New article: Testing subgroup differences in large psychometric models

Equal parameter estimates across subgroups is a substantial requirement of statistical tests. Ignoring subgroup differences poses a threat to study replicability, model specification, and theory development. Parameter invariance implies that an identical statistical model holds for different subgroups (e.g., males and females, older and younger persons, and persons with different ethnic backgrounds) or across measurement occasions. Violations of this assumption can lead to misspecified models, spurious parameter estimates and test results, therefore, concealing differences key for theory development, diagnostic procedures, and treatment design.    

Structural change tests are a powerful statistical technique to assess parameter invariance. They have been adapted to simple linear regression models, but more recently to complex statistical techniques like item response modeling, structural equation modeling, and network analysis. A core element of those tests is the empirical fluctuation process. In the case of parameter invariance, the fluctuation process asymptotically follows a Brownian bridge. This asymptotic assumption further provides the basis for determining the sampling distribution and conducting inference. However, in our paper we show through simulation and mathematical derivation that the empirical fluctuation process does not follow a Brownian bridge in small samples. This situation is amplified in large psychometric models. We simulate simple linear regression models and complex Gaussian graphical models under various sample sizes and number of parameters included. Results show a sub-optimal power and a type 1 error rate below the expected nominal significance-threshold. Therefore, the asymptotic sampling distribution does not exploit the significance level and is of poor quality compared to the exact conditional and conditional asymptotic distribution. In short, the SCT becomes increasingly conservative in small samples.    

We discuss an alternative solution to obtaining the sampling distribution—permutation approaches. Permutation approaches estimate the sampling distribution through resampling of the dataset, avoiding distributional assumptions. We show that permutation approaches provide a correct type I error and tend to increase the test's statistical power in finite samples; the test's performance is improved. We conclude that the permutation alternative is superior to standard asymptotic approximations and allow for the test to be used also in large psychometric models.    

Huth, K.B.S., Waldorp, L.J., Luigjes, J. et al. A Note on the Structural Change Test in Highly Parameterized Psychometric Models. Psychometrika (2022).  

Read the article here.

Centre for Urban Mental Health returns a 2nd year to teach honours course for talented students 

'When one teaches, two learn.' – Robert Heinlein

We are excited to start our second year teaching the Honours Module on Urban Mental Health at the University of Amsterdam

Via the Problem Based Learning approach, we'll guide an interdisciplinary group of 25 honours students through the current state of the art on Urban Mental Health and foster collaboration on urban challenges over the next 9 weeks. 

Find out more here.

A new Frontiers paper on the interplay between resilience and quality of life in later life

Age-related challenges and transitions can have considerable social, psychological, and physical consequences. Maintaining a high quality of life in later life is therefore not self-evident.  Nevertheless, the chance of success seems greater if one has the skills to remain in control over his/her own life and environment. This was concluded from an ongoing aging project led by researchers at the Centre for Urban Mental Health. The project, Senioren Doen Mee, aims to unravel the secret to a vital and happy old age. Particularly, the researchers are interested as to why some older individuals grow old greatly satisfied, whereas others are at risk of succumbing.

According to the authors, the key concept to this matter is resilience. Resilience is defined as the process of successfully adapting to difficult, stressful, and challenging life experiences. It is thought to promote the maintenance of functional ability, well-being, and QoL at old age through the act of all kinds of ‘resilience factors’, including those from the psychological, social and physical domain. While a lot of research has been done on establishing what factors contribute to successful aging and resilience, the relative contribution of each of these factors is largely unknown. However, this is crucial to identify the most opportune target points for interventions that can contribute to successful aging.  

The Senioren Doen Mee project has been set up to explore this interplay. As a primary endeavor in their search for an explanation of individual differences in resilience, the researchers delved into the data of the first 1392 participants (all 55 years or older). They focused on a subset of resilience factors: (1) how people deal with (negative) situations (coping), to what extent they are able to keep a grip on their lives (self-management ability), whether they manage to look at their life in a positive way (positive appraisal style) and the role physical activity plays in all of this. 

Their interplay with (various facets of) QoL was established through network analysis. This is an analysis technique that allows one to estimate and visualize the relationships among a set of variables by controlling for all other variables. The researcher found that QoL was most strongly (and directly) related to positive appraisal style and self-management ability. Among those, the self-management ability subscale ‘taking care of multifunctional resources’(i.e., yielding various benefits at the same time) seemed to be crucial. It was connected directly to ‘satisfaction with past, present, and future activities’, which is a key facet of QoL with strong interconnections to other QoL facets. 

