help blinding
-------------------------------------------------------------------------------

Title

        blinding -- Estimate blinding index


Syntax

        blinding matname [, options]

    options                Description
    -------------------------------------------------------------------------
      bweight(matname)     weights for the main data, to be used in Bang's
                             index when the data is in the 2x5 form
      ancillary(matname)   ancillary data
      ancweight(matname)   weights for the ancillary data
      level(#)             confidence level


Description

    blinding is an implementation of the blinding indexes introduced by James
    et al (1990) and Bang et al (2004), to assess the success of blinding in
    clinical trials. The command performs the estimation of both indexes,
    along with standard errors, confidence intervals and significance levels,
    based on the null hypothesis that the trial is successfully blinded.

    The usual procedure to measure the effectivness of blindness is to ask
    the participants for the treatment allocation they believe they are
    assigned. The command -blinding- works with the two most common
    structures for the responses, that is, the 2x3 setting and the 2x5
    setting.

    The argument matname is a matrix of dimensions either 2x3 or 2x5, where
    the rows correspond to the true allocation of the participants: the first
    row contains the data from the drug arm and the second row contains the
    data from the placebo arm; the columns correspond to the treatment
    allocation that participants believe they are assigned. In the 2x3
    setting, the columns correspond to drug, placebo and "don't know" (DK)
    respectively, and in the 2x5 setting, the columns correspond to "strongly
    believe that the treatment is drug", "somehow believe that the treatment
    is drug", "somehow believe that the treatment is placebo", "strongly
    believe that the treatment is placebo" and "don't know", in this order.

    The Jame's index is only computed for the 2x3 setting. When the original
    data is entered in the 2x5 form, it is transformed to the 2x3 form in
    order to compute the James' index.

Options

    bweight(matname) specifies the weights to be applied to the main data for
        computing the Bang's index when the data is in the 2x5 format. This
        option is used to assign weights to the "somewhat believe" responses
        and the "strongly believe" responses.  It is a matrix in the same 2x5
        format; the default is weight(1,0.5,0.5,1,0\1,0.5,0.5,1,0).
        According to Bang et al(2004), the weights matrix needs to verify the
        following conditions: the elements of this matrix should be numbers
        between 0 and 1; the first and the forth column should be equal; the
        second and the third column should be equal; the last column should
        only contain zeros.

    ancillary(matname) specifies a 2x3 matrix of ancillary data (when
        available), to be used to compute the Bang's index. In some trials,
        participants who answer DK are asked to choose a treatment allocation
        anyway. The ancillary matrix contains this data. Therefore, the total
        of counts for each row in this matrix should be equal to the
        corresponding DK cell in the main data matrix.

    ancweight(matname) specifies a 2x3 matrix containing the weights for the
        ancillary data.  Requires the ancillary() option. The default is
        ancweight(0.25,0.25,0\0.25,0.25,0).  This matrix needs to verify: the
        entries should be numbers between 0 and 1; the first and the second
        column should be equal; the last column should only contain zeros.

    level(#) specifies the level of the confidence intervals. The default is
        level(95).



Examples

    Setup
         . matrix define A=(2,5,3,4,6\2,4,3,5,8)
         . matrix define B=(1,0.6,0.6,1,0\1,0.6,0.6,1,0)
         . matrix define C=(3,3,0\3,4,1)
         . matrix define D=(0.2,0.2,0\0.2,0.2,0)

    Estimate blinding indexes for A
        . blinding A

    Estimate blinding indexes for A using the weight matrix B
        . blinding A, bweight(B)

    Estimate blinding indexes for A using the ancillary data C
        . blinding A, ancillary(C)

    Estimate blinding indexes for A using the ancillary data C and its weight
    D
        . blinding A, ancillary(C) ancweight(D)

    Estimate blinding indexes for A using the weight matrix B and the
    ancillary data C
        . blinding A, bweight(B) ancillary(C)

    Estimate blinding indexes for A using the weight matrix B, the ancillary
    data C and its weight D, and specify the level of confidence interval to
    90
        . blinding A, bweight(B) ancillary(C) ancweight(D) level(90)



Saved results

    blinding saves the following in r():

    Matrices  
      r(b)           vector containing the reported indexes
      r(var)         vector containing the variances of the indexes


References

    Bang, H., L. Ni, and C. E. Davis. 2004. Assessment of blinding in
    clinical trial. Controlled Clinical Trials 25: 143-156.

    James, K. E., K. K. Lee, H. K. Kraemer, and R. K. Fuller. 1990.  An index
    for assessing blindness in a multicenter clinical trial:  disulfiram for
    alcohol cessation-a VA cooperative study.  Statistics in Medicine 15:
    1421-1434.


Author

    Jiefeng Chen, Texas A&M University, Department of Statistics