In charlotte-ngs/asmss2022: Applied Statistical Methods in Animal Science - Spring Semester 2022

knitr::opts_chunk$set(echo = FALSE)

Repeated Measurements

• Same characteristics (Body Weight, BC, ...) measured multiple times for the same animal

if (params$isonline){
  s_rep_obs_path <- "https://charlotte-ngs.github.io/asmss2022/data/asm_bw_bc_rep_obs.csv"
} else {
  s_rep_obs_path <- file.path(here::here(), "docs", "data", "asm_bw_bc_rep_obs.csv")
}
tbl_rep_obs <- readr::read_csv(file = s_rep_obs_path)
knitr::kable(head(tbl_rep_obs, 8),
             booktabs = TRUE,
             longtable = TRUE)

• Column Animal no longer just a counter, it becomes a model factor

Properties

• What is the benefit of repeated measurements?
• Equivalent to more animals in the dataset?
• Assumption: repeated observations of the same animal are more "similar"
• No longer independence of observations

Data Scatterplot

tbl_rep_obs$Animal <- as.factor(tbl_rep_obs$Animal)
ggplot2::ggplot(data = tbl_rep_obs, 
                mapping = ggplot2::aes(x = `Breast Circumference`,
                                       y = `Body Weight`)) + 
  ggplot2::geom_point(ggplot2::aes(color = Animal, 
                                   size = 3))

Statistical Analysis

• Technically, same regression analysis is possible

lm_rep_obs <- lm(`Body Weight` ~ `Breast Circumference`, data = tbl_rep_obs)

• But assumption of independence is violoted
• Consequence of independence

$$var(\mathbf{e}) = \mathbf{I} * \sigma_e^2$$ * Check residuals plot

No Repeated Measures

charlotte-ngs/asmss2022 documentation built on June 7, 2022, 1:33 p.m.