Rijksuniversiteit Groningenfounded in 1614  -  top 100 university

During the Bachelor Thesis semester, a lot of students working on their Bachelor Thesis visit us. Hence, we have experience with the issues you are dealing with and present you with some advice on how to choose a suitable statistical analysis in an empirical Bachelor Thesis. These guidelines are no official requirements or rules and should not be treated as such.

Based on your theoretical framework, you will have formulated a number of hypotheses regarding the association of two or more associated variables. Take your time to define these variables carefully. Next, identify several methods how you might be able to measure them in your study (i.e., self-report, observation, behavioural measure). Preferably, use or adapt validated instruments rather than creating your own measurement. Think critically whether the measurement you have planned on using will yield reliable and valid data. If possible, pre-test your instruments, especially if you have developed them yourself.

What level of measurement will the resulting data be? In many cases, you will want the (in)dependent variable to be as close to ratio scale as possible. If this level is not reached, you might not meet the assumptions of a number of statistical analyses like ANOVA and regression. Hence, selecting or creating valid and reliable measurements is an important task and failing to do so may result in an inconclusive study.

Now it is time to select a statistical technique to analyze the data your study will yield. This decision should be based on the type of association you are testing and the data itself. Often, there are multiple analysis that can be used, varying in difficulty level. We have made a distinction between statistical analysis that are appropriate for a Bachelor Thesis and analysis that are typically too challenging. This separation corresponds to a large degree with the techniques that are part of the Psychology Bachelor curriculum. As a rule of thumb, it's better to use a 'simple' analysis you are (or should be) experienced with than doing a complex analysis you have little or no practice with.

Appropriate analyses for a Bachelor Thesis:

• Non-parametric tests (e.g., Chi-square, sign test)
• t-tests, AN(C)OVA
• Regression analysis
• simple and multiple regression
• mediator or moderator analysis
• Logistic regression

Difficult analyses for a Bachelor Thesis:

• Factor analysis
• Multilevel analysis
• MANOVA
• Determinant analysis
• Structural Equation Modelling

We hope this very short guide will prove helpful in the methodological and statistical part of your Bachelor Thesis. Taking your time early on in the research process to think about the methodological and statistical part of your study will help the statistical analysis to go smoothly. If you do run into problems though, feel free to visit the Methodology shop on working days between 13:00 and 17:00 in room 125 of the Heymans building (2211).

 Laatst gewijzigd: 11 augustus 2016 16:56