Assessing Sample Bias?

I have a large data set, including whites, blacks, and mexicans of both
sexes. I want to determine whether there was sample bias with respect to
gender-race combinations, e.g. were black males under-represented in the
sample?

The textbook formula for a two-sample t test talks about using n1 and n2 for
the standard error expression.

When I want to compare black males (defined as p1) to p2, how should p2 be
defined? Should p2 be defined as the proportion of all nonblack males? Or
should p2 be the proportion of all males in the sample (including blacks)?

Of course, this would affect which value to use for n2 in the standard
error.


.

Relevant Pages

  • Re: Bill Bennett is gambling that no one is listening :)
    ... >>> jail or prison than males who fall on the right side of the curve. ... >>> Black males commit murder at eight times the rate of whites. ... >>the rate of homicide among blacks is higher in young male populations ... White kids caught with drugs don't go to prison nearly as ...
    (soc.retirement)
  • Re: Oh, how I love that John Edwards.
    ... They're all going to be in prison or dead. ... Someone let John Edwards know black males in college outnumber black ... blacks in college: 2.3 million ... whites in college: 12.5 million ...
    (rec.sport.pro-wrestling)
  • Assessing Sample Bias?
    ... I have a large data set, including whites, blacks, and mexicans of both ... the standard error expression. ... Should p2 be defined as the proportion of all nonblack males? ...
    (sci.stat.math)
  • Re: Three score years and ten
    ... for whites, 33.0 for blacks. ... For non-white males, you don't get ... for white females, ... there in 1979-1981, and non-white males get there, again, in 1997. ...
    (alt.usage.english)