There are two different questions, and you should answer both. The questions use different data sets (mat
files), which can be found on Moodle alongside these test instructions. Use Matlab to analyse the data, and write your answers in a Word file.
In your answers, explain what you are doing. Start from the question and explain which statistical method you used and why. Explain any special choices you made, e.g. options to a Matlab command. Then show the code and output, and explain what the results mean, statistically and interpreted with respect to the question. Also explain anything you have done in Matlab which is not obvious from the code, e.g. if you used a Matlab menu option.
Copy your Matlab commands and their outputs into the Word document, with the output directly following the command. Use a monospaced font (typewriter-style, e.g. Courier) for Matlab code and text output, and a regular font for the explanations. If you want to include plots, you can either use a screenshot tool or copy from the Matlab figure window:
Try to answer both questions since they receive equal weighting. Even if you do not reach a satisfactory answer, explain how you were thinking about the question and what you tried to do (including code and output), because you could still receive partial credit.
Question 1
In a now classic experiment, participants were asked to construct short sentences from lists of words. They performed several trials of this task. Unbeknownst to the participants, they had each been allocated to one of two groups.
In one group, a word associated with being elderly (e.g. “wrinkle”, “grey”) appeared amongst the other randomly selected words on every list, so as to unconsciously prime this concept. For the second group, there was no common concept primed across different trials – the words on each list were simply random.
After this test, participants were asked to walk to another room, where they had been told a second study would be undertaken. In reality, however, there was no second study – the walk to the second room was timed with a stopwatch, in order to assess whether people can be unconsciously primed (by the concept of being elderly) to walk more slowly.
The data are in the file SocialPriming.mat, which contains a table with the same name. The variables are:
priming
the experimental condition a participant was assigned to; “neutral” or “elderly”
score
the time the participant took to walk to the other room, in seconds
- Estimate the mean time taken by each group to walk to the second room, and express the uncertainty in your estimates.
- Did walking time vary between conditions?
- The recent “replicability crisis” has led researchers to question many classic findings from psychology, especially social psychology. A researcher wants to run an exact replication of this study, with 95 % power. How many participants should they run?
Question 2
A cohort of put-upon students are in their first year of a BSc Psychology degree. They are required to take a maths course in preparation for learning about inferential statistics. Their lecturer routinely provides a formative assignment mid-way through term for which she provides feedback, but which does not contribute to the final mark. The final mark instead depends on their score in an end-of-term exam.
The lecturer decides to investigate the effect of feedback on the final test score. For this purpose, she randomly splits the cohort into quarters, and provides different types of feedback for the formative assignment to the different quarters: encouraging, neutral, critical, or none.
Prior to the experiment, she calculated each student’s average entry mark (based on how they did at their age 18 school-leaving exams) to control for any differences in their overall academic ability.
The data are in the file Maths.mat, which contains a table with the same name. The variables are:
participant
an integer number identifying the participant (student)
entry
the student’s average entry mark
feedback
the type of feedback provided on the formative assignment; “encouraging”, “neutral”, “critical”, or “none”
score
final test score from 0 to 100
- Perform an analysis of the data to determine whether the type of feedback affects the final test score.
- The result of the previous analysis only tells us whether there is an effect, not what it looks like in detail. What would you expect – which of the four feedback groups will have the highest mean test score, the second highest, etc.? Write this out as a ranking list. (Regarding this list, there is no right or wrong answer.)
After you have written your list, calculate the mean test score in each of the four groups. Does the result correspond to what you expected?
- Take the feedback types you originally ranked highest and lowest, and check whether the test scores are different between the corresponding groups.