Inferencia estadística

pacman::p_load(tidyverse, NHANES)

Concepto básico

p = tamaño del efecto x tamaño de la muestra

Tamaño de la muestra

250

sampSize <- 250
NHANES_sample <- NHANES %>% 
  dplyr::sample_n(sampSize)

NHANES_sample %>% 
  filter(PhysActive != "NA") %>% 
  ggplot(aes(x = PhysActive, y = BMI)) + 
  geom_boxplot()

NHANES_sample %>% 
  filter(PhysActive != "NA") %>% 
  ggplot(aes(x = BMI)) + 
  geom_density(bins = 10) + 
  facet_grid(PhysActive~.)

Ignoring unknown parameters: bins

NHANES_sample %>% 
  filter(PhysActive != "NA") %>% 
  t.test(BMI ~ PhysActive, data = .)

    Welch Two Sample t-test

data:  BMI by PhysActive
t = 1.6625, df = 183.91, p-value = 0.09812
alternative hypothesis: true difference in means is not equal to 0
95 percent confidence interval:
 -0.2616361  3.0637839
sample estimates:
 mean in group No mean in group Yes 
         28.69872          27.29765 

500

NHANES_sample_x_10 <- sapply(NHANES_sample, rep.int, times = 2)
NHANES_sample_x_10 <- as.data.frame(NHANES_sample_x_10)

NHANES_sample_x_10 %>% 
  filter(PhysActive != "NA") %>% 
  t.test(BMI ~ PhysActive, data = .)

    Welch Two Sample t-test

data:  BMI by PhysActive
t = 2.3572, df = 370, p-value = 0.01893
alternative hypothesis: true difference in means is not equal to 0
95 percent confidence interval:
 0.2322911 2.5698566
sample estimates:
mean in group 1 mean in group 2 
       28.69872        27.29765 

NHANES_sample_x_10 %>% 
  filter(PhysActive != "NA") %>% 
  ggplot(aes(x = BMI)) + 
  geom_density(bins = 10) + 
  facet_grid(PhysActive~.)

Ignoring unknown parameters: bins

NHANES_sample %>% 
  filter(!is.na(BMI), !is.na(PhysActive)) %>% 
  group_by(PhysActive) %>% 
  summarise(BMI_promedio = mean(BMI), sd = sd(BMI), n = n())

NHANES_sample_x_10 %>% 
  filter(!is.na(BMI), !is.na(PhysActive)) %>% 
  group_by(PhysActive) %>% 
  summarise(BMI_promedio = mean(BMI), sd = sd(BMI), n = n())

Tamaño del efecto