Dataset Information

Mediation analysis with intermediate confounding: structural equation modeling viewed through the causal inference lens.

ABSTRACT: The study of mediation has a long tradition in the social sciences and a relatively more recent one in epidemiology. The first school is linked to path analysis and structural equation models (SEMs), while the second is related mostly to methods developed within the potential outcomes approach to causal inference. By giving model-free definitions of direct and indirect effects and clear assumptions for their identification, the latter school has formalized notions intuitively developed in the former and has greatly increased the flexibility of the models involved. However, through its predominant focus on nonparametric identification, the causal inference approach to effect decomposition via natural effects is limited to settings that exclude intermediate confounders. Such confounders are naturally dealt with (albeit with the caveats of informality and modeling inflexibility) in the SEM framework. Therefore, it seems pertinent to revisit SEMs with intermediate confounders, armed with the formal definitions and (parametric) identification assumptions from causal inference. Here we investigate: 1) how identification assumptions affect the specification of SEMs, 2) whether the more restrictive SEM assumptions can be relaxed, and 3) whether existing sensitivity analyses can be extended to this setting. Data from the Avon Longitudinal Study of Parents and Children (1990-2005) are used for illustration.

SUBMITTER: De Stavola BL

PROVIDER: S-EPMC4383385 | biostudies-literature | 2015 Jan

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Mediation analysis with intermediate confounding: structural equation modeling viewed through the causal inference lens.

De Stavola Bianca L BL Daniel Rhian M RM Ploubidis George B GB Micali Nadia N

American journal of epidemiology 20141211 1

The study of mediation has a long tradition in the social sciences and a relatively more recent one in epidemiology. The first school is linked to path analysis and structural equation models (SEMs), while the second is related mostly to methods developed within the potential outcomes approach to causal inference. By giving model-free definitions of direct and indirect effects and clear assumptions for their identification, the latter school has formalized notions intuitively developed in the fo ...[more]

PMID: 25504026

Similar Datasets

Project description:ObjectivesThis study used a health belief theory derived framework and structural equation model to examine moderators, mediators, and direct and indirect predictors of childhood vaccination.MethodsA secondary analysis was conducted using data collected from a cross-sectional survey of a random sample of 1599 parents living in urban and rural areas of Mysore district, India. Applying two-stage probability proportionate-to-size sampling, adolescent girls attending 7th through 10th grades in 23 schools were selected to take home a questionnaire to be answered by their parents to primarily assess HPV vaccine intentions. Parents were also asked whether their children had received one dose of Bacillus Calmette-Guérin; three doses of Diphtheria, Pertussis, Tetanus; three doses of oral Polio vaccine; and one dose of Measles vaccine. In addition, parents were asked about their attitudes towards childhood vaccination.ResultsOut of the 1599 parents, 52.2% reported that their children had received all the routine vaccines (fully vaccinated); 42.7% reported their children had missed at least one routine vaccine, and 5.2% reported that their children had missed all routine vaccinations. Perceptions about the benefits/facilitators to childhood vaccination significantly predicted the full vaccination rate (standardized regression coefficient (β) = 0.29) directly and mediated the effect of parental education (β = 0.11) and employment (β = -0.06) on the rate of full vaccination. Parental education was significantly associated indirectly with higher rates of full vaccination (β = 0.11). Parental employment was significantly associated indirectly with decreasing rates of full vaccination (β = -0.05). Area of residence moderated the role of religion (β = 0.24) and the 'number of children' in a family (β = 0.33) on parental perceptions about barriers to childhood vaccination. The model to data fit was acceptable (Root Mean Square Error of Approximation = 0.02, 95% CI 0.018 to 0.023; Comparative Fit Index = 0.92; Tucker-Lewis Index = 0.91).ConclusionsFull vaccination rate was relatively low among children in Mysore, especially among parents who were unsure about the benefits of routine vaccination and those with low educational levels. Interventions increasing awareness of the benefits of childhood vaccination that target rural parents with lower levels of education may help increase the rate of full childhood vaccination in India.

Dataset Information

Mediation analysis with intermediate confounding: structural equation modeling viewed through the causal inference lens.

Publications

Mediation analysis with intermediate confounding: structural equation modeling viewed through the causal inference lens.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets