where," "why" and
- "Who," "what," and "where" questions, by themselves, will usually lead to descriptive hypotheses.
- "Why" questions lead to cause-effect (experimental) hypotheses.
- You can expand "why" hypotheses about the first variable causing changes in
a second variable by adding a third variable. To add a third variable,
expand your "why" question to a
- "Why and When?" "Why and Where?"
or "Why and For whom?"
question to create hypotheses that include a moderator variable--a third variable
that moderates--changes--the first variable's effect.
- "Why and How" question to look for the mechanism, that
is the mediating variable, that comes between the cause and the effect and
is the way the cause has its effect.
- Any study can benefit from replication.
- If you think that the study's reported results may be due to a fluke or to
fraud, repeat the study.
- If you question the generality of the findings (you ask yourself, "when, where, and for
whom would that be not true?), test the external validity of the findings by
replicating the study with that different population or in a different setting. Ideally, you might do a study in which you have one set of
conditions in which you replicate the original findings using the original
participant population or setting and another set of conditions in which you
obtain different findings. If tje results for the different groups differ, the factor that you varied between
between your conditions would be the moderator variable.
- Four places to look for problems with the study's construct validity--
and how to deal with the problems you find :
- Unintentional researcher bias. Consider a systematic replication using
- Participant bias-- participants figuring out the hypothesis: Consider a
replication using the double-blind technique, replicate the study
as a field experiment, change the cover story, use more or better control
conditions, or make the study more interesting so that participants are focused
on just doing the tasks rather than on asking " what is the
hypothesis and how should I do this task to support that hypothesis?"
- Manipulation's validity is questionable: Do conceptual replication
using a different measure.
- Measure's validity is questionable: Do conceptual replication using a
- If a non-experimental study suggests a cause-effect hypothesis, test that
hypothesis by manipulating the suspected cause in an experiment.
- A research idea or topic is not a hypothesis. A hypothesis must
- Predict a difference
- on some observable event or characteristic
A hypothesis should
- not be trivial (producing isolated facts) but should be relevant to theory, practice, or previous research.
- be risky-- nothing ventured, nothing gained
- be interesting. Interesting hypotheses may involve
- Rules that account for exceptions (by including moderator variables)
- Rules that explain how a factor causes an effect (by including mediator variables)
- Rules that allow us to pinpoint the relationship between how much of a factor has how much of an effect (i.e., predictions about the shape of the functional relationship between variables)
Back to Chapter 3 Menu