- can describe different data collection designs (experimental, correlational; cross-sectional versus longitudinal; one-way versus factorial designs) and know which multivariate techniques, including multilevel models, are appropriate given the data collection design and the research questions envisaged.
- can describe and/or explain the key concepts in inferential statistics within the context of multivariate analysis – including but not restricted to (experiment-wise/family-wise) Type I errors, Type II errors, standard errors, sampling distributions, test statistics, confidence intervals, effect sizes, and power.
- can explain the difference between zero-order relationships, partial and part relationships; this includes explaining the concepts at a conceptual level as well as within the context of multivariate techniques (e.g., regression analysis). This includes a thorough knowledge of the concept of (multi)collinearity and confounding.
- can describe the general linear model (GLM) and know how the acquired classical multivariate approaches fit within this general framework.
- can describe the model assumptions underlying the acquired multivariate approaches, including multilevel models, and know how to test these assumptions in real data.
- is able to summarize multivariate data using descriptives (e.g., means, correlation matrices, principal component anlaysis) and graphs (e.g., scatterplots) and is able to screen the data for multivariate outliers (e.g., using the Mahalanobis distance), and to evaluate whether the data are (multivariate) normally distributed.
- is able to apply linear regression models to complex data sets (i.e., containing variables of different measurement levels) and for a variety of research questions; this includes the use of dummy variables, testing and probing of moderating effects, modeling non-constant errors, modeling non-linear relationships, accounting for non-normalities (e.g., data transformation), and examining multicollinearity.
- can apply the classical multivariate methods from the family of general linear models, in particular MANOVA, as well as binary and multinomial logistic regression to (real) data sets from simple and factorial designs, and is able to summarize the results in a correct and comprehensible manner, and in accordance with the APA guidelines. In particular, students are able to combine sample results, statistical significance, and relevant (standardized) effect-size measures to reach sound substantive conclusions regarding the original research questions.
- is able to evaluate most important model assumptions underlying the acquired multivariate appraoches.
- is able to assess, statistically test, and describe zero-order, partial, and part associations within the context of the applied multivariate technique.
- can apply the basic multilevel models (random intercept model, random slopes model) and several of its extensions (e.g., three-level data; logistic multilevel analysis).
- is able to transform substantive research questions into research hypotheses that can be addressed by using the aquired techniques multivariate techiniques; this includes the choice of the technique as well ass formulating tests of contrasts (orthogonal versus non-orthogonal); post-hoc comparisons.
- is able to critically reflect upon the advantages and disadvantages when different multivariate strategies can be applied to address the same research question;
- can detect important flaws in an analysis and clarifying and/or explaining unexpected outcomes and are able to suggest sensible improvements. This includes detection of flaws in the research methodology (e.g., inaccurate sampling design), statistical analysis (i.e., use of less than optimal techniques). Suggested improvements should be realistic and ethically sound.
- General set up: The course consists of 14 two-hour lectures and 12 two-hour lab sessions. In the interactive lecture, the lecturer explains the subject matter, asks questions, and invites students to discuss the subject matter. In the lab sessions, the students apply the acquired methods and techniques to real-data sets from the Social Sciences and Social Psychology using SPSS.
- Study load: The 168 hours in the course consist of 28 (14 × 2) hours attending lectures, 28 (14 × 2) hours attending lab sessions, 32 (2 ×16) hours working on the assignments, and 80 hours self-tuition.
- Practicals: Attendance at the lab sessions is required. Students who fail to attend a lab sessoin three times or more need to complete an alternative assignment to fulfill the practicals requirement.
The final grade is composed of two parts:
- Written exam with open-ended questions assessing knowledge transfer, application of knowledge, and critical judgment
- Four practical assignments assessing application of knowledge and critical judgment. The resit of the assignment covers the four individual assignments; it is not possible to do the resit for individual assignments. Results for the individual assignments only remain valid for the current academic year.
The final grade equals 1/4 times the average grade of the four assignments plus 3/4 times the grade of the written exam.
Only for students who are qualified for the Research Master.
Researchers in the social sciences increasingly are confronted with the analysis of complex data sets, consisting of large numbers of variables often having different measurement levels and the data may be nested. This course provides an introduction into the classical multivariate methods, in particular regression models, starting with the linear regression model (and special topics such as transforming data, dummy variables, unusual and influential data, diagnosing nonlinearity, non-constant error variance and non-normality, and collinearity), logistic regression, multinomial logistic regression, and nonparametric regression. The course also discusses regression analysis for nested data, (i.e., multilevel analysis), including the analysis of longitudinal data. Other methods discussed are factor analysis (briefly; also part of courses 5 en 7) and principal components analysis, discriminant analysis, (M)ANOVA (factorial designs, repeated measurements, contrasts), and log-linear models. In the practical sessions, and for the practical assignments, the students apply the acquired methods and techniques to real-data sets using SPSS and write a report on the main findings.
Type of instructions
Lectures (2 hours) and Compulsory Practicals (2 hours)
Type of exams
Written exam and assignments.
|Written test opportunities|
|Schriftelijk / Written||EXAM_01||BLOK 2||1||12-12-2019|
|Schriftelijk / Written||EXAM_01||BLOK 2||2||16-01-2020|
|Written test opportunities (HIST)|
|Title||:||Applied Multivariate Statistics for the Social Sciences (6th ed.)|
|Author||:||Keenan A. Pituch, James P. Stevens|
|Hand-outs and additional readings will be made available throughout the course. These hand outs are part of the compulsory reading materials for the exam.|
|4 Papers (average score)|