identify_examples: Identify observations that support our results.

In jacobaro/danalyze: Danalyze: Automated Analysis Functions

Identify observations that support our results.

Description

This functions takes a set of results (main and interaction but not mediation) and identifies observations in the dataset used that support the results. Supportive observations are only identified for statistically significant results.

Usage

identify_examples(
  results,
  research.plan,
  unit = NULL,
  time = NULL,
  .time.distance = 0.1,
  .quantile = c(0.8, 0.2),
  .variables = c(".main.variable", ".main.interaction")
)

Arguments

results	The output from 'analyze_plan'.
research.plan	The research plan used in the produce of 'results'.
unit	The unit of analysis to aggregate examples within.
time	The tine variable used in the results. Only necessary for time-series analysis.
.time.distance	If a results was from a survival model, this parameter identifies the quantile of time that indicates a "close" event. If time to the outcome is below this quantile of the time variable, the observation is considered close.
.quantile	The quantile cutoffs for low and high values of a variable. A value above the first cutoff is considered a "high" value while a value below the second cutoff is considered a "low' value.
.variables	Names of the variable columns in the results. Defaults to main and interaction effects. If results contains more than two variables, additional values need to be added here.

Details

For most normal regression models, an observation is supportive when (1) it occurs in the same unit of analysis as a positive outcome and (2) the variable has a high value. For interactions, both the main variable and the interaction need to have the specified values (conditional effects can occur when the interaction variable has a low value–this is accounted for). For event history models, a supportive variable does not need to occur in the same observation as the outcome. Observations that are close in time to the outcome are also considered supportive.

Value

Function returns a dataframe that contains a supportive observation that matches one of the statistically significant results.

See Also

Other post analysis exploration: get_summary_statistics(), get_variable_effects()

jacobaro/danalyze documentation built on Oct. 20, 2022, 8:09 a.m.