In ISU-PrairiePlusPigs/RuckmanMeta: Contains meta data for the Ruckman experiment

library(dplyr)
library(ggplot2)

Here is the actual code used to perform the randomization.

To recreate the process, we use only the columns available in the original csv file.

fields <- RuckmanMeta::fields %>%
  select(field,block,whole_plot,area_acre,manure,x,y)

The design of the experiment is a split-plot design within blocks augmented with a non-randomized study.

Randomization

The randomization has two steps:

1. Randomly assign harvest year to the whole plot
2. Randomly assign treatments to the fields within the whole plot

#################################################
### Randomization
#################################################
set.seed(38) # For reproducibility

fields <- fields %>%

  # Randomize initial harvest
  group_by(block) %>%
  mutate(harvest = as.integer(ifelse(whole_plot == sample(whole_plot,1), 2019, 2020)),
         whole_plot = paste0(block, ifelse(harvest==2019, "a", "b"))) %>% # Rename whole_plot to coincide with harvest year

  # Randomize treatment
  group_by(block,whole_plot,manure) %>%
  mutate(trt = sample(c("FB","LoDiv","HiDiv")),
         trt = ifelse(manure=="no", "HiDiv", trt)) %>% # non-randomization augmentation
  ungroup()

We had two criteria to satisfy for a successful randomization:

• Fields 13a and 13b must have exactly one FB treatment between the two of them
• Mean and SD of field sizes in each treatment are approximately the same

# Check to ensure fields 13a and 13b have exactly one FB treatment
fields %>%
  filter(field %in% c("13a","13b"))

# Check field sizes for each treatment-manure combination to ensure the
# mean and sd are approximately the same
fields %>%
  group_by(trt,manure) %>%
  summarize(n = n(),
            mean = mean(area_acre),
            sd   = sd(area_acre)) %>%
  arrange(manure, trt)

An iterative process was undertaken, updating the set.seed value until these criteria were satisfied. The final randomization was

fields %>% as.data.frame

ggplot(fields, aes(x=x, y=y, label=harvest)) +
  geom_text() +
  theme_bw() +
  xlim(c(0,11)) +
  ylim(c(0,14))

ggplot(fields, aes(x=x, y=y, label=trt)) +
  geom_text() +
  theme_bw() +
  xlim(c(0,11)) +
  ylim(c(0,15))

ISU-PrairiePlusPigs/RuckmanMeta documentation built on May 8, 2019, 1:26 p.m.