student number______
Psychology 403.Laboratory in Cognitive PsychologySpring, 2003
Quiz
Instructions: Write your student number, but not your name, in the blank at the top of each page. Write your answers as concisely as possible.
This quiz has 5 questions, 3 multiple choice or short answers worth 1 point each, a short answer question worth 2 points, and a short essay questions worth 4 points, for a total of 9 points.
Three Very Short Answer Questions- 1 point each
1)Cohen's d, which measures the size of an effect,:
a) is measured in units equal to the standard deviation of the population.
b) can only be assessed using a t test, not a z test.
c) is larger the greater the significance of a t test comparing two means.
d) depends upon the size of the sample.
2) Reaction time is measured in units that have ______properties.
a) nominal
b) ordinal
c) interval
d) ratio
3) Compare and contrast independent variable, dependent variable, and control variable.
Short Answer Question- 2 points
4) If there is a true difference between a control and experimental group, why does increasing sample size in an experiment make it more likely that a statistically significant result will be obtained? (Answer this a precisely as possible.)
Short Essay Question- 4 points
5) The first experiment on the discrimination of line length attempted to illustrate the concept of a speed-accuracy trade-off. Explain what a speed-accuracy trade-off is, and why we cannot interpret reaction time data without also considering the accuracy results at the same time.
Psychology 403.Laboratory in Cognitive PsychologySpring, 2003
1st Midterm Exam
Instructions: Write your student number, but not your name, in the blank at the top of each page. Write your answers as concisely as possible.
This exam has 8 multiple-choice and very short questions worth 1 point each, 3 short answer questions worth 2 points each, 3 longer questions worth 4 points each and one question that must be answered using Excel, worth 6 points, for a total of 32 points.
Multiple Choice and Very Short Answers- 8 questions worth 1 point each
1) A researcher has manipulated only one independent variable at two levels, has held constant as many other variables as possible and has balanced individual differences by using random assignment. The researcher is likely to be able to claim that the independent variable caused the observed changes in the dependent variable because the experiment is
- externally valid.
- reliable.
- sensitive.
- internally valid.
2) Null hypothesis significance testing uses the laws of probability to estimate the likelihood of an outcome assuming that
- the null hypothesis is false.
- an effect is present.
- the independent variable worked.
- only chance factors caused the outcome.
3) Why is a confidence interval also called a “margin of error”?
4) If a researcher were to use a .01 level of significance rather than the .05 level, the probability of a Type I error (falsely rejecting a true null hypothesis) would
- not change.
- increase.
- decrease.
- not able to tell without more information.
5) What is meant by confounding in an experimental design?
Suppose I describe an experimental design as a 3 x 4 within-group (repeated measures) design.
6) How many independent variables were there? ______
7) How many different conditions were there? ______
8) How many interactions would I have to report to describe all of the interactions? ___
Short Answers- 3 questions worth 2 points each
9) Describe briefly how confidence intervals can be used to determine if the means for the two conditions of an experiment differ.
10) Institutional Review Boards consider two important concepts when reviewing proposed research using human participants: informed consent and the cost/benefit (or risk/benefit) ratio. Explain each of these concepts and the role that it plays in the review process.
11) Describe the information that goes into the numerator and denominator of F for a one-way, independent groups ANOVA, and the value of F that you expect to obtain if there is not a statistically significant effect.
Longer Answers- 3 questions worth 4 points each
12) In the figure above, use the axes which are given to draw the following, making your drawing as accurate as possible:
a)the main effect of Factor (variable) A
b)the main effect of Factor (variable) B
c)another graph of the interaction, with B as the variable and A as the parameter distinguishing the two curves.
If the 95% confidence intervals are no larger than the size of the data points,
d)is there a statistically significant main effect of A? ______
e)is there a statistically significant main effect of B? ______
f)is there a statistically significant interaction? ______
13) Suppose that you wanted to determine whether or not there are gender differences in Mental Rotation. Describe the basic experimental design you would employ, identifying the relevant independent and dependent variables. Then describe the null hypothesis that would be most directly relevant to this question. Finally, describe the pattern of the data that you would examine, and the statistical test(s) that you would want to see carried out to reach your conclusion. Describe specifically the pattern of results (and relevant statistics) that would persuade you that there is a gender difference in Mental Rotation.
14). In a t-test, what are the relationships among: a) the alpha level, b) the critical value of t, c) the obtained value of t, and d) the value of p associated with the value of t. In your answer, concisely explain what each of these is, and how they relate to each other.
