pugesek: Bumpus house sparrow data and model.

In stremo: Functions to help the process of learning structural equation modelling

Description

Pugesek and Tomer reanalyse this classic dataset using structural equation modeling.

Usage

data(pugesek)
bumpus
bumpus.log
pugesek.onefactor.model
pugesek.threefactors.model

Format

Two data frames, one untransformed and one log-transformed, with 136 observations on 13 variables. Also, two character vectors with the syntaxes of two of the models described in the paper.

Note

According to the authors, they used log-transformed data in their analyses, so there are two datasets: bumpus and bumpus.log. Not all traits measured were used by the authors in the paper. All analyses used sex as the grouping variable. The one-factor model is in page 392, figure 3 and table 1. The three-factor model is in page 393 and 394, figure 5 and table 2. The survival model in page 395 could not be fitted as there is no support for categorical variables in either lavaan or stremo.

Three-factor models 2 and 3 converge and we get somewhat similar loadings, but there are negative variances, which indicate Heywood cases. This might be caused by the equality constraints we imposed in some of the loadings to try to mimic the results from the paper.

Source

Article:

https://springerlink3.metapress.com/content/m48860525r801047/resource-secured/?target=fulltext.pdf

Data:

http://fm1.fieldmuseum.org/aa/staff_page.cgi?staff=lowther&id=432

References

Pugesek, BH and Tomer, A. 1996. The Bumpus house sparrow data: a reanalysis using structural equation models. Evolutionary Ecology 10: 387-404.

Examples

## Not run: 
data(pugesek)
cat(pugesek.onefactor.model)
cat(pugesek.threefactors.model)

pugesek.onefactor.1 <- sem(pugesek.onefactor.model, data = bumpus.log, fixed.x = FALSE, group =
                           "sex", std.lv = 1, estimator = "MLM")
pugesek.onefactor.2 <- sem(pugesek.onefactor.model, data = bumpus.log, fixed.x = FALSE, group =
                           "sex", std.lv = 1, group.equal = "loadings")
pugesek.onefactor.3 <- sem(pugesek.onefactor.model, data = bumpus.log, fixed.x = FALSE, group =
                           "sex", std.lv = 1, group.equal = c("loadings",
                                                          "residuals"))
pugesek.threefactors.1 <- sem(pugesek.threefactors.model, data = bumpus.log, fixed.x = FALSE, group =
                              "sex", orthogonal = TRUE,
                              std.lv = 1)
pugesek.threefactors.2 <- sem(pugesek.threefactors.model, data = bumpus.log, fixed.x = FALSE, group =
                              "sex", orthogonal = TRUE,
                              std.lv = 1, group.equal = c("loadings"))
pugesek.threefactors.3 <- sem(pugesek.threefactors.model, data = bumpus.log,
                              fixed.x = FALSE, std.lv = TRUE, orthogonal = TRUE)

summary(pugesek.onefactor.1, standardize = TRUE)
summary(pugesek.onefactor.2, standardize = TRUE)
summary(pugesek.onefactor.3, standardize = TRUE)

summary(pugesek.threefactors.1, standardize = TRUE)
summary(pugesek.threefactors.2, standardize = TRUE)
summary(pugesek.threefactors.3, standardize = TRUE)

## End(Not run)

stremo documentation built on Jan. 15, 2017, 7:33 p.m.