Putting affective dynamics into context: Towards a realistic and reliable modeling approach

• Applicant: Dr. Janne Adolf
• Subject Area: Personality Psychology, Clinical and Medical Psychology, Methodology ; General, Cognitive and Mathematical Psychology
• Term: from 2019 to 2021
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 424706845

The aim of recent affect research is to understand affective phenomena as continuously changing within-person processes. By unraveling the dynamics that underlie these processes, one hopes to gain deeper insights into affective functioning and affect regulation. To this end, the field relies on intense longitudinal data, and dynamic, especially vector autoregressive (VAR), time series modeling. The new research paradigm has been embraced by many due to its supposed potential – but it also bears non-trivial challenges. Two of them, developing realistic (i.e., sufficiently complex) modeling approaches to affective functioning, and ensuring adequate reliability of the obtained modeling solutions given typically limited data, form an intriguing trade-off. In the here proposed research project, I aim to address both challenges, and tackle the resulting balancing act. This yields the following twofold research objective.First, for a realistic modeling approach, I draw upon notions of regulatory flexibility, suggesting that affective dynamics depend on and thus change with context. To empirically study this, VAR models are required to include contextual variables that moderate their parameters, and the presumably associated affective dynamics, over time. Fixed moderated time series analysis (fmTSA; Adolf et al., 2017) implements such an extended VAR model by incorporating observed contextual variables as moderators. Here, I propose to further develop this method along two important lines. One concerns the incorporation of moderators of different formats to enable the identification of different patterns or forms of context-dependent affective functioning. The other concerns the rigorous handling of measurement intervals of unequal length, which could otherwise lead to biased modeling solutions.Second, I propose to investigate the reliability of fmTSA under typical longitudinal data conditions. Reliability is thereby evaluated in terms of the precision of modeling solutions (i.e., estimated model parameters), and in terms of the precision of predictions derived from parameter estimates. I thus cover the explanatory as well as the predictive performance of the method. Besides establishing and implementing routines for retrospective reliability diagnosis, I also aim at prospective reliability assessment and improvement. That is, for expected parameter values, I will a-priori determine how reliability varies as a function of data conditions. The knowledge about these associations can then be used to optimize the design of longitudinal studies, and thus produce data that guarantee reliable fmTSA solutions.Taken together, the here proposed research project should culminate in a highly usable and credible modeling approach to putting affective dynamics into context.

• DFG Programme: Research Fellowships
• International Connection: Belgium
• Host: Professorin Dr. Eva Ceulemans