Upper bound =
|
d_hat +
|
|
z( 100 - ( 2 * ( 100 - N ) ) ) *
|
|
square root [
|
|
error s1 ( h1 )
( 1 - error s1 ( h1 ) )
/ n1 +
|
|
error s2 ( h2 )
( 1 - error s2 ( h2 ) )
/ n2 ]
|
Lower bound =
|
d_hat -
|
|
z( 100 - ( 2 * ( 100 - N ) ) ) *
|
|
square root [
|
|
error s1 ( h1 )
( 1 - error s1 ( h1 ) )
/ n1 +
|
|
error s2 ( h2 )
( 1 - error s2 ( h2 ) )
/ n2 ]
|