Multi-treatment Meta Analysis of Summaries

This menu allows a multi-treatment meta analysis of summaries from experiments to be fitted using the VMETA procedure. This is also known as network meta analysis. This allows for means or other summary statistics from treatments tested in multiple experiments to be combined to give an overall estimate for each treatment with approximate standard errors of differences between treatments. Not all treatments are required to be in all experiments, but all experiments should have some treatments in common with some other experiments. A measure of the precision of each experiment (standard errors of differences or sampling variance) is also required.

From the menu select
Stats | Meta Analysis | Multi-treatment Meta Analysis of Summaries.
Fill in the fields as required then click Run.

A range of models can be fitted, and these can be specified in the Options dialog.

Available data

This lists data structures appropriate to the current input field. It lists either factors for use in specifying the treatments, experiments and moderator, or variates for specifying the means, sampling variances, standard errors of differences or moderator. The contents will change as you move from one field to the next. Double-click a name to copy it to the current input field or type the name into the field.

Means

Specify a variate containing the treatment means or other summary statistic from each experiment.

Treatments

A factor to specify the treatments for the means. Note: not all treatments are required to be in all experiments.

Experiments

A factor to specify the experiments (e.g. trials or environments) that the treatments were evaluated in.

Variance information for experiments

This is used to give the variation within each experiment. This is an experiment-level variate and should either be the same length as the number of experiments, or the same length as the means, with the values constant within each experiment. The variance information can be provided in two forms:

Average standard errors of differences of means	Average standard errors of differences between treatment means from each experiment
Sampling variation from each experiment	Sampling variation, i.e. the residual mean square, from each experiment

Moderator

Specify a variate or factor for a moderator which is incorporated into the REML analysis. This is an experiment-level factor or variate and should either be the same length as the number of experiments, or the same length as the means, with the values constant within each experiment. This can be used to account for experiment-specific effects on the estimated treatment means. This is optional and can be left blank if there is no moderator.

Action Icons

	Pin	Controls whether to keep the dialog open when you click Run. When the pin is down the dialog will remain open, otherwise when the pin is up the dialog will close.
	Restore	Restore names into edit fields and default settings.
	Clear	Clear all fields and list boxes.
	Help	Open the Help topic for this dialog.