{smcl}
{* 4feb2003}{...}
{hline}
help for {hi:cij}, {hi:ciji}
{hline}
{title:Binomial confidence intervals for proportions (Jeffreys prior)}
{p 15 19}{cmd:cij}{space 2}[{it:varlist}] [{it:weight}] [{cmd:if} {it:exp}]
[{cmd:in} {it:range}] [{cmd:,} {cmdab:l:evel(}{it:#}{cmd:)} {cmdab:t:otal} ]
{p 15 19}{cmd:ciji} {it:#obs} {space 4} {it:#succ} {space 3}
[{cmd:,} {cmdab:l:evel(}{it:#}{cmd:)} ]
{p}{cmd:by} {it:...} {cmd::} may be used with {cmd:cij} (but not with
{cmd:ciji}); see help {help by}.
{p}{cmd:fweight}s are allowed with {cmd:cij}; see help {help weights}.
{title:Description}
{p}{cmd:cij} computes standard errors and binomial confidence intervals for
each variable in {it:varlist}, which should be 0/1 binomial variables.
{cmd:ciji} is the immediate form of {cmd:cij}, for which specify the number of
observations and the number of successes. See help {help immed} for more on
immediate commands. With both commands confidence intervals are calculated
based on the Jeffreys uninformative prior of a beta distribution
with parameters 0.5 and 0.5.
{title:Remarks}
{p}Suppose we observe {it:n} events and record {it:k} successes. Here as usual
"success" is conventional terminology for whatever is coded 1. For a 95%
confidence interval, for example, we then take the 0.025 and 0.975 quantiles of
the beta distribution with parameters {it:k} + 0.5 and {it:n - k} + 0.5. This
Bayesian procedure has a frequentist interpretation as a continuity-corrected
version of the so-called exact (Clopper-Pearson) confidence interval, produced
by {cmd:ci, binomial}, which takes (in the same example) the 0.025 quantile
of beta({it:k}, {it:n} - {it:k} + 1) and the 0.975 quantile of
beta({it:k} + 1, {it:n} - {it:k}). The lower limit if all values are 0 is taken
to be 0 and the upper limit if all values are 1 is taken to be 1. Among other
properties, note that this interval is typically less conservative than the
exact interval, so that coverage probabilities are on average close to the
nominal confidence level. From a Bayesian point of view, however,
the whole of the posterior distribution is much more fundamental than any
interval derived from it.
{p}See Brown {it:et al.} (2001) for a much fuller discussion and an entry to
the literature. Brown {it:et al.} (2002) provide supporting technical
background to that paper. Among many references, Agresti (2002, pp.14-21),
Agresti and Coull (1998), Newcombe (1998, 2001) and Vollset (1993) provide clear
and helpful context. Williams (2001, Ch.6) provides a lively alternative
treatment of confidence intervals for one-parameter models.
The original work on uninformative priors was by Harold Jeffreys
(1946; 1948, Ch.3.9; 1961, Ch.3.10).
The actuary Wilfred Perks (1947) independently produced very similar ideas.
Later discussions include Good (1965, esp. pp.18-19),
Rubin and Schenker (1987), Lee (1989, esp. p.93; 1997, esp. pp.88-89),
Gelman {it:et al.} (1995, esp. pp.55-56), or Carlin and Louis
(1996, esp. pp.50-54; 2000, esp. pp.42-46). For more on
Jeffreys (1891-1989), see Cook (1990), Lindley (2001) or Lindley
{it:et al.} (1991).
{p}The method of calculating beta quantiles used here is based on the
fact that if {it:Y} is distributed as beta({it:a,b}) and {it:X} is
distributed as {it:F}(2{it:a},2{it:b}), then {it:Y} = {it:aX} /
({it:b} + {it:aX}). See (e.g.) Cram{c e'}r (1946, pp.241-4) for background
or Lee (1989, p.251; 1997, p.291). In Stata 8, it can be done directly
with {cmd:invibeta()}.
{title:Options}
{p 0 4}{cmd:level(}{it:#}{cmd:)} specifies the confidence level, in percent,
for confidence intervals; see help {help level}.
{p 0 4}{cmd:total} is for use with the {cmd:by} {it:...} {cmd::} prefix; it
requests that, in addition to output for each by-group, output be added for all
groups combined.
