t2int(x1,x2,Coverage [, pooled:F]), x1 and x2 REAL vectors or matrices with ncols(x1) = ncols(x2), 0 < Coverage < 1 |

t2int(x1,x2,Coverage), where x1 and x2 are REAL vectors, computes a (two sided) t confidence interval for mu1 - mu2 with confidence coefficient Coverage, where mu1 is the population mean of the data in x1 and mu2 is the population mean of the data in x2. A pooled estimate of the standard error of the difference is used. This assumes equal variances. Coverage must be between zero and one. The value is a vector of length 2 giving the lower and upper endpoints of the interval. t2int(x1,x2,Coverage,pooled:F) computes a confidence interval for mu1-mu2 based on the unpooled estimate sqrt(s1^2/n1+s2^2/n2) of the standard error and Satterthwaite's approximate degrees of freedom. It does not assume equal variances. If there are any missing values, they are omitted from the computation and an informative message is printed When x1 and x2 are REAL matrices with ncols(x1) = ncols(x2) = M, t2int(x1,x2,Coverage [,pooled:T]) computes confidence intervals for each column. The result is a 2 by M matrix with the lower and upper limits in rows 1 and 2, respectively. See also tint(), tval(), and t2val().

Gary Oehlert 2003-01-15