Possible Test 1 Questions
Caution: some of the formatting is lost when I put this in web formatting. not
all of the questions below are laid out properly, and axes for graphs are lost.
I think you can still tell what the question asks however.
SHORT ESSAY. POINT VALUES OF EACH QUESTION ARE GIVEN IN PARENTHESES. Write
legibly. Keep regular left and right margins. It is best to use a pencil so that
you can erase and make corrections. It is generally not necessary to use all the
space allotted to a question. (Don't feel compelled to fill up white space. This
can reduce your score). Label the parts (a, b, c, ...) of multipart questions.
- One researcher finds that the relationship between the length of one's lit
review (in pages) and anxiety level (measured by heart rate
beats per minute "bpm") is r = +.78. Another researcher finds that the
relationship between lit review length (in pages) and anxiety (measured by
fingernail length in cm) is r= -.56! (a) On separate axes draw and label
scatter diagrams for the two correlations and (b) interpret the correlations
using standard interpretation form. Remember to label the axes of the graphs.
Use ovals to indicate regions of scatter. The "fatness" of the ovals for the
two correlations should be have relative accuracy with respect to each other
(absolute accuracy is not expected). (12)
- Imagine the caffeine/RT study is done with five levels of caffeine: 0, 10,
20, 30, and 40 mg. Draw and label a "latin square" that would control order
effects if this experiment were done "within subjects". (5)
- Assume each student in a class of 60 randomly chooses 10 people, divides
them at random into two groups of 5 each, calculates each group's average TV
watching time, and subtracts the averages. We will have 60 differences between
means (like the data shown in class).
a. (6 points) Draw and label an axis below, then construct an ideal frequency distribution
(smooth curve) for the difference between means calculated by the class for
the T.V. watching data. First, draw and label a curve for n=5, as specified
above. (Note: this is a curve for differences between means, not a
curve for means).
Now superimpose (put right on top of one another in the same part of the axis) on the same axis and label an equivalent rough distribution for n = 30. Be sure
to use a reasonable central tendency and dispersion for the second
distribution relative to the first.
b. (2 points) In each case, what is the value we expect to get for the mean of
the curve (remember we're talking about averaging the DIFFERENCES BETWEEN
MEANS)?
n = 5: _______
n = 30: ______
c. (2 points) The EXPECTED difference between any two averages in this
situation is:
_______(a specific numerical value)
d. (2 points) What is the name for the shape of this distribution?
___________________.
- Which of the following techniques for randomly assigning 10 subjects to
two groups might be QUESTIONABLE (because potential biases might creep in) and
which are CERTAIN to produce random assignment to 2 groups, in the case where
the sample itself was randomly chosen (and is therefore already in random
order), and in the case where you are uncertain whether the sample is in
random order.(7)
drawing numbers from a hat, flipping a coin, putting the first 5 in group 5
and the next 5 in group 2, rolling a die, just thinking of the first number
that comes to mind, having a friend decide, using a table of random numbers,
using www.random.org, etc. [other techniques could be listed here]
- The correlations of height, weight, and arm span with day born were NOT
zero. Neither were they zero in other semesters with other sets of students.
Yet we stated the "expected value" of these correlations was zero. Please
explain. (8) Use terms like signal, noise, haze of values, and chance. Mention
what an "expected value" is.
- We stated that the "expected value" of the difference between means for
the TV watching exercise was zero. What was it about our procedure that made
this the expected value? Why didn't we always OBTAIN zero? Please explain. Use
terms like signal, noise, haze of values, and chance. Mention what an
"expected value" is. (8)
- In an experiment on the effects of caffeine on RT, we could balance
gender. Describe (step by step) how to do that for this experiment if you have
16 males and 20 females available for the experiment if the conditions
included are "caffeine" and "no caffeine". (4)
- Indicate which control techniques (a) control a variable precisely; (b)
control a variable by eliminating its variation; (c) control a variable only
on the average by putting a, b, or c in the blanks. [control techniques would
be listed below] (2 each)
- You have a sample of 12 females and 10 males to use in your experiment.
You need three groups of subjects. Describe how to divide the subjects so that
gender is balanced. (Specify how many of each gender is in each group). (5)
- Using an extraneous variable of your choice, describe the TV
violence/aggression
experiment (using boys only) so that it is confounded. (4)
- Confound the caffeine RT experiment (2 points each)
a. state the original hypothesis
b. describe the experiment so that the EV of age is confounded with the IV.
c. what "rival hypothesis" or interpretation does this give rise to?
d. what if age were only PARTIALLY correlated with the IV?
- Describe step by step HOW to match on IQ for an experiment that has three
levels of the IV. Do this by making up 6 IQ scores and showing what should
happen to them. (4)
- We have compared experiments with correlation designs. How does the
experiment solve the causality problems that arise in correlation so that we
can legitimately infer that the IV causes changes in the DV? (4)
- One experiment examined whether caffeine increased alertness. (a) Give 2
possible operational definitions of alertness. (b) Tell how you would
determine whether your definitions are "good" or "poor". (5)
- Some of the subjects in the caffeine/reaction time study discussed in
class never drank coffee or pop and were therefore highly sensitive to the
effects of caffeine. (a) Is this necessarily a confound? Why or why not? (b)
Why does one wish to have no confounds in a study? (8)
- Some of the operational definitions for hunger are
self-report, deprivation of food for six hours, and number of stomach growls
per minute. Using the concept of correlation: (a) Describe how you would
determine whether these definitions are reliable. (b) Describe how you would
determine whether these definitions are valid (that there is a construct of
hunger). (4)
- Identify cases of confounding.