This suggests that the ability to take care of multifunctional resources may have a high intervening potential to promote QoL. According to the researcher, helping older adults to implement daily life adjustments that lead to socially rewarding and productive interactions with various benefits, might be an excellent starting point. Ultimately, the appropriate set of resilience factors to manipulate may depend on the goal and/or facet of QoL that one aims to improve. These first findings will aid future studies (within the Senioren Doen Mee project) in determining specific strategies that can help older adults to gain control of their own lives, enabling them to maintain the functional ability and competence that is vital for well-being and QoL at old age. 

Read the article here.

PNAS Commentary "No robust relation between larger cities and depression"

''Larger cities provide a buffer against depression'' - this statement is from a recent PNAS article by Stier et al. (2021) on how depression rates scale with the population of Metropolitan Statistical Areas (MSAs). However, this statement runs contrary to a wealth of psychological and epidemiological research showing the complex nature of depression and the detrimental influence of cities. 

UMH Researchers have assessed the quality of the data through a short commentary, which was recently published alongside the original article in PNAS. They focused on assessing the stability of the findings with varying city definitions, in particular the spatial extent of cities. Their re-analysis suggests that Stier and colleagues` conclusion is unstable; it holds for cities larger than 78km in radius. 

For a narrower city definition, the original article's finding disappears and even suggests that larger cities pose a risk for depression. Taken together, the analysis by UMH researchers shows that cities can have both a protective and a detrimental effect on depression depending on how they are defined. 

The researchers conclude with three key take-aways: (1) the definition of cities (minimum inhabitants size and spatial area) are heavily influencing findings; (2) cities influence on mental health is not straightforward, both between cities (i.e., between two cities) and within cities (i.e., city core versus suburbs) dynamics exist; (3) always be sceptical of flashy findings.   

Read the commentary here.

New article: Healthy and clinical symptom configuration of alcohol use

Recreational alcohol use is widespread in our society: a Friday-afternoon beer, a celebratory champagne, or an evening a few nights back with a few drinks too much. While alcohol consumption can be associated with short-term benefits -- a nice complement to food, positive mood, loosened social interactions -- its enhanced, uncontrolled use is associated with negative outcomes, termed alcohol use disorder (AUD). Affected individuals struggle to control their alcohol consumption despite detrimental effects on their physical and/or mental health and that of their close social circle. This drove UMH researchers to try to understand what drives individuals from recreational use to alcohol dependence.


Difficulty in studying and treating addictive behaviors lies in its complex nature; like other mental health disorders, AUD is influenced by a multitude of factors on different social, psychological and biological levels. Common approaches to assessing clinical disorders are unable to capture this multitude of interconnected, highly dependent symptoms. Therefore, research in the last decade suggests modeling psychological disorders as a set of interacting entities. This direction was introduced in the field of clinical psychology as the network theory of mental disorders. The network theory models mental disorders as entities (e.g., symptoms) that interact and causally influence each other. In this project, researchers applied the network theory to assess how symptom interactions differ between healthy individuals (i.e., population sample with mainly recreational drinkers) and clinical individuals (i.e., individuals with an alcohol use disorder).  In the healthy sample, the symptoms were strongly connected, with particularly strong connections between emotional problems to work problems and between work problems to dangerous activities. The network of the clinical sample is much sparser connected; the strongest interactions were found between loss of control to dangerous activities. Furthermore, researchers assessed which symptoms are central in each sample's network. While for the clinical sample a central symptom could be considered a treatment target, in the healthy sample, it could be a target of preventive action. In the clinical sample, the loss of control was most strongly connected, corroborating known theories of AUD stating loss of control as the main driver of addiction. This symptom was least strongly connected within the healthy sample; here, time spent drinking alcohol was most strongly connected. 


Symptoms can trigger other symptoms, but it is not understood why this results in substance use in one individual whereas no dependence in another individual. Therefore, the researchers further assessed the effect of socio-demographic characteristics (e.g., gender, age, ethnicity, and income) on the symptom interplay. Results indicated that for recreational drinkers the symptom interactions are influenced by socio-demographics, especially for age subgroups. Thus, subgroup-specific networks should be considered when deriving implications for theory building or prevention planning.

Read the commentary here.

Systematic review published on social instability stress in animal models

 Many aspects of our modern society, such as a high population density (e.g. in an urban environment), increase our frequency of social interactions. While a high number of social interactions can be beneficial and attractive for some individuals, it may provoke stress for others. Furthermore, prolonged social stress represents an important risk factor for the development of psychiatric disorders, including depression and anxiety disorders.

