"Error Analysis" on the AAMC topics list

[mytoechondriac]

The last part about "Error Analysis"-- including chi, and "student t"<-- WTF is that?? seems completely unrelated to anything on the AAMC practice tests I've seen so far. Do you guys think it would be allright to just ignore this section?

Also, the AAMC has alpha, beta, gamma decay listed twice under separate headings of "nuclear decay" and "radioactive decay." WTF is nuclear decay?? Is that same as radioactive decay?

THANKS!!

---

Members don't see this ad.

 

[engineeredout]

Nuclear decay = radioactive decay.

Error analysis? No freakin way!! I'm a statistics major, I love error analysis!! I never got why they called it the "student t" distribution and not just t distribution. Chi-squared turns girls on, trust me.


Is this under the BS topic list?

 

[superduper12]

Nuclear decay = radioactive decay.

Error analysis? No freakin way!! I'm a statistics major, I love error analysis!! I never got why they called it the "student t" distribution and not just t distribution. Chi-squared turns girls on, trust me.


Is this under the BS topic list?

engineeredout...could you explain these topics?

and yes, it's on the BS topic list.

 

[engineeredout]

engineeredout...could you explain these topics?

and yes, it's on the BS topic list.

The most I've even seen the chi-squared or t distributions apply to biology is in genetic analysis. Basically they are just two methods of probability distribution. Normally you would do some calculations about the statistical probability of events and then look up the t or z (normal bell curve) or chi-squared values in a table.

T-distribution is similar to the standard Z (bell curve) distribution , except that it is narrower in the middle. It is generally used for when there are a lower number of samples in which the statistic (usually the standard mean from a smaller group) is being calcuated from. IE if my group has a mean BAC of .10, what is the range of BACs for a much larger group for a specific confidence (like within 95% or 99%). Again the t-distribution applies to smaller sample sizes, usually less than 30. More than that and the Z distribution is used.

The Chi-squared distribution is one sided, meaning only positive values. It is generally used in error analysis and significance testing, and is also used for testing "Null hypothesies", which are hypothetical scenarios set up to test whether an event will happen or not for a given statistical significance.

Chi-squared can also be represented as (Observed - Expected)^2/Expected, when using the chi-squared distribution to calculate the probability that a given series of data is part of another "expected" series of data.

This is what I've used chi-squared for (outside of math classes): I've been working on a research project involving changes in cell growth on cross-linked PDMS substrates with varying levels of clay nanoparticles. Chi-squared is used if I say to myself "Hmm, I have higher numbers with 1% clay then with .5% clay, but are the numbers different enough to have a statistical significance?" meaning are they far enough outside the wrelm of possibility that the higher numbers could have just been random chance.


Now keep in mind that I know nothing about genetics so maybe a bio major will be of more help, but from what I see in this genetics book I bought, a chi-squared test can be used to determine statistical probabilities of genetic events (IE odds of a family having 9 girls 1 boy). It looks like they also go into hypothesis testing. I have a hard time believing that they'd require something like this on the mcat as its only something you start to learn in upper level undergraduate statistics.

 

[mytoechondriac]

WOW, THANK YOU engineered out. You're kind of awesome.

The most I've even seen the chi-squared or t distributions apply to biology is in genetic analysis. Basically they are just two methods of probability distribution. Normally you would do some calculations about the statistical probability of events and then look up the t or z (normal bell curve) or chi-squared values in a table.

T-distribution is similar to the standard Z (bell curve) distribution , except that it is narrower in the middle. It is generally used for when there are a lower number of samples in which the statistic (usually the standard mean from a smaller group) is being calcuated from. IE if my group has a mean BAC of .10, what is the range of BACs for a much larger group for a specific confidence (like within 95% or 99%). Again the t-distribution applies to smaller sample sizes, usually less than 30. More than that and the Z distribution is used.

The Chi-squared distribution is one sided, meaning only positive values. It is generally used in error analysis and significance testing, and is also used for testing "Null hypothesies", which are hypothetical scenarios set up to test whether an event will happen or not for a given statistical significance.

Chi-squared can also be represented as (Observed - Expected)^2/Expected, when using the chi-squared distribution to calculate the probability that a given series of data is part of another "expected" series of data.