a. in the Harlow experiment, only
male monkeys are used.
b.
in the Harlow experiment, males are
in the "alone" group, and females are in the "accompanied" group
c.
in the birth order study, most of the
first borns are from high income families, whereas later borns are mostly from
the middle class
d.
in the birth order study, only middle
class families are included
- Describe the caffeine-reaction time study so that it is confounded (as
some of the examples above). Then name two control techniques and DESCRIBE how
each would prevent the confounding from occurring. (6)
- Which techniques for randomly assigning 20 subjects to 2 groups is
considered acceptable?. (1
each)
a. toss slips of paper labeled with subject ID numbers into one
pile or another
b. flip a coin that you are certain is unbiased
c. draw slips of paper from a hat
d. use a table of random numbers to assign "odd" subjects
(subjects who receive an odd random number) to one group, and "even" subjects
to the other
e. roll a die (singular of dice) that is unbiased. Rolling a 1,
2, or 3 assigns the subject to group 1; anything else assigns the subject to
group 2
f. ask a friend to just think of a series of digits. Assign the
digits one at a time to the list of subjects, then use odd and even as above
in d.
g. put every other subject in group 1, the remainder in group 2
h. put the first 10 in group 1; the second 10 in group 2
i. ask subjects to choose a number... 1 or 2. Let that number be
their group.
j. line the subjects up according to height and put the first 10
in group 1, etc.
k. line the subjects up alphabetically and put the first 10 in
group 1, etc.
- We have used as an example the caffeine-reaction time experiment, in which
one group of individuals receives a standard amount of caffeine, the other
group does not, and reaction time is measured.
A possible confound in this experiment is the fact that some subjects may
be "immune" to caffeine because they ordinarily consume large quantities of
it. (a) Describe a situation in which this EV is a confound in this
experiment. (b) Describe how to use matching to control this EV. (c) Explain
WHY matching will control this EV, even though it will still vary in the
experiment. (d) Describe what will happen to the EV if instead of matching it
we just randomly assign subjects to conditions. (e) Compare randomization to
matching with respect to the control achieved over the EV. (f) Describe how to
do this experiment as "within subjects". (g) Tell how to control order and
carryover effects in part f. Draw a diagram to illustrate. (14)
- We have used the term "expected value" regarding correlations (e.g.,
between height and day born the expected value = 0) and also regarding means
(the expected value of the mean TV hours watched is 5.5; the expected value of
the difference between two means is 0). In either case what can we do to make
it more likely that we will get the expected value? Use terms like law of
large numbers, n, haze of values, noise (versus signal). (8)
- State the null hypothesis and the alternative hypothesis for the caffeine
RT study and/or for other example studies from class. Can you identify their
IVs? DVs? EVs? (6)
- Using physical size as an example, explain how we determine whether an
operational definition is VALID. (4)
- Consider a study that examines the relationship between IQ and gpa.
Describe (interpret in words, using the standard format developed in class) a
strong correlation, a weak correlation, and a moderate correlation between the
two variables. (6)
- Why do we need to know "n" to determine how "seriously" we should
take the result of an experiment? (4)
- We NEVER "accept the null hypothesis". Rather we "fail to reject" the null
hypothesis. Why? (Use terms such as signal, noise, law of large numbers, and
haze of values in your answer). (4)
- Draw a rough scatter diagram for a -.78 correlation. (4)
- A group of researchers replicated the study which placed a monkey in a
room with or without a surrogate mother and then measured the monkey's
exploration. The group of monkeys with the surrogate present was tested in the
morning, and the group without the surrogate was tested in the afternoon. (a)
What is the flaw in this design? (b) What would you do to correct it in a
second study? (6)
- A study was done on the effects of caffeine on RT in adults over 30. The
subjects were given either 0mg, 5mg, or 10mg of caffeine. List: (for 3 points
each)
The IV ______________________
Levels_______________________
The DV______________________
3 possible EV's_______________, _________________, _________________
If we were to change the above study to where the subjects were now given
either 0mg, 5mg, 10mg, or 15mg, what would change? (circle the letters of the
correct alternatives) (3)
the IV
number of levels
the DV
nothing would be affected
- In an experiment on caffeine and RT, the subjects will be given 0mg, 5mg,
10mg, and 15mg of caffeine. This will be a within subjects experiment. Draw
and label a Latin square showing how you would counterbalance. (8)
- State both the null and alternative hypothesis for the following "Students
who exercise get better grades than students who do not exercise." (4)
HA: ___________________________________________________________________
H0: ____________________________________________________________________
- Show how you would control for age with matching for the following
participant ages. There are 2 groups (17, 21, 14, 25, 22, 16, 18, 19). State
the steps you would go through, then apply the steps to this data. Show all
work and end up with two columns of numbers, one for group 1 and another for
group 2. (5)
- Explain how you would determine whether your new test of a personality
trait called "sociability" is reliable. How would you determine whether
it is valid?
34 (a) Why can we NOT make a causal statement based on a
correlation? State the 2 problems and give an example of each.
(b) Explain how we solve each of those problems in doing an experiment so that
we CAN make a causal statement regarding the IV's influence on the DV.
35. The volunteers for participating in your experiment include 17
Republicans, 14 Democrats, and 7 libertarians. You want to BALANCE the
factor of political affiliation across the two groups of your experiment.f
a.
the first person is a Republican. How do you assign that person to one of
your groups?
b. as you assign Democrats to the two groups, you reach a point
where group A has 7 Democrats, and Group B only has 3. What do you do
next?
c. two of the political affiliations has an odd number of people?
Is this a problem? How do you solve it?