Disease Detectives B/C

[GHamski]

I don't know too much about chi squared distributions, but I would assume the "expected" would be the null hypothesis, i.e. that there is no association between the exposure and the condition, and thus the expected value of people affected by the condition would be the attack rate of the whole population times the population exposed.

I'm currently working on finding a good way to calculate this. Apparently a method I got from a practice test was wrong, so I gave out some misinformation earlier. So far it's been hard finding ways to calculate chi-squares in an epidemiology context, but I'm trying. I'll share the correct calculations here once I find them.
Sorry for double post.

EDIT: If anyone could look at the 2017 Princeton test and figure out how they got the answer to question 7 all our problems will be solved. It'a on the test exchange. I'll link to the test and key:
https://scioly.org/tests/files/diseased ... n_test.pdf
https://scioly.org/tests/files/diseased ... on_key.pdf

I'm going to go out on a limb here and assume he meant number 7 from the statistics section of the test. As for how they got the numbers on the key, I was able to get them by doing the Chi-Square test in the TI-84. From what I've tested, it seems like how you organize your numbers does not really matter so long as you are consistent. If you've ever worked with similar triangles, I believe its a similar concept. You would either want to set it up as:
Category 1 Category 2
Yes 41 64
No 216 180

Or:
Yes No
C1 41 216
C2 64 180
But in this scenario C1 represents Placebo, C2 represents Drug A, and Yes and no represent Symptoms relieved or not relieved.
Whenever you set up your matrix like the tables above and perform the test in the calculator, you'll find that it stores the expected values in a second matrix that you can then go and view to get the answers they got on the key. The calculator gave me a p value of .0047 and a matrix of
53.862 203.14
51.138 192.86
for the expected values.

[Nerd_Bunny]

I'm currently working on finding a good way to calculate this. Apparently a method I got from a practice test was wrong, so I gave out some misinformation earlier. So far it's been hard finding ways to calculate chi-squares in an epidemiology context, but I'm trying. I'll share the correct calculations here once I find them.
Sorry for double post.

EDIT: If anyone could look at the 2017 Princeton test and figure out how they got the answer to question 7 all our problems will be solved. It'a on the test exchange. I'll link to the test and key:
https://scioly.org/tests/files/diseased ... n_test.pdf
https://scioly.org/tests/files/diseased ... on_key.pdf

I'm going to go out on a limb here and assume she meant number 7 from the statistics section of the test. As for how they got the numbers on the key, I was able to get them by doing the Chi-Square test in the TI-84. From what I've tested, it seems like how you organize your numbers does not really matter so long as you are consistent. If you've ever worked with similar triangles, I believe its a similar concept. You would either want to set it up as:
Category 1 Category 2
Yes 41 64
No 216 180

Or:
Yes No
C1 41 216
C2 64 180
But in this scenario C1 represents Placebo, C2 represents Drug A, and Yes and no represent Symptoms relieved or not relieved.
Whenever you set up your matrix like the tables above and perform the test in the calculator, you'll find that it stores the expected values in a second matrix that you can then go and view to get the answers they got on the key. The calculator gave me a p value of .0047 and a matrix of
53.862 203.14
51.138 192.86
for the expected values.

FTFY

How does the TI-84 perform its calculations? I don't have one (yet) so I'd like to be able to do this just with pencil and paper. I had trouble figuring out how the test got the expected values, I think I can figure out the rest from there.

EDIT: Yes, I did mean Q7 in the statistics section, sorry that I wasn't clear on that.

[WhatScience?]

I'm going to go out on a limb here and assume she meant number 7 from the statistics section of the test. As for how they got the numbers on the key, I was able to get them by doing the Chi-Square test in the TI-84. From what I've tested, it seems like how you organize your numbers does not really matter so long as you are consistent. If you've ever worked with similar triangles, I believe its a similar concept. You would either want to set it up as:
Category 1 Category 2
Yes 41 64
No 216 180

Or:
Yes No
C1 41 216
C2 64 180
But in this scenario C1 represents Placebo, C2 represents Drug A, and Yes and no represent Symptoms relieved or not relieved.
Whenever you set up your matrix like the tables above and perform the test in the calculator, you'll find that it stores the expected values in a second matrix that you can then go and view to get the answers they got on the key. The calculator gave me a p value of .0047 and a matrix of
53.862 203.14
51.138 192.86
for the expected values.

