kappa.hlp

是一个经济学管理应用软件 很难找的 但是经济学学生又必须用到

{smcl}
{* 07mar2005}{...}
{cmd:help kap}, {cmd:help kapwgt}, {cmd:help kappa}{...}
{right:dialogs:  {dialog kap_uniq:kap (2 raters)}  {bf:{dialog kapwgt}}}
{right:{dialog kap_nonuniq:kap (2+ raters)}  {bf:{dialog kappa}}}
{hline}

{title:Title}

{p2colset 5 18 20 2}{...}
{p2col :{hi:[R] kappa} {hline 2}}Interrater agreement{p_end}
{p2colreset}{...}


{title:Syntax}

{phang}
Interrater agreement, two unique raters

{p 8 12 2}
{cmd:kap} {it:{help varname:varname1}} {it:varname2} {ifin}
{weight} [{cmd:,} {it:{help kappa##options:options}}]

{phang}
Weights for weighting disagreements

{p 8 15 2}
{cmd:kapwgt} {it:wgtid} [{cmd:1} {cmd:\} {it:#} {cmd:1} [{cmd:\} {it:#}
{it:#} {cmd:1} {it:...}]]

{phang}
Interrater agreement, nonunique raters, variables record ratings for each rater

{p 8 12 2}
{cmd:kap} {it:{help varname:varname1}} {it:varname2}
{it:varname3} [{it:...}] {ifin} {weight}

{phang}
Interrater agreement, nonunique raters, variables record frequency of ratings

{p 8 14 2}
{cmd:kappa} {varlist} {ifin}

{synoptset 14 tabbed}{...}
{marker options}{...}
{synopthdr}
{synoptline}
{syntab :Main}
{synopt :{opt t:ab}}display table of assessments{p_end}
{synopt :{opt w:gt(wgtid)}}specify how to weight disagreements; see 
          {help kappa##Options:Options} for alternatives{p_end}
{synopt :{opt a:bsolute}}treat rating categories as absolute{p_end}
{synoptline}
{p2colreset}{...}
{p 4 6 2}
{opt fweight}s are allowed; see {help weight}.


{title:Description}

{pstd}
{cmd:kap} (first syntax) calculates the kappa-statistic measure of
interrater agreement when there are two unique raters and two or more ratings.

{pstd}
{cmd:kapwgt} defines weights for use by {cmd:kap} in measuring the importance
of disagreements.

{pstd}
{cmd:kap} (second syntax) and {cmd:kappa} calculate the kappa-statistic
measure when there are two or more (nonunique) raters and two outcomes, more
than two outcomes when the number of raters is fixed, and more than two
outcomes when the number of raters varies.
{cmd:kap} (second syntax) and {cmd:kappa} produce the same results; they
merely differ in how they expect the data to be organized.

{pstd}
{cmd:kap} assumes that each observation is a subject.  
{it:{help varname:varname1}} contains
the ratings by the first rater, {it:varname2} by the second rater, and so on.

{pstd}
{cmd:kappa} also assumes that each observation is a subject.  The variables,
however, record the frequencies with which ratings were assigned.  The first
variable records the number of times the first rating was assigned, the second
variable records the number of times the second rating was assigned, and so
on.


{marker Options}{...}
{title:Options}

{dlgtab:Main}

{phang}
{opt tab} displays a tabulation of the assessments by the two raters.

{phang}
{opt wgt(wgtid)} specifies that {it:wgtid} be used to weight disagreements.
You can define your own weights using {cmd:kapwgt}; in that case, {opt wgt()}
specifies the name of the user-defined matrix.  For instance, you might define

{phang3}
{cmd:. kapwgt mine 1 \ .8 1 \ 0 .8 1 \ 0 0 .8 1}

{pmore}
and then 

{phang3}
{cmd:. kap rata ratb, wgt(mine)}

{pmore}
In addition, two prerecorded weights are available.

{pmore}
{cmd:wgt(w)} specifies weights 1-|i-j|/(k-1),
  where i and j index the rows and columns of the ratings by the two
  raters and k is the maximum number of possible ratings.

{pmore}
{cmd:wgt(w2)} specifies weights 1 - {c -(}(i-j)/(k-1){c )-}^2.

