Page 15 of 17
Re: Disease Detectives B/C
Posted: April 21st, 2018, 8:29 am
by Froggie
Asking questions a week before states yay
1). Should attack rate be in a percentage form? Should relative risk be in percentage form? Should odds ratio be in percentage form?
2). This was a question on a test: Number ill and exposed = 51, Number not ill and exposed = 34, Number ill and not exposed = 2, number not ill and not exposed = 13. The answer sheet says that the attributable risk is 46.67%, but I keep getting 362%. Can someone explain?
Re: Disease Detectives B/C
Posted: April 21st, 2018, 8:42 am
by UTF-8 U+6211 U+662F
Froggie wrote:Asking questions a week before states yay
1). Should attack rate be in a percentage form? Should relative risk be in percentage form? Should odds ratio be in percentage form?
2). This was a question on a test: Number ill and exposed = 51, Number not ill and exposed = 34, Number ill and not exposed = 2, number not ill and not exposed = 13. The answer sheet says that the attributable risk is 46.67%, but I keep getting 362%. Can someone explain?
1) AFAIK it doesn't matter, but I would just use numbers.
2) (51/85) - (2/15) = 0.4667 = 46.67%
Re: Disease Detectives B/C
Posted: April 21st, 2018, 9:36 am
by Nerd_Bunny
wzhang5460 wrote:Does anyone have any info on the quality of the Disease test for Div B nats?
Last year the test was very long and very thorough. There weren't as many calculations as we expected there to be, but it was a very good test with two good-sized cases. They gave you all the information you needed, then asked you questions about it. All of them followed the rules, as far as I could tell, so we were really pleased with the quality of the test. I don't know about this year's test- I'm unsure if the proctor was from Wright State or will be also writing the test for this year.
Re: Disease Detectives B/C
Posted: April 21st, 2018, 1:00 pm
by Tailsfan101
Froggie wrote:1). Should attack rate be in a percentage form? Should relative risk be in percentage form? Should odds ratio be in percentage form?
As far as I know, attack rate should be in percentage form, while relative risk and odds ratio should not.
Re: Disease Detectives B/C
Posted: April 21st, 2018, 4:28 pm
by UTF-8 U+6211 U+662F
Since the tables are 2x2 anyway, why not just do a 2-proportion z-test instead of a chi-squared test of homogenity?
Re: Disease Detectives B/C
Posted: April 21st, 2018, 4:33 pm
by Unome
UTF-8 U+6211 U+662F wrote:Since the tables are 2x2 anyway, why not just do a 2-proportion z-test instead of a chi-squared test of homogenity?
Pretty sure a z-test wouldn't work because of how the individuals are assigned to categories (e.g. what would your standard deviation be?)
Re: Disease Detectives B/C
Posted: April 21st, 2018, 4:35 pm
by UTF-8 U+6211 U+662F
Unome wrote:UTF-8 U+6211 U+662F wrote:Since the tables are 2x2 anyway, why not just do a 2-proportion z-test instead of a chi-squared test of homogenity?
Pretty sure a z-test wouldn't work because of how the individuals are assigned to categories (e.g. what would your standard deviation be?)
(is not equal to or greater than or less than, depending on the question)
}{n_{\textrm{exposed}}}+\frac{\hat{p}_C(1-\hat{p}_C)}{n_{\textrm{not exposed}}}})
 - 0}{\sigma})
Not sure if pooling the proportions is absolutely necessary.
Re: Disease Detectives B/C
Posted: April 21st, 2018, 5:26 pm
by Unome
UTF-8 U+6211 U+662F wrote:Unome wrote:UTF-8 U+6211 U+662F wrote:Since the tables are 2x2 anyway, why not just do a 2-proportion z-test instead of a chi-squared test of homogenity?
Pretty sure a z-test wouldn't work because of how the individuals are assigned to categories (e.g. what would your standard deviation be?)
(is not equal to or greater than or less than, depending on the question)
Not sure if pooling the proportions is absolutely necessary.
Ok, per internet research z^2 = chi-squared, which is interesting.
Re: Disease Detectives B/C
Posted: April 21st, 2018, 6:42 pm
by UTF-8 U+6211 U+662F
But with a two-proportion z-test, you get the freedom of doing one-tailed tests
Re: Disease Detectives B/C
Posted: April 23rd, 2018, 2:29 pm
by BasuSiddha23
UTF-8 U+6211 U+662F wrote:Froggie wrote:Asking questions a week before states yay
1). Should attack rate be in a percentage form? Should relative risk be in percentage form? Should odds ratio be in percentage form?
2). This was a question on a test: Number ill and exposed = 51, Number not ill and exposed = 34, Number ill and not exposed = 2, number not ill and not exposed = 13. The answer sheet says that the attributable risk is 46.67%, but I keep getting 362%. Can someone explain?
1) AFAIK it doesn't matter, but I would just use numbers.
2) (51/85) - (2/15) = 0.4667 = 46.67%
I'm certain that attack rate is a percentage. As for odds ratio and relative risk, they are shown as decimal numbers as a ratio to 1. For example, an odds ratio or relative risk could be 43 as in 43 to 1.