Linking lyrics to statistics

[Quita]

The text in parentheses explains what relationship the words in this poem have to the concepts in epidemiology and stats... these are some tough relationships to remember but I thought that using something I like to do (writing poems and lyrics) would help me learn this high level math. Hope you can use this to assist you. It may not help you or make sense to you but it really has helped me.

Namaska

 

I'm trying to get a PhD

I took statistics but they still tryna school me

to evaluate the health of the community

I was forced to take epidemiology

 

Now what does it mean

Epidemiology's purpose is to halt or prevent disease

Improves health

In not just one but everybody (ind vs population)

 

The gold standard is a RCT (randomized clinical trial)

For evaluating

New treatments,interventions and screenings

Some times we wanna see

If its better than old things (use epi to evaluate screening programs)

 

But I gotta be careful

Pick my subjects right

So my study is replicable

My criteria must be precise (selection of participants to prevent bias)

 

I don’t know where they will go

some over here over there (randomization)

I study series in a case

If I don’t compare (case studies and series)

 

But sometimes I should

To check the relation

If the treatment made a difference

What caused the assumption?(importance of RCT)

 

Then I try to control

With historical groves (historical controls)

And at the same time select my folks out of control (simultaneous randomized control groups)

 

Pick a day, pick a month, pick a time pick a year

Whatever so I don’t see who you are my dear (blinding)

 Because if I know might be inclined to pick a side my dear (bias)

 

So I flip a coin

Tails or head

More likely that they’ll be

Representative(randomization)

 

So I match what I know and on those things control

Age, black or white … stratify like clothes (lol stratified randomization based on a characteristic)

 

I crossover  like a ball player

But I washout before

I play on another team

And I feel my own control (crossover groups receive both treatments and serve as their own contols)

 

Now here comes the fun when the hypothesis is done

Do I accept or reject

I choose the outcome

From four possibilities

O what have I done

 

There’s a difference but I said there was none

type 1 error man what have I done

If I say that it is and there really isnt

 type 2 on my mind man I did it again ( the two types of errors in stats are type 1 and type 2)

 

Probability I make

Errors mistakes

what do you say?

Alpha is the first (type 1 stats use alpha symbol)

The second B.E.T.A (type 2 stats use beta symbol)

 

Alpha is first...also is p

The value of 0.5 or less I like to see ( this value denotes statistical significance)

 

I'm like public enemny

To fight the power that be

In my study Take away 1 from  B (1 subtract beta)

If its hi this tells me (power)

How strong the difference is in my study (correctly id difference)

And also how many people I need (sample size)

 

First I gotta set my expectations

Estimate response to move to perfection

I’m satisfied if alpha <0.1 or 0.5

 

Now the question

Do I do one or do I figure 2 sides( one direction or both directions on a normal distribution curve)

 

If its effective I put every body and take away over the fake (placebo-tx/placebo)- chks efficacy

And this will tell me if what I did worked or if it’s a waste

 

The number I need to treat prevents people from going away

From death or disease I wish they all could stay

The number needed to harm

Tells me who else could be harmed

Don’t wanna feel no cold just wanna stay warm

 

Validate from withing make sure everythings done right

Validate send it out

generalize  hope it works out!

 

So they made me take it

And I gotta learn it

They won't give it to me

so I gotta earn it

 

Gettin  my PhD

A scientist I must be

gotta lot to figure out

plus I wanna lot of cheese...

[Raven Writes]

So typing this ish over and over helps u to retain it as well. Well redpill gonna gt tired of me when I go back to nursing school. But I get it tho

[Quita]

So basically epidemiology is the study of populations. Statistics denotes characteristics of samples.   Symbols and formulas used for statistics dependend on whether a sample or population is being described. A population includes all members in a defined group; a sample is a subset of a population. characteristics of a population is called parameter- characteristics of samples are called statistics... to distinguish them different symbols are used; usually lower case Greek letters denote parameters and Roman letters denote stats... There lies some of the differences between epidemiology and statistics .

There are several study designs out there but the gold standard or best design is termed a randomized control trial or RCT. This is also referred to as an experimental design. What does this mean.  The test relationship of these variables one must conduct an experiment. To do this in an unbiased manner the scientist would first need to randomize the groups before the intervention. This is like tossing a coin and putting all that land on heads in one group and all that land on tails in another. This is a form of control a researcher uses to make sure that the result that is given is not only "by chance" After the groups are randomized...the researcher would want to match on characteristics we "know are related to a dependent variable" such as age, race, weight, etc. ( this is called matching ). Sometimes a researcher does not have enough subjects to do a RCT , so they will use a crossover design in which they subjects serve as their own control. It would be unethical to use this in certain cases, (such as with HIV experimentation, investigation of lung cancer) but this type of design could be used if one wants to test two different interventions. For example... say I have a person who takes care of a patient with alzheimers who is depressed  (this is tested by validated/reliable instruments). I want to help people deal with or eliminate depression... but how?? How do you know what is effective and what is not... ??

After it is assessed that they are depressed we give the group education about alzheimers disease (how to manage the patient, respite care etc.); the test for depression is taken again and scored;  then for a period of time the group has to go through what is termed a washout... to make sure that no "residual" effects are present after first intervention. Then they are given a focus group intervention tailored for reflections, self awareness, etc... and tested for depression again. This is an example of a crossover study.

When a researcher samples, cases are selected from a predetermined population ; everything can not be controlled so the sample may deviate from the defined population's true nature - this is  deviation is called a sampling error - thus inferences from samples to populations are always PROBAILISTIC... always... drawing wrong conclusion is called an error of inference... there are two types of erros... type 1 and type 2.

 

if  the null hypothesis is true and is accepted the response is correct.

if the null is true but is rejected this is called a type 1 error (alpha)

if the null is false and is rejected this is correct response

if null is false and is accepted this is type 2 error.  (beta)

 

to decrease probability of making a type 1 error (which could be detrimental) the researcher sets a significance level called p value at 0.05 then there would be 5 chances in 100 ; setting significance also determines power of  the test...

IMPORTANT NOTE

Measures like this test what works and what doesn't and in what type of person or populaton it works and does not work in ....  People claim that certain things work for certain ailments with no proof. such as taking simylarin(milk thistle)  for hepatitis c ... RCT show no evidence that it is effective but people claim that it works based off anectodotal information (just cuz a nigga told you so lol) I have seen where people will rely on something they beleive will work and it just doesnt) ... I've had patients who used this with no change in quantitative viral replication of HCV RNA... it actually continues to rise with the use of this... so this math is a way to "true" results...  Stats is not perfect

any way this can get confusin and my poem helped me ...lol