Disease Detectives B/C

[yang573]

Does anyone know how to calculate OR and RR when one of the numbers in the 2x2 table is 0?

[LIPX3]

Does anyone know how to calculate OR and RR when one of the numbers in the 2x2 table is 0?

I think it is 0 or undefined, defending on what variable is 0.

[ampy1234567]

I saw on one of the nationals tests that you're supposed to add 0.5 to all the numbers on the table, but it's probably a better idea to just go with 0 or undefined anyways

[QuantumTech]

@people who have participated in state competition for Disease Detectives for multiple years: Have you seen questions or a table where they provide a disease name and ask what type of disease it is (bacterial, viral, fungal, etc.) and what vector causes it? I need to make space for stats tables (t-table, z-table etc.) so I need to free up some space. Right now my cheat sheet has background info on all the common diseases (where it is endemic, historical outbreaks, amount of people affected). Will that be useful on tests (I will memorize if I have too) or should I just delete it?

[maxxxxx]

@people who have participated in state competition for Disease Detectives for multiple years: Have you seen questions or a table where they provide a disease name and ask what type of disease it is (bacterial, viral, fungal, etc.) and what vector causes it? I need to make space for stats tables (t-table, z-table etc.) so I need to free up some space. Right now my cheat sheet has background info on all the common diseases (where it is endemic, historical outbreaks, amount of people affected). Will that be useful on tests (I will memorize if I have too) or should I just delete it?

I've only competed in Disease Detectives at states twice(and one of them was 4 years ago so I don't remember much) but they both had sections regarding an outbreak of a disease which asked questions about spread, treatment, and cause. Also my regionals test this year was 20% writing what kind of agent caused a disease. You should probably keep the specific disease info. I was also under the impression that if they expected you to look up z and t statistics they would provide the tables because that's a lot to ask the competitors of(the tables would take up a significant amount of a note sheet just to be readable, and it wouldn't comprise very much of the test).

[dcrxcode]

@people who have participated in state competition for Disease Detectives for multiple years: Have you seen questions or a table where they provide a disease name and ask what type of disease it is (bacterial, viral, fungal, etc.) and what vector causes it? I need to make space for stats tables (t-table, z-table etc.) so I need to free up some space. Right now my cheat sheet has background info on all the common diseases (where it is endemic, historical outbreaks, amount of people affected). Will that be useful on tests (I will memorize if I have too) or should I just delete it?

I've definitely seen tests (at all levels) that list a disease name and then ask what agent transmit the disease. Usually the questions will use Bacteria, Virus, Fungal, Protozoan, and sometimes I've also seen Helminth, Apicomplexan, and Prion. Definitely keep the disease information, but try to familiarize yourself with it as much as possible since Disease Detectives Tests are usually less about the disease themselves but rather about investigation of these diseases. For stats tables, you won't need entire tables - it is usually more than sufficient to have z, t, and chi-squared areas for . That being said, I've seen very few questions that involve having access to entire distribution tables.

[QuantumTech]

@people who have participated in state competition for Disease Detectives for multiple years: Have you seen questions or a table where they provide a disease name and ask what type of disease it is (bacterial, viral, fungal, etc.) and what vector causes it? I need to make space for stats tables (t-table, z-table etc.) so I need to free up some space. Right now my cheat sheet has background info on all the common diseases (where it is endemic, historical outbreaks, amount of people affected). Will that be useful on tests (I will memorize if I have too) or should I just delete it?

I've definitely seen tests (at all levels) that list a disease name and then ask what agent transmit the disease. Usually the questions will use Bacteria, Virus, Fungal, Protozoan, and sometimes I've also seen Helminth, Apicomplexan, and Prion. Definitely keep the disease information, but try to familiarize yourself with it as much as possible since Disease Detectives Tests are usually less about the disease themselves but rather about investigation of these diseases. For stats tables, you won't need entire tables - it is usually more than sufficient to have z, t, and chi-squared areas for . That being said, I've seen very few questions that involve having access to entire distribution tables.

OK thanks for the tip! Do you know when to use Bessel's correction for standard deviation? My understanding right now is to use it only when I am doing an actual stats test like z and t test. I did a practice test and one of the questions was testing for understanding of the concept only: "Find standard deviation of this list of numbers: 27,33,50,67,78,89,103,4. " The answer does not use Bessel's correction, so is my understanding correct?

Another question I have is for reading the stats table for z test... If the critical point we need is in between two numbers, like say 1.64 and 1.65, do we take the one nearest to the critical point or the average (1.645)?

[yang573]

OK thanks for the tip! Do you know when to use Bessel's correction for standard deviation? My understanding right now is to use it only when I am doing an actual stats test like z and t test. I did a practice test and one of the questions was testing for understanding of the concept only: "Find standard deviation of this list of numbers: 27,33,50,67,78,89,103,4. " The answer does not use Bessel's correction, so is my understanding correct?

Bessel's correction is used when you're finding the statistics of a sample. The standard deviation of some given data is based on the difference of individual values from the mean, therefore sample standard deviation will (almost) always be smaller than the population standard deviation because the sampled values are (almost) always closer to the sample mean than the population mean. Thus, Bessel's correction increases the standard deviation and, in this way, decreases this bias when estimating the population standard deviation from a sample standard deviation. The problem you were given is interesting, because that's straight up testing stats knowledge rather than application of stats in analytical epidemiology. Regardless, you're right; the standard deviation of those numbers would not use Bessel's correction because that is not a sample.

[yang573]

Another question I have is for reading the stats table for z test... If the critical point we need is in between two numbers, like say 1.64 and 1.65, do we take the one nearest to the critical point or the average (1.645)?

I think you take the number closest to the critical point because critical points are based off of the normal model, and you can't just use averages to find points on a curved graph. If you take a look at a distribution table, you'll see that the jumps between critical points aren't exactly predictable. Don't quote me on this though; I'll have to verify this assumption somewhere.

[dcrxcode]

OK thanks for the tip! Do you know when to use Bessel's correction for standard deviation? My understanding right now is to use it only when I am doing an actual stats test like z and t test. I did a practice test and one of the questions was testing for understanding of the concept only: "Find standard deviation of this list of numbers: 27,33,50,67,78,89,103,4. " The answer does not use Bessel's correction, so is my understanding correct?

Another question I have is for reading the stats table for z test... If the critical point we need is in between two numbers, like say 1.64 and 1.65, do we take the one nearest to the critical point or the average (1.645)?

If you look at the normal distribution function, it's a curve that does have a formula that can be found here: http://mathworld.wolfram.com/NormalDistribution.html. The curve means the slope of the function changes depending on the test statistic (which for the z-test, is ). So it's a little unpredictable to guess which value is closest (I suppose you could try using derivatives and such).

But using a TI system, the area under a curve can be returned using normCdf(low, high) - this would return the actual value of the probability of a value being in the selected part of the normal distribution. I doubt that a disease detectives test would use something that would not be on a "nice" value, i.e. 90%, 95%, or 99% levels. If they did not stick to such values, then they would likely provide the probability for the value that you will need. If you did end up getting this type of question on a test, I'd just use your judgement and make a "best guess" (although you might want to check to make sure you are computing the test statistic correctly).