Beyond path diagrams: Enhancing applied structural equation modeling research through data visualization

The effect of general practice contact on cancer stage at diagnosis in Aboriginal and non-Aboriginal residents of New South Wales

PURPOSE

Older age, risks from pre-existing health conditions and socio-economic disadvantage are negatively related to the prospects of an early-stage cancer diagnosis. With older Aboriginal Australians having an elevated prevalence of these underlying factors, this study examines the potential for the mitigating effects of more frequent contact with general practitioners (GPs) in ensuring local-stage at diagnosis.

METHODS

We compared the odds of local vs. more advanced stage at diagnosis of solid tumours according to GP contact, using linked registry and administrative data. Results were compared between Aboriginal (n = 4,084) and non-Aboriginal (n = 249,037) people aged 50 + years in New South Wales with a first diagnosis of cancer in 2003-2016.

RESULTS

Younger age, male sex, having less area-based socio-economic disadvantage, and fewer comorbid conditions in the 12 months before diagnosis (0-2 vs. 3 +), were associated with local-stage in fully-adjusted structural models. The odds of local-stage with more frequent GP contact (14 + contacts per annum) also differed by Aboriginal status, with a higher adjusted odds ratio (aOR) of local-stage for frequent GP contact among Aboriginal people (aOR = 1.29; 95% CI 1.11-1.49) but not among non-Aboriginal people (aOR = 0.97; 95% CI 0.95-0.99).

CONCLUSION

Older Aboriginal Australians diagnosed with cancer experience more comorbid conditions and more socioeconomic disadvantage than other Australians, which are negatively related to diagnosis at a local-cancer stage. More frequent GP contact may act to partly offset this among the Aboriginal population of NSW.

Two-stage path analysis with definition variables: An alternative framework to account for measurement error

When estimating path coefficients among psychological constructs measured with error, structural equation modeling (SEM), which simultaneously estimates the measurement and structural parameters, is generally regarded as the gold standard. In practice, however, researchers usually first compute composite scores or factor scores, and use those as observed variables in a path analysis, for purposes of simplifying the model or avoiding model convergence issues. Whereas recent approaches, such as reliability adjustment methods and factor score regression, has been proposed to mitigate the bias induced by ignoring measurement error in composite/factor scores with continuous indicators, those approaches are not yet applicable to models with categorical indicators. In this article, we introduce the two-stage path analysis (2S-PA) with definition variables as a general framework for path modeling to handle categorical indicators, in which estimation of factor scores and path coefficients are separated. It thus allows for different estimation methods in the measurement and the structural path models and easier diagnoses of violations of model assumptions. We conducted three simulation studies, ranging from latent regression to mediation analysis with categorical indicators, and showed that 2S-PA generally produced similar estimates to those using SEM in large samples, but gave better convergence rates, less standard error bias, and better control of Type I error rates in small samples. We illustrate 2S-PA using data from a national data set, and show how researchers can implement it in Mplus and OpenMx. Possible extensions and future directions of 2S-PA are discussed. (PsycInfo Database Record (c) 2022 APA, all rights reserved).

Assessing the Global and Local Uncertainty of Scientific Evidence in the Presence of Model Misspecification

Scientists need to compare the support for models based on observed phenomena. The main goal of the evidential paradigm is to quantify the strength of evidence in the data for a reference model relative to an alternative model. This is done via an evidence function, such as ΔSIC, an estimator of the sample size scaled difference of divergences between the generating mechanism and the competing models. To use evidence, either for decision making or as a guide to the accumulation of knowledge, an understanding of the uncertainty in the evidence is needed. This uncertainty is well characterized by the standard statistical theory of estimation. Unfortunately, the standard theory breaks down if the models are misspecified, as is commonly the case in scientific studies. We develop non-parametric bootstrap methodologies for estimating the sampling distribution of the evidence estimator under model misspecification. This sampling distribution allows us to determine how secure we are in our evidential statement. We characterize this uncertainty in the strength of evidence with two different types of confidence intervals, which we term “global” and “local.” We discuss how evidence uncertainty can be used to improve scientific inference and illustrate this with a reanalysis of the model identification problem in a prominent landscape ecology study using structural equations.

