Inferential Statistics

At the end of the course, candidates will be able to, in general terms:

• correctly select from a set of the most important univariate, bivariate and multivariate inferential statistical methods to describe and test characteristics of variables and relationships between variables;
• carry out the most important univariate, bivariate and multivariate inferential statistical analyses using R for statistics;
• correctly interpret and report about output of these univariate, bivariate and multivariate inferential statistical analyses.

More specifically students will be able to:

• explain the role and main assumptions of inferential statistics in the process of scientific research and its relationship with descriptive statistics, and know the main concepts used in the context of inferential statistics;
• construct confidence intervals and perform tests for both proportions and means;
• describe and statistically assess the relationship between one independent variable (dichotomous or nominal) and a dependent dichotomous or nominal variable;
• describe a relationship between one independent variable (dichotomous, nominal and scale) and a dependent scale variable using the linear model;
• construct confidence intervals and perform tests in the context of a bivariate relationship between one independent (dichotomous, nominal and scale) variable and a dependent scale variable using the linear model;
• describe a relationship between several independent variables and a dependent scale variable using the linear model (both in the context of addition and in the context of interaction);
• construct confidence intervals and perform tests in the context of several independent variables and a dependent scale variable using the linear model (both in the context of addition and in the context of interaction);
• assess whether the output of a parametric test should lead to adjusting the model (and the test) used and more generally assess whether the data allow using a parametric test to construct confidence intervals and perform tests in the context of a simple and multivariate relationships;
• construct a test for a mean, the difference between means and the association between scale variables when the assumptions for a parametric test are not fulfilled.

Assumed previous knowledge

It is assumed students are very well versed in the distinction between units and variables; the measurement levels of variables (dichotomous, nominal, scale); the main ‘statistics’ describing variables (‘mean’, ‘standard deviation’ and ‘variance), the ‘standardization’ of variables, and with the (standardized) normal distribution (and the associated ‘empirical rule’). These topics are covered in the course Research Methods and Descriptive Statistics.

Content

In this course the basic notions of data analysis are introduced that would allow to make inferences about populations on the basis of a randomly sampled data set. The course uses the regression (or ‘linear’) model as the basic skeleton and in this context introduces confidence intervals and tests. In addition, it familiarizes students with the logic and implementation of some non-parametric statistical analyses (methods that do not use a concepts like ‘the mean’ and ‘variance’). Usage of these methods is illustrated using research examples. The software used in both teaching and in the assessment is R for statistics.

Assignments and exams

The assessment in this course consists of some assignments and two written exams.
The assignments count for 20% of the final grade, while the exams both count for 40% of the final mark.
The minimum mark for an exam must be at least 5.0.
The individual assignments are mandatory. They are only accepted as a ‘valid attempt’, if you try to seriously answer all questions in the assignments and hand in the assignment before the deadline. There is no minimum grade for the assignments though. If you do not make all assignments and hand them in before the deadline, you will not be allowed to take part in the written exams.
The final grade of the course should be at least a 5.5.

For both partial exams a retake will be offered.
For the assignments there is no graded retake, although you are allowed to repair the omission to hand in a valid assignment by handing in a repair assignment for which you will not get points. 

Time Table

PhDs can choose one of the 3 tutorial groups: