summary.metaConvert function

Synthesize information of an object of class metaConvert into a dataframe

## S3 method for class 'metaConvert'
summary(object, digits = 3, ...)

Arguments

• object: an object of class metaConvert
• digits: an integer value specifying the number of decimal places for the rounding of numeric values. Default is 3.
• ...: other arguments that can be passed to the function

Returns

This function returns a dataframe with many columns. We present below the information stored in each column of the returned dataframe

1. Raw user information.

The first columns placed at the left of the returned dataset are simply information provided by the users to facilitate the identification of each row. If the following columns are missing in the original dataset, these columns will not appear in the returned dataset.

row_id	Row number in the original dataset.
study_id	Identifier of the study.
author	Name of the author of the study.
year	Year of publication of the study.
predictor	Name of the predictor (intervention, risk factor, etc.).
outcome	Name of the outcome.
info_expected	Types of input data users expect to be used to estimate their effect size measure.

2. Information on generated effect sizes.

Then, the function returns information on calculations. For example, users can retrieve the effect size measure estimated, the number and type(s) of input data allowing to estimate the chosen effect size measure, and the method used to obtain a unique effect size if overlapping input data were available. These columns could have several suffix.

• If users requested to separate crude and adjusted estimates, then the following columns will be presented with both a "_crude" suffix and a "_adjusted" suffix.
• If users did not request to separate the presentation of crude and adjusted estimates, the following columns will have no suffix.

For example, let's take column "all_info". It can be "all_info_crude" (all input data used to estimate any crude effect size), "all_info_adjusted" (all input data leading to estimate any adjusted effect size), or "all_info" (all input data leading to estimate any crude or adjusted effect sizes).

To facilitate the presentation, we thus refer to these columns as name_of_the_column*, the * meaning that it could end by _crude, _adjusted or "".

all_info*	list of input data available in the dataset that was used to estimate any effect size measure.
measure*	effect size measure requested by the user.
info_measure*	input data available to estimate the requested effect size measure.
n_estimations*	number of input data available to estimate the requested effect size measure.
es_selected*	method chosen by users to estimate the main effect size when overlapping data are present.
info_used*	type of input data used to estimate the main effect size.

3. Main effect size.

The following columns contain the key information, namely, the main effect size + standard error + 95% CI.

Again, the suffix of these columns can vary depending on the separation of effect sizes estimated from crude and adjusted input data.

es*	main effect size value.
se*	standard error of the effect size.
es_ci_lo*	lower bound of the 95% CI around the effect size.
es_ci_up*	upper bound of the 95% CI around the effect size.

4. Overlapping effect sizes

These columns are useful ONLY if a given comparison (i.e., row) has multiple input data enabling to compute the requested effect size measure.

These columns identify the smallest/largest effect size per comparison, and some indicators of consistency.

Again, the suffix of these columns can vary depending on the separation of effect sizes estimated from crude and adjusted input data.

min_info*	type of input data leading to the smallest effect size for the comparison.
min_es_value*	smallest effect size value for the comparison.
min_es_se*	standard error of the smallest effect size for the comparison.
min_es_ci_lo*	lower bound of the 95% CI of the smallest effect size for the comparison.
min_es_ci_up*	upper bound of the 95% CI of the smallest effect size for the comparison.

max_info*	type of input data leading to the largest effect size for the comparison.
max_es_value*	largest effect size value for the comparison.
max_es_se*	standard error of the largest effect size for the comparison.
max_es_ci_lo*	lower bound of the 95% CI of the largest effect size for the comparison.
max_es_ci_up*	upper bound of the 95% CI of the largest effect size for the comparison.

diff_min_max*	difference between the smallest and largest effect size for the comparison.
overlap_min_max*	% of overlap between the 95% CIs of the largest/smallest effect sizes for the comparison.
dispersion_es*	standard deviation of all effect sizes for the comparison.

Details

Summary method for objects of class metaConvert produced by the convert_df

function. This function automatically:

1. computes all effect sizes from all available input data
2. selects, if requested, a main effect size for each association/comparison using the information passed by the user in the es_selected argument of the convert_df function
3. identifies the smallest and largest effect size for each association/comparison
4. estimates the absolute difference between the smallest and largest effect size for each association/comparison
5. estimates the percentage of overlap between the 95% confidence intervals of the smallest and largest effect size for each association/comparison

Examples

### generate a summary of the results of an umbrella object
summary(
  convert_df(df.haza, measure = "g"),
  digits = 5)

See Also

metaConvert-package for the formatting of well-formatted datasets

convert_df for estimating effect sizes from a dataset