Week 6 - Quantitative Research Methods

Question 1

What are the two major data analytic approaches in psychological research?

Answer 1

• exploring relationships between variables

- comparing groups

Question 2

what analyses are used to conduct statistical analyses to explore relationships?

Answer 2

• correlation / partial correlation
  structure: cluster analysis, factor analysis, multidimensional scaling (MDS)
  prediction: multiple regression (continuous), logistic regression (dichotomous)

casual links: modelling techniques (path analysis, confirmatory factor analysis, structural equation modelling (SEM), hierarchical linear modelling)

Question 3

what analyses are used to conduct statistical; analyses to compare groups?

Answer 3

• t-tests
• analysis of variance (ANOVA)
• analysis of covariance (ANCOVA) / multivariate analysis of variance (MANOVA)
• discriminant analysis

Question 4

what is the difference between univariate statistics and multivariate statistics?

Answer 4

• multivariate statistics involve multiple IVs and/or DVs (outcome/s)
• univariate statistics involve a single DV (outcome)
• UNI: correlation, t-test, ANOVA, ANCOVA
• MULTI: MANOVA, discriminant function analysis, multiple regression, logistic regression, canonical correlation, cluster analysis, factor analysis, MDS

classification not very strict

Question 5

what is the general linear model (GLM)?

Answer 5

• an approach to assume and test a linear relationship between IVs (predictor) and DV. In GLM, the degrees to which
  (1) a predictor is associated with DV, and
  (2) a set of predictors as a set explains the variance in the DV (outcome), are examined.
• straightforward to choose a statistical procedure suitable for a study with a clear research design, somewhat difficult cases are encountered when choosing between ANCOVA, Regression, Mixed-design ANOVA, and MANOVA

Question 6

what is classified as the GLM Family?

Answer 6

DV

• Single (continuous, categorical)
• Multiple (continuous)

IV (and covariate)

• Categorical
• categorical and continuous
• continuous

ANOVA (single/categorical)
ANCOVA (single/categorical and continuous)
REGRESSION (single/continuous)

(not GLM but similar) LOGISTIC REGRESSION (single/categorical and continuous)
DISCRIMINANT FUNCTION (single/continuous)

MANOVA (multiple/categorical)
MANCOVA (multiple/categorical and continuous)

Question 7

what are the research objectives in quantitative data analysis?

Answer 7

• measure group means and compare between groups
• test casual models
• assess relationship among variables
• assess structure in and across complex relationships

Question 8

what are the analytic methods in quantitative data analysis?

Answer 8

• t-test (35% Hons Thesis)
• ANVOVA (70% Hons Thesis)
• ANCOVA (70% Hons Thesis)
• MANOVA (70% Hons Thesis)
• Discriminant analysis
• correlation (50% Hons Thesis)
• reliability analysis (80% Hons Thesis)
• multiple regression (25% Hons Thesis)
• logistic regression (<5% Hons Thesis)
• path analysis (10% Hons Thesis)
• SEM (10% Hons Thesis)
• PCA and Factor analysis (10% Hons Thesis)
• cluster analysis, MDS

Question 9

Multiple regression and logistic regression analysis

Answer 9

• Multi-causality
• Unique contributions of predictors (partial stats, ect)
• Interaction effects, too!

non-parametric = categorical

assumption: distribution of errors is normally distributed

Question 10

what is the principal component analysis (PCA) and the exploratory factor analysis (EFA)

Answer 10

• PCA: how can we combine these variables into a smaller set of variables? explaining components (big list to represent happiness, want to reduce items)
• EFA: what is the underlying structure of these variables? how they relate to each other (levels of happiness, different factors and structure of measure) exploratory

Latent Factor 1
Latent Factor 2

Question 11

what is confirmatory factor analysis (CFA)?

Answer 11

CFA starts with a theory to guide your analysis
used to confirm
• Distinctions are made between latent constructs that are error free, and their indices that involve error. (e.g., authoritarian parent {construct} and smacking {indices/error})

The analysis informs about an overall ‘fit’ of the model

Question 12

what is a path analysis (mediation analysis)

Answer 12

Path Analysis examines the causal structure among the variables.

indirect effects
direct effects

Question 13

what is the structural equation modelling (SEM)?

Answer 13

• combination of CFA and PA
• measurement model = CFA
• structural model = PA (causal)
• models fit is analysed

Question 14

what is cluster analysis and multidimensional scaling (MDS)?

Answer 14

• clustering assigns variables, or cases, that are similar to one another in groups
• MDS is a method of visualising patterns of relationship among variables or cases based on similarities