Statistical Analysis and Research Design

Cronbach's Alpha

Please use your own words if referring to textbook definitions and please use references if appropriate.  Please type your answers in the boxes below.

1)Cronbach’s alpha (4 marks)

2)An interaction effect (as might be found in a factorial ANOVA) (4 marks)

3) Theme (as per Thematic Analysis, with reference to what a theme is and how the researcher might determine a theme from a dataset) (4 marks)

4) Levene’s test (4 marks)

Multiple Choice

Please UNDERLINE THE SINGLE STATEMENT IN BOLD FONT which you feel best completes the sentence/answers the question.  (2 marks each)

5) When using ANOVA, what is the effect of an Independent variable (IV) (or Factor) on a Dependent variable (DV) known as?
A)A Direct effect.
B)An Interaction effect.
C)An Indirect effect.
D)A Main effect.
E)A simple effect.

6) The statistic ‘R²’ in linear regression analysis:
    A) May be ‘adjusted’ and then generalised to the population
    B) Refers to the difference between group scores
    C) Is the variance in the outcome/criterion variable that is predicted
    D) Both A and C
E) Tells the researcher if findings are statistically significant or not

7) In relation to obtaining ethical approval for research:
A) Your study must be approved the very first time you submit to the ethics committee with no chance of resubmission
B) You must wait for ethics committee approval before beginning any data collection or even approaching potential participants
C) Pilot studies do not need ethical approval
D) You may obtain data before ethical approval if conducting the study online
E) All of the above

8) Factor Analysis:
    A) Needs correlations to exist in the data
    B) Can help distinguish groups of questionnaire items that ‘fit’ together
C) Is a process of data reduction
D) Can be both exploratory and confirmatory
    E) All of the above

9) In a bottom-up qualitative thematic analysis, the researcher should aim to:
    A) Code all data to the point of ‘saturation’
    B) Reject the null hypothesis
    C) Infer cause-effect between variables
    D) Generalise findings to the population
    E) Apply pre-determined codes to the data

10)  To analyse data from a two-factor unrelated design testing for potential effects of participant pet ownership (yes/no), as well as gym use (yes, no or occasionally) on self-reported neuroticism score, you would conduct:
    A) A 2 (pet ownership) by 4 (gym use) Repeated measures ANOVA
    B) A 2 (pet ownership) by 2 (gym use) Unrelated ANOVA
    C) A 2 (pet ownership) by 3 (gym use) Unrelated ANOVA
    D) 2 One-way ANOVAs   
    E) None of the above

11) How do you check whether data has met the assumption of Sphericity in a repeated measures ANOVA:
    A) Interpret whether Mauchly’s test of Sphericity is p>.05
B) Just report the Greenhouse-Geisser correction to be cautious
    C) Checking Levene’s test for p>.05
    D) Examine the descriptive statistics
    E) Both A and B

Interaction Effects in Factorial ANOVA

Section C: Interpreting Statistical findings
A study tracked depression scores over time in a group of 53 self-selecting individuals who were not receiving any treatment.  Depression was measured using the Beck Depression Inventory (BDI, Beck, 1961) where higher scores represented greater reported depression.  

BDI scores were taken at the study outset (baseline), at 3 months and again at 12 months.  The research team tested the first hypothesis that there would be an improvement in BDI scores over time with no intervention.    
Tests of Within-Subjects Effects

Based on estimated marginal means
*. The mean difference is significant at the .05 level.
b. Adjustment for multiple comparisons: Bonferroni.

Questions 12-16 are based on the study SPSS output directly above and on the previous page (4 marks each unless stated).

12) Name in full the statistical test used that has produced the SPSS output here (2 marks).

13) Name the ‘independent variable’ (or factor) and each of the ‘levels’.  Please also identify and precisely name the dependent variable.

14) Can Sphericity be assumed in this study?  Please explain your interpretation.

15) Was the first hypothesis supported? (Please explain/interpret the findings in full sentences using statistical notation i.e. f-ratio, dfx2 and significance/p-value).

16) In testing pairwise comparisons (Bonferroni post-hoc test), which BDI mean score comparisons were significantly different from each other over time?  Please also refer to the direction of any significant differences.

As part of the same study, optimism ratings (from the life orientation test revised, LOT-R, Carver et al., 2010; higher scores = more reported optimism) and amount of physical exercise (in hours, per week) were also captured when the study commenced.  

Researchers then tested the second hypothesis that levels of optimism (from the LOT-R) and amount of physical exercise would predict BDI baseline scores using a different statistical test.  

a. Predictors: (Constant), Physical exercise (in hours) per week, Optimism score
           
a. Dependent Variable: Baseline BDI Depression Scores
b. Predictors: (Constant), Physical exercise (in hours) per week, Optimism score

a. Dependent Variable: Baseline BDI Depression Scores

Questions 17-22 are based on the SPSS output directly above (4 marks each unless stated):

17) From examining the SPSS output for the second hypothesis, please name in full the statistical test that was conducted (2 marks).  

18) Please write the formula for a regression line below (for a single predictor variable).

Then define each component of the formula with reference to the research above (you may choose to cite relevant values from the coefficients table)

19) Please then interpret if the research team’s second hypothesis was supported (2 marks each).  

a) How much variance in BDI scores was predicted?

b) Does the regression model significantly predict outcome? (include any relevant statistics here to support your interpretation – please write in a sentence)

20) Which was the strongest individual predictor of variance in BDI scores and how can the relationship be explained?  (examine the output closely to determine this and please answer in a full sentence, citing relevant statistics, e.g. Beta coefficient, p-value)

21) Did the second predictor variable significantly predict variance in BDI scores (please answer in a full sentence, citing relevant statistics)

22) Please answer either part (a) or (b) (both 4 marks).

a) If physical exercise was theorised to mediate (i.e. explain) the relationship/pathway between optimism and BDI scores, please outline briefly one way in which this could be tested statistically (4 marks).  

b) Please consider some implications of the second part of the study.  Could you suggest anything that might have a positive impact on BDI scores in those reporting being affected by depression?  And/or outline briefly some future research possibilities? (4 marks)

A health psychologist has received funding to design a study to see how exercise could be promoted in adults working in offices.  Your task is to design a study to explore this and fill in the APA report sections starting below.  You may find it helps to plan the study on a separate piece of paper before beginning to answer.

Think about how you could promote exercise.  This may involve testing a particular intervention which may potentially increase or initiate exercise.  Think how you could test and measure whether any intervention has had any effect.  

Alternatively, you could consider some psychological and demographic (e.g., age) factors you think will influence exercise.  Study findings may then indicate where an intervention to promote exercise could be targeted.  Do also consider very briefly how such an intervention might be implemented as you consider your results.   

Perhaps a qualitative approach would be just as beneficial?  This could involve office workers who already exercise.  What could you learn from such individuals?

You may wish to use references in places (e.g., for questionnaires) for this section.

Do be guided by the marks available per question in terms of how much you should write for each section (e.g., one mark could be gained by one relevant sentence).