{title:Examples}
{inp:. cij foreign}
{inp:. ciji 10 1}{right:(10 binomial events, 1 observed success)}
{title:Author}
Nicholas J. Cox, University of Durham, U.K.
n.j.cox@durham.ac.uk
{title:Acknowledgements}
{p 8 8}Alan Feiveson suggested the Cram{c e'}r reference.
John R. Gleason increased my interest in this problem.
{title:References}
{p}Agresti, A. 2002. {it:Categorical data analysis.}
Hoboken, NJ: John Wiley.
{p}Agresti, A. and Coull, B.A. 1998. Approximate is better than "exact" for
interval estimation of binomial proportions. {it:American Statistician}
52: 119-126.
{p}Brown, L.D., Cai, T.T., DasGupta, A. 2001. Interval
estimation for a binomial proportion. {it:Statistical Science}
16: 101-133.
{p}Brown, L.D., Cai, T.T., DasGupta, A. 2002. Confidence intervals
for a binomial proportion and asymptotic expansions. {it:Annals of Statistics}
30: 160-201.
{p}Carlin, B.P. and Louis, T.A. 1996/2000.
{it:Bayes and empirical Bayes methods for data analysis.}
Boca Raton, FL: Chapman and Hall/CRC (1996: London: Chapman and Hall.)
{p}Cook, A.H. 1990. Sir Harold Jeffreys.
{it:Biographical Memoirs of Fellows of the Royal Society}
36: 303-333.
{p}Cram{c e'}r, H. 1946. {it:Mathematical methods of statistics.}
Princeton, NJ: Princeton University Press.
{p}Gelman, A., Carlin, J.B., Stern, H.S., Rubin, D.B. 1995.
{it:Bayesian data analysis.} London: Chapman and Hall.
{p}Good, I.J. 1965. {it:The estimation of probabilities:}
{it:an essay on modern Bayesian methods}. Cambridge, MA: MIT Press.
{p}Jeffreys, H. 1939/1948/1961. {it:Theory of probability.}
Oxford: Oxford University Press.
{p}Jeffreys, H. 1946. An invariant form for the prior probability
in estimation problems.
{it:Proceedings of the Royal Society A} 186: 453-461.
Reprinted in Jeffreys, H. and Jeffreys, B.S. (eds) 1977.
{it:Collected papers of Sir Harold Jeffreys on geophysics and other sciences.}
{it: Volume 6: Mathematics, probability and miscellaneous other sciences.}
London: Gordon and Breach, 403-411.
{p}Lee, P.M. 1989/1997.
{it:Bayesian statistics: an introduction.}
London: Edward Arnold.
{p}Lindley, D.V. 2001. Harold Jeffreys. In Heyde,
C.C. and Seneta, E. (eds) {it:Statisticians of the centuries.}
New York: Springer, 402-405.
{p}Lindley, D.V., Bolt, B.A., Huzurbazar, V.S., Jeffreys,
B.S., Knopoff, L. 1991. [articles on Harold Jeffreys]
{it:Chance} 4(2): 10-26.
{p}Newcombe, R.G. 1998. Two-sided confidence intervals for
the single proportion: comparison of seven methods.
{it:Statistics in Medicine} 17: 857-872.
{p}Newcombe, R.G. 2001. Logit confidence intervals and
the inverse sinh transformation.
{it:American Statistician} 55: 200-202.
{p}Perks, W. 1947. Some observations on inverse probability
including a new indifference rule.
{it:Journal, Institute of Actuaries} 73: 285-334.
{p}Rubin, D.M. and Schenker, N. 1987. Logit-based interval
estimation for binomial data using the Jeffreys prior.
{it:Sociological Methodology} 17: 131-144.
{p}Vollset, S.E. 1993. Confidence intervals for a
binomial proportion. {it:Statistics in Medicine} 12: 809-824.
{p}Williams, D. 2001.
{it:Weighing the odds: a course in probability and statistics.}
Cambridge: Cambridge University Press.
{title:Also see}
Manual: {hi:[R] ci}
{p 0 19}On-line: help for {help ci}, {help bitest}, {help immed}{p_end}