Description, prediction and causation: Methodological challenges of studying child and adolescent development

Abstract

Scientific research can be categorized into: a) descriptive research, with the main goal to summarize characteristics of a group (or person); b) predictive research, with the main goal to forecast future outcomes that can be used for screening, selection, or monitoring; and c) explanatory research, with the main goal to understand the underlying causal mechanism, which can then be used to develop interventions. Since each goal requires different research methods in terms of design, operationalization, model building and evaluation, it should form an important basis for decisions on how to set up and execute a study. To determine the extent to which developmental research is motivated by each goal and how this aligns with the research designs that are used, we evaluated 100 publications from the Consortium on Individual Development (CID). This analysis shows that the match between research goal and research design is not always optimal. We discuss alternative techniques, which are not yet part of the developmental scientist's standard toolbox, but that may help bridge some of the lurking gaps that developmental scientists encounter between their research design and their research goal. These include unsupervised and supervised machine learning, directed acyclical graphs, Mendelian randomization, and target trials.

Keywords: Causality; Developmental science; Directed acyclical graphs; Mendelian randomization; Radical randomization; Target trials.

Fig. 1

Number of papers in a random sample of 100 CID studies that indicate a descriptive/association, predictive/forecasting, or explanation/causal interest in the research question, hypotheses, discussion, and conclusion. Note that a study may include multiple goals.

Fig. 2

Number of papers in the random sample of 100 CID studies based on a particular design and that indicate a descriptive/association, predictive/forecasting, or explanation/causal interest in the research question. Papers may include multiple goals and multiple study designs. Experimental research focused on causation is further divided into “M” when the assumed cause X was actively manipulated by the researchers, and “O” when the assumed cause X was observed (e.g., when X is an outcome variable in the experiment, as in the mediation paradox, or a variable measured prior to the experiment; see main text for further details).

Fig. 3

Three different directed acyclical graphs (DAGs) that show the three different roles of a third variable Z, that is: a confounder or common cause, a mediator on an indirect causal path, and a collider or effect of both variables.

Fig. 4

Three different directed acyclical graphs (DAGs) that show the three different roles of a third variable Z, that is: a confounder or common cause, a mediator on an indirect causal path, and a collider or effect of both variables.

Fig. 5

Three directed acyclical graphs (DAGs). Left DAG shows an observational study in which the causal effect of X on Y is biased due to unobserved confounding U. Middle DAG shows how randomizing treatment (through a flip of a coin), breaks the causal effect of U on X, thereby allowing for the estimation of the causal effect of X on Y. Right DAG shows the use of an instrumental variable Z in a non-experimental study: By dividing the effect of Z on Y (a*b) by the effect of Z on X (a), one can estimate the unbiased effect of X on Y (b).

Fig. 6

DAG for intergenerational Mendelian experiment for investigating the causal in uterus effect of maternal BMI on future offspring BMI, with family eating habits as unobserved confounders, maternal genotype as the instrumental variable, and offspring genotype as mediator that needs to be controlled for. Based on Figure 1 in Richmond et al. (2017).