a.  I collect information about people's income on a survey.  I want to compute the strength of the relationship between income and number of shopping trips.
Correlation ("relationship" is the key word).
b.  I also ask if people would like a shuttle to run from campus to Nashville.  I want to know if the distribution of “yes” and “no” responses is different for the two sexes.
Chi-square (the key word is "distribution", also nominal variables).

a. Which part of the hypothesis has to be operationally defined to make an independent variable (copy it)? (2 points)
"people make a fist"

b. The treatment group would be? (3 points)
Making a fist

c. The control group would be? (3 points)
Not making a fist

d. How would you operationally define the treatment and control groups (this is a different question from parts b and c above)? (5 points)
The operational definition would be the operations you would perform to get participants to make a fist or not make a fist. It is NOT sufficient to say "have them make a fist," what are the specific operations? The worry with a poor definition is that is can be interpreted in multiple ways and will not be accurately replicated.

a. (see here)
Do part b. first, then fill this in based on your decision.
Which design will you use (circle one: 2-group after-only, 2-group pre-post, matched, within)? Fill in the chart below. (3 points)
If a part is not in your design, put a “-” under it. For the parts in your design, fill in what you will do (for example, in our class experiment we might put “measure pre-existing love” under “Screen”).

Screen	Match	Assign	Observe	Manipulate	Observe
- (not needed)	- (not needed)	Random	+ (measure feelings of power)	+ (fist or not)	+ (measure feelings of power)

b. Why have you chosen this design? (4 points)
Start here:
--Make a worry list (confounds, things that undermine the implementation of the IV, things that undermine the measurement of the DV). Mine: People unable to make a fist, people in emotional states that make it hard to experience feelings of power, people with life events that make it hard to experience feelings of power. (Admittedly, there doesn't seem to be a lot to worry about for this one.) I think I can deal with all of these by eliminating those people from the design. If not, I would have to match it.
--Do you have a homogenous DV? Feelings of power could vary, I think pre-post.
--If I hadn't decided on pre-post, then I'd do after-only.
Remember: The answer is far less important than the reasoning. We don't have to agree, but I have to see that you based your answers on the right process. Help me to help you.

a. What is one threat to the internal validity of the IV? How will you deal with that threat? (5 points)
The answer needs to be the ONE threat most relevant to this specific design. You don't need to show me all of your reasoning, but you do need to justify the one you picked. Copying something at random from the cheat sheet isn't going to get you very far.
--Effective manipulation? It kind of depends on your operational definition. In this case, I don't see how it could apply. Generally, I frown on choosing this one. Why didn't you just do a better job in number 4d? However, sometimes, as in the love study, we have to check the manipulation even after doing the best we can.
--Appropriate control group? Not making a fist seems appropriate. Sometimes how you choose the control will be the answer.
--Confounds? I generally frown on this one as an answer. If you did a good job in 5b, you already took out all of the confounds. If there's an obvious one left, why didn't you handle it with your design? In actual research situations, this is always a concern and it's a good idea to check again by reviewing the literature and your study.
--Demand characteristics? It is possible that people could figure out that making a fist could be related to power.

I would pick demand characteristics.

b. What is one threat to the external validity of the IV? How will you deal with that threat? (5 points)
Again, pick the MOST relevant one.
--Participants? I did take some people out. That makes it a relevant (if uninspired) answer.
--Variables? Do people make fists in this way? (Depends on the operational definition.) Possible.
--Setting (ecological validity)? Do people self-administer power manipulations in this kind of setting? It's technically a possibility, but seems the least important in this design.

I think I would go with participants.

a. Which part of the hypothesis has to be operationally defined to make a dependent variable (copy it)? (2 points)
"they will feel more power"

b. How would you operationally define it? (3 points)
It's kind of cheating, but: "I would use a scale from the literature that had been shown to be reliable and valid." It is very hard to find fault with that, but it closes off options for number 7c.

c. What is one threat to the internal validity of the DV? How will you deal with that threat? (5 points)
Again, MOST relevant.
--Scale? Given my answer to 7b, how can I complain about this? It would need to be interval.
--Sensitivity? Can this DV detect changes caused by the IV? You don't really have the tools on the exam to evaluate effect sizes and see what kinds of studies the DV has been used in previously. So, you will have to note in general terms that there is a chance that the DV won't work.
--Reliability? See 7b.
--Validity? See 7b.

I guess that means I'm going to worry about sensitivity.

d. What is one threat to the external validity of the DV? How will you deal with that threat? (5 points)
Sometimes I won't ask this one. The evaluation of participants and setting is pretty much the same, so you're really looking at variables. Are feelings of power normally measured this way by participants? Probably not.