I have data which shows the incidences of a health condition as a % of the population. Can these % measures be used for t-testing?
I cannot use the actual number of incidences as the population size is growing.
2007-01-30 20:28:49 · 3 answers · asked by Tatiana 3 in Science & Mathematics ➔ Mathematics
I believe that the percentages can be used, but you have to show that those percentages are statistically relevant by showing that the percentage is more than 2 or 3 standard deviations from the norm. Remember, a t-test provides statistical relevance of the data. It does not tell you that a percent of a population that has the health condition. A t-test tells you whether the subset that you are studying has the condition to a higher or lower extent than the average.
For example, suppose I want to show that it is statistically relevant that 10% of people in New York city are over six feet tall. I first need to know the percentage of the total population of the US (or ideally the world) that is over 6 feet tall. If I find out that 5% of the US population is over 6 feet tall, then I know to a certain percentage certainty (in this case 65%, I think) that it is statistically relevant because 10% lies 2 standard deviations away from the norm.
2007-01-30 21:50:49 · answer #1 · answered by Matthew 2 · 3⤊ 0⤋
M-maybe.
2007-01-31 04:36:33 · answer #2 · answered by Anonymous · 0⤊ 2⤋
Thats a nasty stutter you've got there girl!
2007-01-31 05:01:23 · answer #3 · answered by Eegit 1 · 0⤊ 3⤋