1 Excel Question worth 6 points
15) On your disk you will find a file entitled Mid1.xls that contains the results of the Sternberg experiment on searching short-term memory. Use that file to answer this question. For the portions of this question that have to be answered using Excel, please put all of your work on the same page of the Excel workbook so that I can print a single page to grade your answer.
a) Write your student number onto the label on the disk.
b) Type your student number (not subject number) into the yellow box in the upper left-hand part of the spreadsheet.
c) (.5 points) Calculate the means for all of the conditions for both reaction time and accuracy.
d) (.5 points) Create tables to display the data as clearly as possible that illustrate the pattern of results that would be of interest to someone trying to decide which hypothesis about how items in short-term memory are searched is most plausible: parallel, unlimited capacity; serial, self-terminating; or serial-exhaustive search. Label your tables appropriately.
e) (2 points) Create the graphs to display these results, again focusing on the format that makes it easiest to see the patterns that will make discriminating between these hypotheses possible. Make the graphs be as good as possible. Include the linear regression lines, with equations, that will help in deciding among the alternative memory search hypotheses.
d) (1.5 points) Given the data that you have, describe the statistical analyses that you would undertake (or wish to have someone do for you if they cannot be done in Excel) to support your interpretation of the data. State explicitly what you would do. That is, describe the test that you would use and the specific data that you would use it on, and how you would interpret the outcome of the test. (Be sure to include descriptions of the analyses of both the Reaction Time and Accuracy data.)
e) (.5 points) Do the appropriate t-test to compare the slopes of the probe present and probe absent conditions, and write a statement that reports the test as you would if you were trying to determine whether the target absent intercept differed from the target present intercept, with no prior prediction about the expected direction of difference (to show me that you know how to do t-tests).
f) (1.0 points) Discuss how consideration of the Accuracy data influences or qualifies the conclusions that you might draw based on consideration of the Reaction Time data.
Psychology 403.Laboratory in Cognitive PsychologySpring, 2003
2nd Midterm Exam
Instructions: Write your student number, but not your name, in the blank at the top of each page. Write your answers as concisely as possible.
This exam has 8 multiple-choice or very short questions worth 1 point each, 4 short answer questions worth 2 points each, 2 longer questions worth 4 points each, and 2 questions that must be answered using Excel, worth 8 points, for a total of 32 points.
If you have studied something very hard and I have not given you a chance to demonstrate your knowledge on the exam, you may replace one question in the exam with a question of comparable difficulty. Write the question and then the answer. Use the back of the first page for your "replacement" question and answer, indicating the question that it replaces.
Multiple Choice and Very Short Answers- 8 questions worth 1 point each
1) Referring to the data in the figure above, what is the best prediction about the result of a one-way, independent groups ANOVA assessing the effect of the independent variable?
a) It would definitely be statistically significant.
b) It would probably be statistically significant.
c) It definitely would not be statistically significant.
d) It probably would not be statistically significant.
The Figure above, represents hypothetical individual data from Males and Females, whose memories were each tested at two different times.
3) Is this an independent groups or repeated measures design? (Check one.)
___independent groups ___repeated measures
4) Suppose that I had an hypothesis that predicted that males would have a lower memory score than females at Time 1, but that they would be equal at Time 2. How could I use the proper 2-way ANOVA to assess the truth of the hypothesis? (That is, if I had F and p values for the main effects of Time of Testing and Gender, and for the Time x Gender interaction, which would I look at and what outcomes would be consistent with my hypothesis?)
5) Suppose that two different investigators have conducted an identical experiment involving two groups, except that investigator A has used 25 participants/group and investigator B has used 100 participants. If we know (but they do not) that the true state of the world is that the treatment that distinguishes the experimental group from the control group does have an effect, so that the populations from which the two groups are drawn differ in the distribution of the dependent measure, what can we say about the likelihood that the appropriate t-test will show that the two groups differ significantly.
a) We have no basis for making a prediction.
b) Investigator A is more likely to find a statistically significant difference than investigator B.
c) Investigator B is more likely to find a statistically significant difference than investigator A.
d) Both investigators will always find a statistically significant difference.
6) Given the data above, what can you confidently say about whether the Time x Gender interaction in a 2-way ANOVA would be significant?
a) It will be statistically significant.
b) It will not be statistically significant.
c) Without running the actual test it is not possible to judge simply by looking at these data.
d) It is not possible to answer without knowing whether it is an Independent Groups or Repeated Measures ANOVA.
7) Suppose that I have designed an experiment that requires that people participate in two sessions, separated by one week, and that they receive payment on an hourly basis. Do you think that the Institutional Review Board would or would not permit me to say that participants who did not return for the second session would not receive compensation for their participation in the already completed first session. Explain what guiding principle about the use of human participants would lead you to expect them to reach that conclusion.
8) When examining data from an independent groups design with a one-way analysis of variance,
2)what properties of the data does the numerator of F depend on?
b) what properties of the data does the denominator of F depend on?