FTFY

How does the TI-84 perform its calculations? I don't have one (yet) so I'd like to be able to do this just with pencil and paper. I had trouble figuring out how the test got the expected values, I think I can figure out the rest from there.

EDIT: Yes, I did mean Q7 in the statistics section, sorry that I wasn't clear on that.

-__-
Thought you meant number 7 overall

[GHamski]

I'm going to go out on a limb here and assume she meant number 7 from the statistics section of the test. As for how they got the numbers on the key, I was able to get them by doing the Chi-Square test in the TI-84. From what I've tested, it seems like how you organize your numbers does not really matter so long as you are consistent. If you've ever worked with similar triangles, I believe its a similar concept. You would either want to set it up as:
Category 1 Category 2
Yes 41 64
No 216 180

Or:
Yes No
C1 41 216
C2 64 180
But in this scenario C1 represents Placebo, C2 represents Drug A, and Yes and no represent Symptoms relieved or not relieved.
Whenever you set up your matrix like the tables above and perform the test in the calculator, you'll find that it stores the expected values in a second matrix that you can then go and view to get the answers they got on the key. The calculator gave me a p value of .0047 and a matrix of
53.862 203.14
51.138 192.86
for the expected values.

FTFY

How does the TI-84 perform its calculations? I don't have one (yet) so I'd like to be able to do this just with pencil and paper. I had trouble figuring out how the test got the expected values, I think I can figure out the rest from there.

EDIT: Yes, I did mean Q7 in the statistics section, sorry that I wasn't clear on that.

I would highly recommend getting a graphing calculator, it's very useful for many things and it makes your math life way easier. Not to mention you will need it for any higher math courses you go into. For calculating the expected values you will still need a calculator since you will be dealing with proportions (unless you think you will have the time to perform them all by hand), but it would be possible with just a regular old calculator. Expected values are calculated using a lot of proportions, and, similar to before, there are multiple ways to do it that are ultimately equivalent. In this case you have a 2 by 2 table, which makes things a whole lot easier. You want to start off by totaling each column and row individually, then adding those two totals together (either way columns+columns or rows+rows doesn't matter, they should be the same)
41 216 =257
64 180 =244
105 396 501
Then, assuming that the events in the table are independent of each other, you would expect the proportion of each part in the total sample to match that in its own category. You could either use the Proportion of your categories out of the total (Placebo and Drug A) mulitplied by your total number of Yeses and Nos (Relieved or not Relieved), which would get you
(257/501)*(105)= 53.86 people expected to be relieved with the placebo.
(244/501)*(105)= 51.14 people expected to be relieved with Drug A.
(257/501)*(396)= 203.14 people expected to not be relieved with the placebo.
(244/501)*(396)= 192.86 people expected to not be relieved with Drug A.
Or you could do the same thing but reversed, using the proportion of your Yeses and Nos out of the total multiplied by your total number of people in each category, which would get you
(105/501)*(257)= 53.86 people expected to be relieved with the placebo.
(105/501)*(244)= 51.14 people expected to be relieved with Drug A.
(396/501)*(257)= 203.14 people expected to not be relieved with the placebo.
(396/501)*(244)= 192.86 people expected to not be relieved with Drug A.
So it just ends up being the same thing since multiplication is commutative, just pick a proportion of the total of one column or row out of the total in the sample, then multiply by the opposite of what you picked (ie if you picked to use the proportion of the total of a row out of the whole sample, you would multiply by the total of one column and vice versa).
The same process would apply with a 3 by 3 table or other sizes, it would just take more time and work.

