Causation
Experiments
Frequency Distributions and Crosstabs
Hypotheses and Significance
Applications!
100
These are the three requirements that must be met to be able to claim causation.
What is empirical association, time order, and nonspuriousness?
100
This is the requirement of a true experiment that is the most difficult to achieve.
What is random assignment (to groups)?
100
A frequency distribution looks at this many variables at a time.
What is one variable?
100
This hypothesis says that there is no relationship between the IV and the DV.
What is the null hypothesis?
100

The time of day is hypothesized to affect the likelihood of victim injury in a robbery. This is an example of this type of hypothesis.

What is a two-tailed (nondirectional) alternative hypothesis?

200
This wonderful adage reminds us to address possible spurious relationships before inferring a causal relationship exists.
What is "Correlation does not imply causation!"?
200
This is not a requirement of a true experiment, but it can help determine the effectiveness of a treatment, particularly because it can allow for within-group comparisons to be made.
What is a pretest?
200
These types of variables work best with a crosstabulation analysis.
What are discrete (usually nominal or ordinal) variables?
200
This is the name of the chosen probability level at which point a big difference is considered "big enough" to be real.
What is the alpha level?
200

Suzy's IV is childhood victimization. Jon's IV is unemployment. Both studies have the same DV: adult criminality. Thus, Suzy's study addresses this element much better than Jon's.

What is time order (and, thus, causality)?

300
If I want to see a change in a population over time, the best way to do this would be to conduct one of these types of nonexperimental studies.
What is a repeated cross-sectional design?
300
This type of quasi-experiment uses matching (either at the individual or group level) to try to make the control group look as much like the treatment group as possible.
What is a nonequivalent control group design?
300
This is the type of frequency distribution we can use to help make it more readable when there are too many variable values.
What is a grouped frequency distribution?
300
If the p-value of our statistic is very high, it would suggest that the pattern that we see is likely due to this.
What is random error?
300

A new program to address alcoholism is randomly provided to half of the members of an AA group to test if it helps maintain their sobriety. The control group finds out the extra help that the treatment group is receiving, and decides that it's unfair and gives up on their AA program as a result. This is an example of this effect, which can invalidate the results of the study.

What is demoralization (a type of contamination)?

400
This type of variable can help explain the causal mechanism and is the variable that the IV works through to have the causal effect on the DV.
What is an intervening variable?
400
When subjects learn from the testing process and change their results on the posttest due simply to this learning process, this is an example of this source of invalidity in experiments and quasi-experiments.
What is endogenous change?
400
This describes a crosstab that has more values in one variable than the other, which can make it a little more difficult to interpret.
What is a "non-square" table?
400
If our p value is .05, this would be our level of confidence that the relationship is real.
What is .95 (95%)?
400

If I run a true experiment, it really helps me be confident when I talk about a causal relationship. However, it makes it a little tougher to be confident about this.

What is the generalizability of my findings?

500
High research costs and subject attrition (losing people from your sample) and fatigue (people becoming bored with the study) are major problems with this type of nonexperimental design.
What is a fixed-sample panel design?
500
This is what makes before-and-after designs different from any other experimental or quasi-experimental design.
What is the fact that they do not have a control group?
500
In a frequency distribution, this is the percentage of cases at each value, with the missing values removed.
What is the valid percent?
500
Just because a statistical finding is significant does not mean it is this.
What is important?
500

Our alpha level has been set at .05. If the probability of our statistic happening by chance is around 3%, this is the hypothesis we would reject.

What is the null hypothesis?

M
e
n
u