This is what I've used chi-squared for (outside of math classes): I've been working on a research project involving changes in cell growth on cross-linked PDMS substrates with varying levels of clay nanoparticles. Chi-squared is used if I say to myself "Hmm, I have higher numbers with 1% clay then with .5% clay, but are the numbers different enough to have a statistical significance?" meaning are they far enough outside the wrelm of possibility that the higher numbers could have just been random chance.


Now keep in mind that I know nothing about genetics so maybe a bio major will be of more help, but from what I see in this genetics book I bought, a chi-squared test can be used to determine statistical probabilities of genetic events (IE odds of a family having 9 girls 1 boy). It looks like they also go into hypothesis testing. I have a hard time believing that they'd require something like this on the mcat as its only something you start to learn in upper level undergraduate statistics.

 

[futuredoctor10]

So should we be able to do "Error Analysis" including chi, and "student t" calculations on the MCAT?

I haven't seen anyone mention it on SDN that it came up on the MCAT, and it's odd AAMC would put it on a list of possible topics randomly....

 

[engineeredout]

So should we be able to do "Error Analysis" including chi, and "student t" calculations on the MCAT?

Impossible. You have to use tables or excel to be able to find out the probability values.

Example: http://www.medcalc.be/manual/t-distribution.php


And yeah I know I'm awesome. I pick up girls with my statistics knowledge all the time.

 

[Vihsadas]

They could very easily give you a small DOF/Conf Interval table in a passage.

 

[engineeredout]

They could very easily give you a small DOF/Conf Interval table in a passage.

But what is the use of one of those tables if you're unable to do hypothesis testing? Even if they wanted to ask someone about it, it takes far too long.

 

[Vihsadas]

But what is the use of one of those tables if you're unable to do hypothesis testing? Even if they wanted to ask someone about it, it takes far too long.

I dunno, there are some pretty simple Chi-Squared problems that could be asked. Think about flipping a coin, and then the results of say a 100 Tosses, and comparing expected vs. observed.

For hypothesis testing, they could just ask something really simple, like looking at the table to find the conf intervals, and then just finding whether a certain observation is likely in the distribution or not.

 

[engineeredout]

I dunno, there are some pretty simple Chi-Squared problems that could be asked. Think about flipping a coin, and then the results of say a 100 Tosses, and comparing expected vs. observed.

Eh I guess. I don't even see how they would come up with a question like that that relates though. That example would need the person to be testing a coin to see if the coin is properly balance on either side.

A question would have to be like "If the person had three boys and seven girls, and knowing that the expected number of boys and girls were to be even, prove the null hypothesis that the mother of three boys and seven girls isn't a freak or a monster and can reproduce like any normal human". Even in that situation something like N=10 wouldn't be statistically significant.

Wait a minute, that question would be a t distribution, as would your original question I think.




Hell though if they decided to all of a sudden change the BS section to probability and statistics, I'm set

 

[superduper12]

my bad, it's on the PS section...not the BS section.

 

[futuredoctor10]

I dunno, there are some pretty simple Chi-Squared problems that could be asked. Think about flipping a coin, and then the results of say a 100 Tosses, and comparing expected vs. observed.

For hypothesis testing, they could just ask something really simple, like looking at the table to find the conf intervals, and then just finding whether a certain observation is likely in the distribution or not.

Would the background for this be provided in a passage or we should know this going into the exam?

Has this come up on anyone's MCAT before?

 

[Vihsadas]

Would the background for this be provided in a passage or we should know this going into the exam?

Has this come up on anyone's MCAT before?

I've never heard of it coming up before, and it's not in any of the major review books (well at least Kap, TPR and EK)...nevertheless, it is still on that topic list. :/
I'd assume that you'd have enough information in the passage to answer the questions that might come up for a stats problem, but since I've never heard of this being on the exam aside from the topics list the AAMC puts out, I really would have no idea what you should study for.

 

[futuredoctor10]

OK thanks. I think I am going to continue working on other areas of PS for now and then maybe closer to test date I can spend a few hours looking up the listed topics under Error Analysis in wikipedia, the internet or a statistics book.