{phang}
{cmd:absolute} is relevant only if {opt wgt()} is also specified.  Option 
{opt absolute} modifies how i, j, and k are defined and how corresponding
entries are found in a user-defined weighting matrix.  When {opt absolute} is
not specified, i and j refer to the row and column index, not to the ratings
themselves.  Say that the ratings are recorded as {c -(}0,1,1.5,2{c )-}.  There
are four ratings; k=4, and i and j are still 1, 2, 3, 4 in the formulas above.
Index 3, for instance, corresponds to rating=1.5.  This is convenient but
can, with some data, lead to difficulties.

{pmore}
When {opt absolute} is specified, all ratings must be integers, and they
  must be coded from the set {c -(}1,2,3,...{c )-}.  Not all values need be
  used; integer values that do not occur are simply assumed to be unobserved.


{title:Remarks}

{pstd}
You have data on individual patients.  There are two raters, and the possible 
ratings are 1, 2, 3, and 4, but neither rater ever used rating 3
In this case, {cmd:kap} would determine the ratings
are from the set {c -(}1,2,4{c )-} because those were the only values
observed.  {cmd:kap} would expect a user-defined weighting matrix would be 3x3
and, if it were not, {cmd:kap} would issue an error message.  In the
formula-based weights, {cmd:wgt(w)} and {cmd:wgt(w2)}, the calculation would be
based on i,j = 1,2,3 corresponding to the three observed ratings 
{c -(}1,2,4{c )-}.

{pstd}
Specifying the {cmd:absolute} option would make it clear that the ratings are
1, 2, 3, and 4; it just so happens that rating==3 was never assigned.  If a 
user-defined weighting matrix were also specified, {cmd:kap} would expect it
to be 4x4 or larger (larger because one can think of the ratings being 1, 2,
3, 4, 5, ... and it just so happens that ratings 5, 6, ... were never
observed, just as rating==3 was not observed).  In the formula-based weights,
the calculation would be based on i,j = 1,2,4.

{pstd}
If all conceivable ratings are observed in the data, specifying
{cmd:absolute} makes no difference.


{title:Example:  two raters}

{phang2}{cmd:. kap rada radb}{p_end}
{phang2}{cmd:. kap rada radb, tab}

{phang2}{cmd:. kap rada radb, wgt(w)}{p_end}
{phang2}{cmd:. kap rada radb, wgt(w2)}

{phang2}{cmd:. kap rada radb, wgt(w) absolute}


{title:Example:  two raters, continued}

{pstd}
Assume that the two raters rate patients into four categories.  You want to use
the weighting matrix:

{center:Rater A {c |}  normal  benign  suspect  cancer}
{center:{hline 8}{c +}{hline 33}}
{center: normal {c |}    1       .8       0        0  }
{center: benign {c |}   .8        1       0        0  }
{center:suspect {c |}    0        0       1       .8  }
{center: cancer {c |}    0        0      .8        1  }

    You type

{phang2}{cmd:. kapwgt xm 1 \ .8 1 \ 0 0 1 \ 0 0 .8 1} {space 13} (define matrix){p_end}
{phang2}{cmd:. kapwgt} {space 30} (print matrix, check correct){p_end}
{phang2}{cmd:. kap rada radb, wgt(xm)} {space 17} (calculate weighted kappa)

{phang2}{cmd:. kap rada radb, wgt(xm) absolute} {space 9} (some ratings unobserved)


{title:Example:  more than two raters, two ratings}

{phang2}{cmd:. kappa pos neg}

{pstd}
{cmd:pos} records the number of raters assessing "positive", {cmd:neg} the
number of raters assessing "negative".


{title:Example:  more than two raters, more than two ratings, fixed number of raters}

{phang2}{cmd:. kappa cat1 cat2 cat3}

{pstd}
{cmd:cat1} records the number of raters assessing category 1; {cmd:cat2}, the
number assessing category 2; and {cmd:cat3}, the number of raters assessing
category 3.  The sum of {cmd:cat1}, {cmd:cat2}, and {cmd:cat3} is constant for
each observation in the data.


{title:Example: more than two raters, more than two ratings, varying number of raters}

{phang2}{cmd:. kappa cat1 cat2 cat3}

{pstd}
This is the same as with a fixed number of raters, except {cmd:cat1} +
{cmd:cat2} + {cmd:cat3} is not constant across observations.  Kappa will be
calculated, but there is no statistic for testing kappa>0 and so none will be
reported.


{title:Also see}

{psee}
Manual:  {bf:[R] kappa}

{psee}
Online:  {helpb tabulate twoway}
{p_end}