[Nerd_Bunny]

I'm going to go out on a limb here and assume she meant number 7 from the statistics section of the test. As for how they got the numbers on the key, I was able to get them by doing the Chi-Square test in the TI-84. From what I've tested, it seems like how you organize your numbers does not really matter so long as you are consistent. If you've ever worked with similar triangles, I believe its a similar concept. You would either want to set it up as:
Category 1 Category 2
Yes 41 64
No 216 180

Or:
Yes No
C1 41 216
C2 64 180
But in this scenario C1 represents Placebo, C2 represents Drug A, and Yes and no represent Symptoms relieved or not relieved.
Whenever you set up your matrix like the tables above and perform the test in the calculator, you'll find that it stores the expected values in a second matrix that you can then go and view to get the answers they got on the key. The calculator gave me a p value of .0047 and a matrix of
53.862 203.14
51.138 192.86
for the expected values.

FTFY

How does the TI-84 perform its calculations? I don't have one (yet) so I'd like to be able to do this just with pencil and paper. I had trouble figuring out how the test got the expected values, I think I can figure out the rest from there.

EDIT: Yes, I did mean Q7 in the statistics section, sorry that I wasn't clear on that.

I would highly recommend getting a graphing calculator, it's very useful for many things and it makes your math life way easier. Not to mention you will need it for any higher math courses you go into. For calculating the expected values you will still need a calculator since you will be dealing with proportions (unless you think you will have the time to perform them all by hand), but it would be possible with just a regular old calculator. Expected values are calculated using a lot of proportions, and, similar to before, there are multiple ways to do it that are ultimately equivalent. In this case you have a 2 by 2 table, which makes things a whole lot easier. You want to start off by totaling each column and row individually, then adding those two totals together (either way columns+columns or rows+rows doesn't matter, they should be the same)
41 216 =257
64 180 =244
105 396 501
Then, assuming that the events in the table are independent of each other, you would expect the proportion of each part in the total sample to match that in its own category. You could either use the Proportion of your categories out of the total (Placebo and Drug A) mulitplied by your total number of Yeses and Nos (Relieved or not Relieved), which would get you
(257/501)*(105)= 53.86 people expected to be relieved with the placebo.
(244/501)*(105)= 51.14 people expected to be relieved with Drug A.
(257/501)*(396)= 203.14 people expected to not be relieved with the placebo.
(244/501)*(396)= 192.86 people expected to not be relieved with Drug A.
Or you could do the same thing but reversed, using the proportion of your Yeses and Nos out of the total multiplied by your total number of people in each category, which would get you
(105/501)*(257)= 53.86 people expected to be relieved with the placebo.
(105/501)*(244)= 51.14 people expected to be relieved with Drug A.
(396/501)*(257)= 203.14 people expected to not be relieved with the placebo.
(396/501)*(244)= 192.86 people expected to not be relieved with Drug A.
So it just ends up being the same thing since multiplication is commutative, just pick a proportion of the total of one column or row out of the total in the sample, then multiply by the opposite of what you picked (ie if you picked to use the proportion of the total of a row out of the whole sample, you would multiply by the total of one column and vice versa).
The same process would apply with a 3 by 3 table or other sizes, it would just take more time and work.

Thank you so much! I'll definitely try to get one of those calculators, but this is really helpful.

[UTF-8 U+6211 U+662F]

Anyone have any information about who to contact after an investigation has been conducted?

Still wondering about this :/

[IcsTam]

Anyone have any information about who to contact after an investigation has been conducted?

Still wondering about this :/

I would recommend looking at breakdowns of the ten steps. Generally, if it was purely a field investigation, you would have to notify the proper authorities (i.e. policymakers in the CDC if you were an investigator.) Past that, it would go to the press and media.

See step 13 here: https://www.cdc.gov/ophss/csels/dsepd/s ... tml#step13
Page 15 of this document: http://www2.wpro.who.int/internet/files ... gation.pdf

[Nerd_Bunny]

Are experimental or trial studies the same as randomized controlled trials? I feel like this is a stupid question but I had to ask just in case.

[UTF-8 U+6211 U+662F]

Are experimental or trial studies the same as randomized controlled trials? I feel like this is a stupid question but I had to ask just in case.

I think so.

[WhatScience?]

I have decided to take Unome's advice and start prepping for next year. I will be in C for the first time.

Does anyone have a list of statistics i need to know.