plot.bootnet {bootnet}

Plots bootnet results

Description

This function can be used to plot bootnet results by plotting all bootstrapped statistics as line or by plotting confidence intervals.

Usage

## S3 method for class 'bootnet'
plot(x, statistics, plot, graph, CIstyle = c(
                 "quantiles", "SE"), rank = FALSE, sampleColor =
                 "darkred", samplelwd = 1, meanColor = "black", meanlwd
                 = 0.5, bootColor = "black", bootAlpha = 0.01, bootlwd
                 = 0.9, areaAlpha = 0.1, order = c("id", "sample",
                 "mean"), decreasing = TRUE, perNode = FALSE,
                 legendNcol = 2, labels = TRUE, legend = TRUE,
                 subsetRange = c(100, 0), area = !perNode, alpha =
                 0.05, onlyNonZero = FALSE, differenceShowValue,
                 differenceEdgeColor = TRUE, verbose = TRUE, panels =
                 TRUE, split0 = FALSE, prop0 = ifelse(split0, TRUE,
                 FALSE), prop0_cex = 1, prop0_alpha = 0.8,
                 prop0_minAlpha = 0.25, subset, ...)

Arguments

x

A bootnet object

statistics

The types of statistics to plot.Defaults to "edge" for regular bootstrap and c("strength", "outStrength", "inStrength") for node and person drop bootstrap. Use "all" to obtain all pairwise statistics tested for regular bootstraps and all nodewise statistics tested for person and node drop bootstarps.

plot

Character string indicating what to plot. Can be "area" to produce a graph with the area indicating the confidence region, or "difference" producing a plot showing significant differences. Other options are "line" and "interval", which are currently unstable and not recommended to use.

graph

If multipe graphs are estimated, which graph should be plotted? Currently used for default = "graphicalVAR" to plot a temporal network using graph = "temporal" or a contemporaneous network using graph = "contemporaneous"

CIstyle

Style of CIs to construct. "SE" shows the sample statistic plus and minus two times the standard deviation of bootstraps, and "quantiles" the area between the 2.5th and 97.5th quantile. Defaults to "quantiles".

rank

Logical, should plots show rank of statistics instead of statistics?

sampleColor

Color of the original sample line

samplelwd

Line width of the original sample line

bootColor

Color of the bootstrap lines

bootAlpha

Alpha of the bootstrap lines

bootlwd

Line width of the bootstrap lines

areaAlpha

Alpha of the area

order

String indicating how to order nodes. "id" will order nodes based on their name, "mean" will order nodes based on the average bootstrapped value of the first statistic in statistics, and "sample" will order the nodes as done in "mean" but orders ties based on their sample value.

decreasing

Logical indicating if the ordering is decreasing or increasing.

perNode

Logical, should centrality estimates per node be plotted instead of correlation with original parameter. Only used in node and person drop bootstrap.

legendNcol

Number of columns in the legend if perNode = TRUE.

labels

Logical, should labels be plotted?

legend

Logical, should the legend be plotted?

subsetRange

Range in percentages of the x-axis in node and person drop plots.

area

Logical, should the confidence area be plotted?

alpha

Signifiance level used in plot = "difference".

onlyNonZero

Logical used when plot = "difference", statistics = "edge", should only edges be included that were nonzero in the estimated network structure?

differenceShowValue

Logical used when plot = "difference". Should values be shown in the diagonal of the difference plot?

differenceEdgeColor

Logical used when plot = "difference", statistics = "edge". Should diagonal blocks be colored according to default edge colors?

verbose

Should expected alpha be printed?

panels

Logical, should panel titles be printed?

meanColor

Color of the bootstrap means.

meanlwd

Line width of the bootstrap means

split0

Logical. When set to TRUE, the displayed intervals are based on occasions when the parameter was not estimated to be zero, and an extra box is added indicating the number of times a parameter is estimated to be zero.

prop0

Logical, should boxes indicating the proportion of times parameters were estimated to be zero be added to the plot?

prop0_cex

Size of the boxes indicating number of times a parameter was set to zero.

prop0_alpha

Transparency of the boxes indicating number of times a parameter was set to zero.

prop0_minAlpha

Minimal transparency of the *lines* of plotted intervals as the proportion of times an edge was not included goes to 0.

subset

Vector indicating labels of nodes to include in the plot. This can be used to show, for example, only edges related to one particular node.

...

Not used.

Value

A ggplot2 object.

Author(s)

Sacha Epskamp <[e-mail]>