27.10.2014 Views

Russel-Research-Method-in-Anthropology

Russel-Research-Method-in-Anthropology

Russel-Research-Method-in-Anthropology

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Sampl<strong>in</strong>g Theory 179<br />

Calculat<strong>in</strong>g Sample Size for Estimat<strong>in</strong>g Means<br />

Nowwearereally close to answer<strong>in</strong>g the question about sample size. Suppose<br />

we want to get the standard error down to RM200 <strong>in</strong>stead of RM400. We<br />

need to solve the follow<strong>in</strong>g equation:<br />

SEM sd 4,000/n 200<br />

n<br />

Solv<strong>in</strong>g for n:<br />

n 4,000<br />

200 20 n 202 400<br />

In other words, to reduce the standard error of the mean from RM400 to<br />

RM200, we have to <strong>in</strong>crease the sample size from 100 to 400 people.<br />

Suppose we <strong>in</strong>crease the sample to 400 and we still get a mean of<br />

RM12,600 and a standard deviation of RM4,000. The standard error of the<br />

mean would then be RM200, and we could estimate, with 95% confidence,<br />

that the true mean of the population was between RM12,208 and 12,992. With<br />

just 100 people <strong>in</strong> the sample, the 95% confidence limits were RM11,816 and<br />

RM13,384. As the standard error goes down, we get narrower—that is, more<br />

precise—confidence limits.<br />

Let’s carry this forward another step. If we wanted to get the standard error<br />

down to RM100 and the 95% confidence <strong>in</strong>terval down to RM200 from<br />

RM400, we would need a sample of 1,600 people. There is a pattern here. To<br />

cut the 95% confidence <strong>in</strong>terval <strong>in</strong> half, from RM800 to RM400, we had to<br />

quadruple the sample size from 100 to 400. To cut the <strong>in</strong>terval <strong>in</strong> half aga<strong>in</strong>,<br />

to RM200, we’d need to quadruple the sample size aga<strong>in</strong>, from 400 to 1,600.<br />

There is another pattern, too. If we want to <strong>in</strong>crease our confidence from<br />

95% to 99% that the true mean of the population is with<strong>in</strong> a particular confidence<br />

<strong>in</strong>terval, we can raise the multiplier <strong>in</strong> formula 7.2 from roughly 2 standard<br />

deviations to roughly 3. Us<strong>in</strong>g the confidence <strong>in</strong>terval of RM400, we<br />

would calculate:<br />

n 3 4,000<br />

400<br />

30 n 30 3 900<br />

We need 900 people, not 400, to be about 99% confident that our sample mean<br />

is with<strong>in</strong> RM400, plus or m<strong>in</strong>us, of the parameter.<br />

Small Samples: The t-Distribution<br />

In anthropology, even do<strong>in</strong>g surveys, we often have no choice about the<br />

matter and have to use small samples.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!