PROBABILIDAD Flashcards
Probabilidad objetiva y subjetiva
Empirical probability dado STATA
Binomial probability
Binomial probability
Binomial probability
Poisson
Poisson
Bar graph binomial distribution
Poisson formula
En STATA
2 dice in STATA
PROBABILITY PROBLEM
STATA IS LEFTLIST
Pnorm graphs to check normality should be used with… sample sizes
Large
In medicine we usually have a long R tail, so we use transformations of the variable so that they can follow a normal distribution
NORMAL DISTRIBUTION SUMMARY
Random error
Systematic error-bias
Sampling in STATA
Sampling in STATA
Para no perder el trabajo cuando hacemos RANDOM SAMPLING IN STATA, debemos usar los comandos…
preserve
….
restore
Count
Random allocation “randomization”
RANDOMIZATION IN STATA
What can you do to get again exactly the same random allocation or the same random sampling in STATA?
Set seed 9999
In internal validity, what is more important? Biological inference or statistical respresentativeness?
BIOLOGICAL INFERENCE: for causal inference.
Representativeness can become a fallacy and it is needed for external validity.
REPLICAR EL CENTRAL LIMIT THEOREM EN STATA: sacar una muestra y su media
PROBABILITY INTERVALS VS CI (VERY IMPORTANT)
With age we cannot apply normal distribution, it is calculating a probability interval. We can only calculate the CI with a sample in ESTIMATORS.
¿Qué observamos?
200 intervalos de confianza del 95% para la media de edad calculados en 200 muestras, cada una de 36 sujetos.
AL FINAL TENEMOS 5 CI WHICH FAIL, NO TOCAN LA REALIDAD, por ello 95% CI.
PROBLEMA RESUELTO CON DISTRIBUCIÓN NORMAL Y BINOMIAL
PROBLEMA A RESOLVER CON NORMAL Y POISSON
Probability following a normal distribution
SE
T tail STATA
Rejecting the null hypothesis
Hypothesis testing for a proportion
Hypothesis testing for a proportion in STATA
Always look at 2-tailed
Hypothesis testing for a mean
INTERPRETACIÓN p value
Two-sample test of proportions in STATA
Z test for proportions in STATA
Chi 2 en STATA
Getting expected values in STATA
If expected values in chi are lower than 5, what do we use?
FISHER EXACT TEST
Pearson chi 2 with rows and columns
Here we have 3 DF.
En la última foto, a la izquierda sacamos expected values y a la derecha observed
Chi 2 y z with only one variable and many categories
Mc Neman paired test for proportions
Chi2 for linear trend
1df, ordinal variables
T test parametric indecent variables STATA
corregir CI en t-test
F distribution para comparar SD, si es mayor que 0.05 cumple homogeneidad.
Mirar t-tail más pequeña
Sd test con calculadora en STATA
T test heterogeneous variances STATA
Paired T-test en STATA
Paired t test but with mean vs external reference
ESQUEMA COMPARISON OF MEANS
Mann Whitney U test STATA