clique_test: The generalized main randomization test function.

In dpuelz/CliqueRT: Randomization-Based Testing of Causal Effects under General Interference

The generalized main randomization test function.

Description

The generalized main randomization test function.

Usage

clique_test(Y, Z, teststat, biclique_decom, alpha = 0.05, one_sided = T)

Arguments

Y	The observed outcome vector.
Z	A vector that gives the realization of treatment assignment. Its length should match Y and is the number of units in the experiment.
teststat	The test statistic used. See details for further illustration.
biclique_decom	Output from biclique.decompose function that contains a biclique decomposition and controls.
alpha	The significance level. By default it's 0.05.
one_sided	Logical, whether to use a one-sided p-value or a two-sided p-value. Default is TRUE which uses a one-sided p-value. See details for further illustration.

Details

teststat specifies the test statistic used in the conditional clique.

• By default, one_sided is set to be TRUE, which calculates the p-value by the function one_sided_test. It hence requires that a large value of the test statistic provides evidence against the null hypothesis. However, the user can choose other statistics and use the two-sided p-value by setting one_sided=F, which calculates the p-value by two_sided_test.

• It should contain at least (with order) y, z, focal_unit_indicator as inputs, where y is the outcome vector, z is the treatment vector and focal_unit_indicator is a 0-1 vector indicating whether a unit is focal (=1) or not (=0). All three inputs should have length equal to number of units and have the same ordering.

• Users may use other global variables in the function, such as adjacency matrix of the network, but they do not need to be written as parameters of the function.

• We provide several default test statistics for no-interference null that can be generated using function gen_tstat.

Sometimes the test statistic or its randomization distribution contains NA. It may be due to poor selection of focals. In this case the returned p.value is set to be 3 and the user may consider rerun the biclique decomposition.

It might also happen that the randomization distribution is degenerate, which means that the resampled test statistics are all identical, and the p-value will not be available. In this case the returned p.value is set to be 2 and the user may consider changing the test statistic used.

Value

A list of items summarizing the randomization test. It contains the p-value p.value, test statistic statistic, the randomization distribution of the test statistic statistic.dist, and a list MNE of clique decomposition. Each element of MNE records one biclique decomposed from a multi-null exposure graph and contains its focal units and focal assignments.

See Also

gen_tstat, one_sided_test

dpuelz/CliqueRT documentation built on Jan. 6, 2023, 11:20 p.m.