1. Week 3: Literature Reviews and Methodologies
Craft of Research
Week 3: Literature Reviews and Methodologies

2. Literature Review Key Objectives
Gain and maintain readers’ attention
Demonstrate knowledge on topic
Set the context. What does the reader need to know about the topic. How much should you cover?
Claim the relevance of your chosen topic
Identify an area that needs to be addressed by research (gap in knowledge)
Introduce present research

3. BEFORE Writing a Literature Review
Consider the following questions:
Why is this problem important?
How does the study relate to previous work in the area? How will it build upon this work?
What are the primary and secondary hypotheses and objectives of the study?
How do the hypotheses and research design relate to one another?
What are the theoretical and practical implications of the study?

4. Pair and Share from your projects
Consider the following questions:
Why is this problem important?
How does the study relate to previous work in the area? How will it build upon this work?
What are the primary and secondary hypotheses and objectives of the study?
How do the hypotheses and research design relate to one another?
What are the theoretical and practical implications of the study?

5. Basic Organization of Literature Reviews
General Background (1 to 2 paragraphs)
Addressing Specific Area (3+ paragraphs)
Purpose of Current Research (1 to 2 paragraphs)
introducing/announcing present research
Presenting research questions/hypotheses
Clarifying relevant definitions
Summarizing methods
Announcing principle anticipated outcomes
Stating the value of the present research
Outlining the structure of the paper
***these basic sections may be integrated together; They are not mutually exclusive
Culminates to current research questions/aims/hypotheses

6. Literature Review Content
Establishing a Territory
Identifying a Niche
Addressing the Niche
Claiming centrality
Indicating a gap
Introducing/announcing present research descriptively or purposefully
General background
Highlighting a problem
Reviewing previous research
Raising general questions
Presenting research questions/hypotheses
Proposing general hypotheses
Presenting justification
Clarifying definitions
Summarizing methods
Announcing principal outcomes
Stating value of present research
Outlining structure of the paper

7. Example of Literature Table

8. https://thelogicofscience

9. My strategies
What types of articles should you look for when starting a lit review?
Start with Review Articles and Meta-analyses
Original articles by the big names in the field. What is the historical context?
What studies are you building off of? When discussing give more detail.
How to find articles…
What are the key words? Are there other words that mean the same thing also used?
Start with newer articles and find citations to dig deeper into.

10. Reading the results p-values t-tests
Not Going to cover but be aware of
ANOVA’s
Bayesian Statistics
Structural Equations
Post hoc tests and alpha inflation
ICA’s and PCA’s
Correlations
Dealing with missing data
Regressions
ANCOVAs
Chi-squared tests (likelihood)
Confidence Intervals
Partial correlations
Effect Sizes
Power analysis

11. t-tests and p-values
Measures whether or not something is different (Group1 vs Group2) can be used with 2 factor levels only. Differences are in relation to overall variability.
p-value: What is the likelihood that your findings are due to chance? (Scale from 0 to 1).
Arbitrarily set at 0.05 or a 1 in 20 chance. So if someone reports that their results are statistically significant with a p = 0.05 then they are saying…

12. ANOVA – Analysis of variance
Measures whether or not something is different and can be used with 2 or more factor levels. Differences are in relation to overall variability.
Within Subjects factors, Between subjects factors, Mixed design (2x3)
Main and Interaction Effects
Post-Hoc Analysis

13. Post-hoc analysis and alpha inflation
Post-hoc analysis are only done with ANOVAs to tell when there is a main effect with 3 or more levels of analysis, or when there is an interaction effect.
Alpha inflation 1- (1-0.05)^# of tests… So ten tests means alpha now equals 0.4.
Bonferonni correction: alpha/# of tests.

14. Figure 2 A shows the results for letter height.
A significant interaction effect was revealed for size x pace, and pace x group. No other interaction effects were revealed.
For the pace x group interaction effect, post hoc analysis revealed that across both sizes, persons with PD decreased letter height at the fast pace, while HOAs did not.
What is driving this the 1cm size or the 2 cm size for PD individuals?
How many comparisons did I make? What should my alpha level be now? Which should not be displayed as significant. (Bonferonni correction: alpha/# of tests.)

15. Correlations Pearson’s r: ranges from -1 to 1.
Measures covariance of two variables divided by their product of their standard deviations.
Degree to which a pair of variables are linearly related
Does not imply causation

17. Regressions: How well does one variable predict the other?
Is essentially a test of correlation. R^2 how much variance is explained by the predictor variable? How well does it predict?
Can use multiple variables… How do we determine what variables are the most important for predictions?
Linear Regressions (scalar variables), Logistic Regression (Categorical variables, See chi-square tests).