Short Answers- 4 questions worth 2 points each
9) If you were given the opportunity to make one change in this course to improve it, what would you recommend? Your suggested improvement should be intended to increase mastery of the course material by the students. (Any answer offered in good faith will receive full credit.)
10a. Compare and contrast the sources of information about "error" or "random" variance used when conducting an F test as part of an Analysis of Variance in the following two situations;
a) analyzing an independent groups experimental design and b) analyzing a repeated measures experimental design, including in your answer definitions of each design.
OR
10b. Describe the source of variance that is used in the numerator of an F to test for an interaction in an independent groups Analysis of Variance.
11) According to the textbook, “Controlling extraneous (control) variables by holding them constant leads to more sensitive experiment with less external validity, whereas controlling extraneous variables by balancing them leads to lower sensitivity but increased external validity.” Explain what the authors meant by this claim.
12a) I have conducted an experiment to evaluate a new drug that is alleged to improve memory performance. The drug is so expensive that I will consider using it only if it improves memory performance by at least 10%. I ran a group of subjects under two conditions. They learned two sets of items matched for difficulty, one after taking the drug and the other without the benefit of the drug. The order in which the conditions were run was counterbalanced across subjects. I conducted the appropriate t-test to compare the mean number of items recalled in the two conditions and found that the mean number of items recalled after taking the drug was larger than the number recalled without the drug, t(37)= 5.48, p= 0.0000032. Should I be persuaded to use the drug? Why or why not?
The textbook has an extensive discussion of experimental designs, including various techniques for randomization and counterbalancing in the chapter on Repeated Measures Designs. Why is there such concern about these issues in repeated measures designs? That is, what problems are they trying to avoid by using these designs. Pick one example of one of these techniques and describe how it reduces the problem that requires this kind of solution.
Longer Answers- 2 questions worth 4 points each
Questions from Prof. Hambrick
1.3a) What is an interaction? Your answer should include two parts: First, provide a general definition. Second, illustrate the concept of an interaction in terms of an experiment designed to investigate state-dependent memory--that is, the idea that memory is best when one's internal state (e.g., arousal level) when attempting to retrieve some information is the same as was when that information was encoded. Assume that the experiment occurred in two phases separated by 24 hours—an encoding phase in which subjects studied a list of words and a retrieval phase in which they attempted to recall the words. In addition, assume that arousal level (low vs. high) was manipulated at both encoding and retrieval. Describe a 2 x 2 interaction that would support the hypothesis of state-dependent memory.
OR
13b) Students in introductory statistics courses are often reminded that correlation doesn't imply causation. However, in class, we discussed the idea that significant differences in experiments also do not necessarily imply causation. That is, just because experimental manipulation of some independent variable has an effect on some dependent variable doesn't necessarily justify a causal conclusion. Unpack this statement. In particular, to what extent are causal conclusions justified in research, regardless of whether the research is correlational or experimental?
13. What is statistical power? Use the figure above to discuss how increasing the sample size (all other things being equal) gives you an increase in power. Be sure to label all of the distributions, add whatever else is required to show what corresponds to statistical power, and then explain how the two panels A and B relate to the effect of increasing sample size.
2 Excel Questions worth 8 points
The file labeled Mid2ExcelS02.xls contains the information for the following two questions. It has 2 different pages in it. Please type your student number in the upper left-hand cell of each of the spreadsheet pages and write your student number on the disk label.
15) (2 points) The page labeled tTestIllustrationConInt.xls contains the illustration of the workings of an independent groups t-test that was on the course website. Modify the data (the individual subjects scores on the relevant condition(s) in a way that results in the absolute value of t increasing to 4.48±.2 without changing the denominator in the calculation of the t statistic, and save the file with the new data values.
16) On the disk there is also a page entitled Mid2DataFile, which contains a set of data from an imaginary experiment comparing the results of a memory training program. There were two groups of subjects, those who had low GPAs and those who had high GPAs. Both groups were tested before and after utilizing the memory training exercises. The experimental design was a 2(groups of subjects) x 2 (before and after memory training). The data were examined using a mixed ANOVA.
a) Draw a graph to represent the results of the experiment in a manner, which makes it possible to see the pattern of results for both groups of subjects before and after training. Please draw the graph on the same Excel Workbook page as the data.
Assume that the mixed ANOVA yielded the following results:
Main effect of training: F(1,18)=44.31, p<.001
Main effect of GPA: F(1,18)=4.76, p=.04
Interaction of training x GPA: F(1,18)=.68, p=.42
Referring to the data, your graph and to the statistics:
b) What would you conclude about whether the effect of training depends upon the GPA of the subjects, supporting your statement by reference to the appropriate data and analyses by including the specific values of the means that are relevant as well as the specific results of the relevant F test.