Logo Utrecht University

S4 conference

Pre-conference lectures

We will start the conference on June 8 with four lectures to set the stage for the rest of the conference
by introducing some basics. Registration is for the combination of the lectures (including lunch).

 

Reaching Hard-to-Reach Populations & Achieving High Measurement Quality: Insights from Survey Methodology
Researchers working with hard-to-reach populations are often faced with the challenge of small sample sizes. Each participant in this setting is precious. On the one hand, sampling and contact strategies should be optimal to reach as many participants as possible. On the other hand, the measurement instrument (e.g., the questionnaire) should be thoroughly developed to achieve high measurement quality. After all, a unit or item nonresponse weighs heavier in small samples than in large-scale surveys.

Survey methodological research provides a rich literature on multiple challenges that might arise when collecting data with small samples. This workshop will discuss different strategies from survey methodology on how to access hard-to-reach populations (e.g., sampling approaches, contact strategies), how to deal with issues of social desirability (e.g., mode selection, question formulation, interviewer training), and how to design a valid, reliable, and comparable questionnaire (e.g., questionnaire design, question formulation, pretesting approaches).

Single-case experimental designs: Studying treatment effectiveness and mediators of treatment outcomes
Single-case experimental designs (SCEDs) are extremely useful to investigate intervention impact on individual clients’ functioning, and working mechanisms of change. In this talk: (i) definition and basic characteristics of SCEDs are provided; (ii) statistical mediation analysis is placed in a single-case context; and (iii) examples of intervention and mediator research are provided using real-life clinical data.
SEM with small samples in Lavaan: problems and solutions
In this lecture, there will be two parts. In the first part, I will give an overview of issues that arise if small sample sizes are used in SEM: non-convergence, non-admissible solutions, biased estimates, and poor quality of standard errors and fit measures. In the second part, I will discuss several old and recent solutions to handle small samples in SEM in a more effective way.
A Gentle Introduction to Bayesian Statistics
Many of the solutions to deal with small sample sizes rely on Bayesian statistics. However, as of now Bayesian methods are not a part of the statistics curricula in most graduate programs. During this lecture, participants will be gently introduced to Bayesian statistics using many empirical examples. I will clarify the differences between the philosophies and interpretations in classical and Bayesian frameworks, and will illustrate why Bayesian methods might be able to overcome issues with small sample sizes. I will also demonstrate some dangers and commonly encountered difficulties, because using Bayesian methods for small sample sizes is not without challenges.