Associations of developmental imbalance between sensation seeking and premeditation in adolescence and heavy episodic drinking in emerging adulthood

BACKGROUND

Dual systems theories suggest that greater imbalance between higher reward sensitivity and lower cognitive control across adolescence conveys risk for behaviors such as heavy episodic drinking (HED). Prior research demonstrated that psychological analogues of these systems, sensation seeking and premeditation, change from childhood through emerging adulthood, and each has been independently linked with HED. However, few studies have assessed whether change over time in these developing analogues is prospectively associated with HED. Moreover, we know of no research that has shown whether within-person differences between higher sensation seeking and relatively lower premeditation across the adolescent period predict HED in emerging adulthood.

METHODS

Prospective data from the National Consortium on Alcohol and NeuroDevelopment in Adolescence study (n = 715) were used to examine the association of sensation seeking and premeditation with HED among adolescents ages 16 to 20 years. We used novel applications of latent difference score modeling and growth curve analysis to test whether increasing sensation seeking, premeditation, and their imbalance over time are associated with HED across the study period, and whether these associations differed by sex.

RESULTS

Whereas premeditation increased linearly from adolescence through emerging adulthood across sexes, males reported growth and females reported decline in sensation seeking. Sensation seeking in adolescence (and not premeditation) was associated with higher levels of HED by emerging adulthood. Importantly, greater imbalance between sensation seeking and premeditation was associated with higher levels of HED by emerging adulthood though we note that variability capturing this imbalance correlated highly (r = 0.86) with baseline levels of sensation seeking.

CONCLUSIONS

Developmental imbalance between higher sensation seeking and lower premeditation in late adolescence may be a risk factor for greater HED in emerging adulthood.

A graph for every analysis: Mapping visuals onto common analyses using flexplot

For decades, statisticians and methodologists have insisted researchers utilize graphical analysis much more heavily. Despite cogent and passionate recommendations, there has been no graphical revolution. Instead, researchers rely heavily on misleading graphics that violate visual processing heuristics. Perhaps the main reason for the persistence of deceptive graphics is software; most software familiar to psychological researchers suffer from poor defaults and limited capabilities. Also, visualization is ancillary to statistical analysis, providing an incentive to not produce graphics at all. In this paper, we argue that every statistical analysis must have an accompanying graphic, and we introduce the point-and-click software Flexplot, available both in JASP and Jamovi. We then present the theoretical framework that guides Flexplot, as well as show how to perform the most common statistical analyses in psychological literature.

Visualisation and network analysis of physical activity and its determinants: Demonstrating opportunities in analysing baseline associations in the Let's Move It trial

Background: Visualisations and readily-accessible web-based supplementary files can improve data reporting and transparency. In this paper, we make use of recent developments in software and psychological network analysis to describe the baseline cohort of a trial testing the Let's Move It intervention, which aimed to increase physical activity (PA) and reduce sedentary behaviours (SB) among vocational school students. Methods: At baseline, 1166 adolescents, distributed across 6 school clusters and four educational tracks, completed measures of PA and SB, theoretical predictors of these behaviours, and body composition. Within a comprehensive website supplement, which includes all code and analyses, data were tabulated and visualised, and network analyses explored relations between predictor variables and outcomes. Results: Average daily moderate-to-vigorous PA was 65 min (CI95: 57min-73 min), and SB 8h44 min (CI95: 8h04min-9h24 min), with 25.8 (CI95: 23.5-28.0) interruptions to sitting. Cluster randomisation appeared to result in balanced distributions for baseline characteristics between intervention and control groups, but differences emerged across the four educational tracks. Self-reported behaviour change technique (BCT) use was low for many but not all techniques. A network analysis revealed direct relationships between PA and behavioural experiments, planning and autonomous motivation, and several BCTs were connected to PA via autonomous motivation. Visualisation uncovered a case of Simpson's paradox. Conclusions: Data-visualisation and data exploration techniques (e.g. network analysis) can help reveal the dynamics involved in complex multi-causal systems - a challenging task with traditional data presentations. The benefits of presenting complex data visually should encourage researchers to publish extensive analyses and descriptions as website supplements, which would increase the speed and quality of scientific communication, as well as help to address the crisis of reduced confidence in research findings. We hope that this example will serve as a template for other investigators to improve upon in the future.