2009
DOI: 10.1890/08-1034.1
|View full text |Cite
|
Sign up to set email alerts
|

Confirmatory path analysis in a generalized multilevel context

Abstract: Abstract. This paper describes how to test, and potentially falsify, a multivariate causal hypothesis involving only observed variables (i.e., a path analysis) when the data have a hierarchical or multilevel structure, when different variables are potentially defined at different levels of such a hierarchy, and when different variables have different sampling distributions. The test is a generalization of Shipley's d-sep test and can be conducted using standard statistical programs capable of fitting generaliz… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
2
2
1

Citation Types

3
960
0
1

Year Published

2015
2015
2024
2024

Publication Types

Select...
8

Relationship

0
8

Authors

Journals

citations
Cited by 848 publications
(964 citation statements)
references
References 9 publications
3
960
0
1
Order By: Relevance
“…This multivariate technique is useful for testing a priori defined models and quantifying the relative importance of explanatory variables. SEM can also test whether a given effect is direct (e.g., IAP cover influences native plant diversity) or indirect (e.g., flow regime influences IAP cover, which influences native plant diversity) (Shipley 2009;Lefcheck 2016). We developed a conceptual model (meta-model) detailing potential cause-effect relationships based on biological relevance in the literature or logical arguments to guide the modelling process (Fig.…”
Section: Structural Equation Modellingmentioning
confidence: 99%
See 1 more Smart Citation
“…This multivariate technique is useful for testing a priori defined models and quantifying the relative importance of explanatory variables. SEM can also test whether a given effect is direct (e.g., IAP cover influences native plant diversity) or indirect (e.g., flow regime influences IAP cover, which influences native plant diversity) (Shipley 2009;Lefcheck 2016). We developed a conceptual model (meta-model) detailing potential cause-effect relationships based on biological relevance in the literature or logical arguments to guide the modelling process (Fig.…”
Section: Structural Equation Modellingmentioning
confidence: 99%
“…Upon model validation a significant missing path was identified, the effect of mean annual flood frequency on the diversity of propagules, and added to the SEM. Fisher's C [Shipley's test of directed separation; (Shipley 2009)] was used to evaluate SEM fit, where higher P values ([ 0.05) indicate that the data supports the model (i.e., H 0 = no difference between the data and the hypothesised paths). However, it should be noted that alternative models could also support the data, and for one response variable (BCI spring v summer Yr2), two alternative SEMs were constructed based on a priori hypotheses, and the best model in this case was selected using AICc.…”
Section: Structural Equation Modellingmentioning
confidence: 99%
“…We tested the hypothesis that abiotic conditions affected Trientalis performance via their relationships with biotic conditions using structural equation models (Grace, 2006) and mediation tests (d-separation tests, Shipley, 2009;Clough, 2012). With the structural equation models we were able to test for potential direct and indirect associations between broad categories of abiotic conditions (soil and topographic microclimate) and plant performance, while with the mediation tests we were able to select the combination of biotic conditions that explained a significant proportion of the effect of abiotic variables on plant performance.…”
Section: Discussionmentioning
confidence: 99%
“…'mediating' effects (defined in Grace et al, 2012), were examined with mediation tests (d-separation tests, Shipley, 2009;Clough, 2012). We first considered which abiotic variable (serpentine soils or elevation) was significantly related to each biotic condition (overstorey shading, litter, understorey cover, mycorrhizal or septate fungal colonization) and whether these biotic conditions were significantly associated with plant variables (presence, abundance or reproduction).…”
Section: Discussionmentioning
confidence: 99%
“…From full models of litter size and juvenile survival, we selected a baseline model for these two reproductive traits by retaining only the confounding variables with statistically significant effects on a given reproductive trait (see electronic supplementary material, section B). To test the fit of the data to the indirect scenario (scenario 2), we compared the C value associated with scenario 2 to a x 2 -distribution with 2k degrees of freedom (where k is the number of pairs of variables in the graph that are not associated in the focal model) [27]. We rejected scenario 2 when the p-value was below 0.05 [27].…”
Section: (E) Sample Sizesmentioning
confidence: 99%