The social instability stress (SIS) paradigm induces social stress in animal models and is used to promote our understanding of the mechanisms underlying social stress-induced depression and/or anxiety on a molecular and behavioral level. The SIS paradigm entails the frequent introduction of unfamiliar animals into the home cage. The subsequent recurring formation of a new social hierarchy results in chronic and unpredictable social stress. 

In this systematic review, published in Neurobiology of Stress, a team of researchers, including Amber Koert, Claudi Bockting, Paul Lucassen, Anouk Schrantee, and Joram Mul from the Centre for Urban Mental Health, investigated the effect of two commonly applied SIS protocols on several stress-related outcomes in rats and mice. Both the shorter and longer SIS models induced a wide range of stress-related behavioral and physiological effects that are relevant for the pathophysiology of depression and anxiety disorders. However, some of the effects appear to be species-dependent and/or duration-dependent. To increase the standardization and reproducibility of this etiologically relevant preclinical model of social stress, we provide practical suggestions for optimal design and reporting of SIS protocols.

Read the review here.

Article on psychopathological networks: theory, methods and practice

In recent years, network approaches to psychopathology have sparked much debate and have had a significant impact on how mental disorders are perceived in the field of clinical psychology. However, there are many important challenges in moving from theory to empirical research and clinical practice and vice versa. Therefore, in this newly published article, a group of researchers, including Professor Claudi Bockting and Denny Borsboom at the Centre for Urban Mental Health, bring together different points of view on psychological networks by methodologists and clinicians to give a critical overview on these challenges, and to present an agenda for addressing these challenges.

In contrast to previous reviews, they especially focus on methodological issues related to temporal networks. This includes topics such as selecting and assessing the quality of the nodes in the network, distinguishing between- and within-person effects in networks, relating items that are measured at different time scales, and dealing with changes in network structures. These issues are not only important for researchers using network models on empirical data, but also for clinicians, who are increasingly likely to encounter (person-specific) networks in the consulting room.

Read more here.

UMH PhD Retreat to Groet, NL 2021

We had a wonderful time in November 2021 working on UMH research with the PhD students. 

UMH PhD Candidates & Co-director Claudi Bockting 

Several UMH PhD Candidates

During the UMH PhD Retreat 2021 in Groet

Recap of the 'Cities & Health' Urban Studies Networking Event

On the 1st of June the Centre for Urban Studies (CUS) co-hosted the fifth edition of the Urban Studies Networking Event. This event, organized by the CUS, was co-hosted with the Centre for Urban Mental Health (UMH) and the Centre for Social Sciences and Global Health (SSGH). The event aimed to explore shared interests and to stimulate future collaborations between scholars looking at health aspects in urban areas from different methodological, epistemological and ontological perspectives.

Read the recap here.

Centre for Urban Mental Health featured in Lancet Psychiatry profile

The Centre for Urban Mental Health is featured in a recent Lancet Psychiatry profile of our co-director prof. Claudi Bockting, where she discusses the potential of complexity science and urban mental health research.

For more information, please read the profile here

UMH funded by Wellcome Trust

We’ve been funded by Wellcome Trust to advance science so that no one is held back by mental health problems

8 June 2020

We are excited to announce that the Centre for Urban Mental Health and the Amsterdam University Medical Centre at the University of Amsterdam are one of a select number of teams funded by Wellcome Trust to review the evidence on which aspects of interventions really make a difference in preventing and treating youth anxiety and depression. Each commissioned team will review the evidence for one active ingredient.

The active ingredients are diverse and cover the range of options, from improving gut microbiome function to increasing financial resources via cash transfer, from the use of antidepressants to increased self-compassion.

Our team will look at neighbourhood cohesion, the extent of connectedness within communities in which young people grow up, especially in an urban environment. During the commission, we will be reviewing the available evidence to present an insight analysis into what the evidence says about how effective increasing neighbourhood cohesion is for preventing anxiety and depression.

Read the press release in full here.

Read more about the rewarded projects here.

Effective digital mental health more necessary than ever during COVID-19 pandemic

20 September 2020

A study published today by the research group of Professor Claudi Bockting (Centre for Urban Mental Health, University of Amsterdam) shows that digital psychological interventions can effectively diminish the symptoms of mental disorders including depression and substance use, even in low and middle income countries.

Read the press release in full here.

Read the article here.