18. Multivariate Analysis
Variable
Univariate Analysis
Multivariate Analysis β p
95%CI
Demographic
Gender
-.37*
.05
-.73, -.01
-.52**
.01
-.87, -.18
BMI
-.13
.50
-.51, .26
-.03
.83
-.35, .28
Marital Status
.33
.08
-.04, .69
.32*
.03
.32, .61
Education Level
.04
.90
-.35, .43
.30
-.00, .61
Employment Status
.18
.34
-.20, .56
-.08
.61
-.38, .23
Age
.07
.71
-.32, .46
-.04
.76
-.33, .25
Motivators
Enjoyment
.48**
.14, .82
.36*
.02
.06, .65
Emotional Health
.29
.16
-.13, .51
Family
-.31
-.76, .15
Independence
.22
.25
-.16, .59
PD Health
-.19
-.59, .21
Spouse
-.25
-.78, .28
Social
.19
.36
-.23, .60
Friends
-.15
.57
-.70, .40
Physical Health
-.01
.98
-.44, .44
Barriers
Health
-.43*
-.83, -.03
-.41*
-.70, -.13
Falling
-.35
.06
-.72, .02
-.41**
-.75, -.08
PD Symptoms
.13
-.71, .10
Weather
.20
-.22, .62
Access
.35
-.23, .61
Financial
.37
-.63, .24
Dislike
-.12
-.54, .30
No Partner
.63
-.63, .39
Time
-.10
.67
-.58, .38
Personal Factors
Active Lifestyle
.26
-.19, .69
PD Worry
-.20
-.60, .19
Health Worry
.11
-.29, .51
Active Social
.72
-.39, .56
Active Spouse
.78
-.42, .56
Active Family
.95
-.39, .42
Table 2: Linear Regression. Factors associated with time spent exercising (minutes/week).

19. Variable
Chi-Square p
Demographic
Gender
4.84*
.03
BMI
.08
.77
Marital Status
5.60*
.02
Education Level
0.41
.84
Employment Status
1.08
.30
Age
0.82
.36
Motivators
Enjoyment
0.97
.32
Emotional Health
0.45
.50
Family
0.98
Independence
1.68
.19
PD Health
0.46
Spouse
0.05
.82
Social
0.54
.46
Friends
0.09
Physical Health
0.64
.42
Barriers
Health
0.04
.85
Falling
4.29*
.04
PD Symptoms
12.45**
<.001
Weather
5.06*
Access
0.20
.66
Financial
2.15
.14
Dislike
0.22
.64
No Partner
13.39**
Time
0.08
.79
Personal Factors
Active Lifestyle
0.02
.88
PD Worry
2.26
.13
Health Worry
0.56
Active Social
0.11
.75
Active Spouse
3.45
.06
Active Family
Table 3: Logistic regression showing factors that predict those who meet exercise recommendations (>150 per week).

20. Effect Sizes
Cohen’s d (effect size of mean differences ANOVA, t-tests) = difference in means divided by the pooled standard deviation.
Small = 0.3, medium =0.5, large = 0.7
Pearsons r
Small = 0.1, medium = 0.3, large = 0.5
Hedges g (meta analysis) – takes into account sample size from each study
Same as Cohen’s d
Effect size calculator (

21. Power Analysis
How big of a sample size do you need to detect an effect of a given size with a given degree of confidence
Power or beta is usually set to 80% or 0.8. (Again this is an arbitrary number).
There are many different calculators out there.
GPower

22. Danger of type II error: n=10

23. Power Analysis: Repeated Measures ANOVA
Β= 0.8, α = 0.05,
r and Cohen’s d will be estimated using the results from other studies for optimistic estimate. Conservative estimate using a smaller r and Cohen’s d.
r and d in lit review
Optimistic
Conservative
Cortisol
d = 1 , 1.2, r = 0.76
N = 16 (d = 1, r = 0.76)
N = 46 (d = 0.5, r = 0.3)
Digit Span Backwards
d = 0.77, r = 0.67
N = 12 (d = 0.77, r = 0.67)
Auditory Habituation
d = 0.67, r = 0.44
N = 22 (d = 0.67, r = 0.44 )
UPDRS
d = 0.725, r = 0.63, 0.75
N = 14 (d = 0.725, r = 0.6)

24. Power Analysis: Multiple Regression
Computed with Power and Precision: Power (1- β) = 0.8, α = 0.05
No studies have examined the relationship between working memory or auditory habituation and UPDRS motor scores.
The smallest covariate r found for individual measures in UPDRS is r = 0.6 (overall motor score: r = 0.89).
Optimistic: (covariate r = 0.6, predictor r =0.5) suggests that we need 16 participants.
Conservative: (covariate r = 0.5 predictor r =0.3) suggests that we need 60 participants.

25. Confidence Intervals
A range of values that you can be certain contains the true mean of the population.
Large sample sizes will give you smaller confidence intervals.

26. Reading the results P-values T-tests
Not Going to cover but be aware of
ANOVA’s
Bayesian Statistics
Structural Equations
Post hoc tests and alpha inflation
ICA’s and PCA’s
Correlations
Dealing with missing data
Regressions
ANCOVAs
Chi-squared tests (likelihood)
Confidence Intervals
Partial correlations
Effect Sizes
Power analysis