Agreement (version 0.8-1)

unified.agreement: A Unified Approach for Assessment of Agreement

Description

The unified approach calculates the agreement statistics for both continuous and categorical data to cover multiple readings from each of the n subjects.

Usage

unified.agreement(dataset, var = NA, k, m, CCC_a = 0.9775, CCC_a_intra = 0.995, CCC_a_inter = 0.9775, CCC_a_total = 0.9775, CP_a = 0.9, TDI_a = 150, TDI_a_intra = 75, TDI_a_inter = 150, TDI_a_total = 150, tran = 0, error = "prop", dec = 4, alpha = 0.025, iter = 35, toe = 1e-10)

Arguments

dataset

Name of your dataset.

var

Name of the dependent variables used in the dataset, if it is not specified, all the variables in the dataset will be used.

k

Number of methods/raters/instruments/assay, etc.

m

Number of replications for each methods/raters/instruments/assay, etc.

CCC_a

CCC allowance when there is no replications, the default is CCC_a=0.9775.

CCC_a_intra

Intra CCC allowance.

CCC_a_inter

Inter CCC allowance.

CCC_a_total

Total CCC allowance.

CP_a

CP allowance for continuous data.

TDI_a

TDI allowance when there is no replications for continuous data. The default is TDI_a=150.

TDI_a_intra

Intra TDI allowance for continuous data.

TDI_a_inter

Inter TDI allowance for continuous data.

TDI_a_total

Total TDI allowance for continuous data.

tran

If transformation is used for the dataset. tran=0: no transformation will be used for statistical inference. tran=1 transformation such as Z, logit and log will be used for statistical inference. tran=1 can be used for categorical data but the TDI and CP outputs would become irrelevant. Therefore, tran=0 is recommended for all categorical data and tran=1 is recommended for all continuous data.

error

error="const": constant error structure for continuous data. When error="const", TDI is an absolute difference with the same measurement unit as for the original data. error="prop": proportional error structure for continuous data. When error="prop", TDI is the percent change. Log transformation will be applied to original data. For categorical data, use error="const".

dec

Significant digits after the decimal point printed for TDI when the error="const" is specified. The default is dec=2.

alpha

100(1-alpha)

iter

Number of iterations in the GEE step. The default is iter=35.

toe

The error tolerance of the GEE iteration. The default is toe = 1e-10.

Value

An object of class unified_agreement is returned, which contains all the estimated statistics and their confidence limits.

See Also

agreement, summary.unified_agreement, html.unified_agreement

Examples

data(DCLHb);
ua <- unified.agreement(dataset=DCLHb, var=c("HEMOCUE1","HEMOCUE2","SIGMA1","SIGMA2"), k=2, m=2, CCC_a_intra=0.9943, CCC_a_inter=0.9775, CCC_a_total=0.9775, CP_a=0.9, tran=1, TDI_a_intra=75, TDI_a_inter=150, TDI_a_total=150, error="const", dec=1, alpha=0.05);
summary(ua);