sTeamTraen

library(rpsychi)

# Function to display the possible ranges of the F or t statistic from a one- or two-way ANOVA.
f_range <- function (m, s, n, title=FALSE, show.t=FALSE, dp.p=-1, labels=c()) {
  m.ok <- m
  if (class(m.ok) == "matrix") {
    func <- ind.twoway.second
    useF <- c(3, 2, 4)
    default_labels <- c("col F", "row F", "inter F")
  }

sTeamTraen

# Find words (in a dictionary file) that are made up of other words run together.
#  e.g.: "deliberating" = "deli" + "berating"
#  e.g.: "categorising" = "cat" + "ego" + "rising"
# By Nick Brown (nicholasjlbrown@gmail.com), June 2025.

# Some parameters. I hope their meaning is reasonably obvious.
min.split.words <- 2
max.split.words <- 3
shortest.single.word <- 3
longest.single.word <- 8

sTeamTraen

# Simulation of BMI values from Figure 1 of Rizk et al. 10.1016/j.fertnstert.2004.07.960

exact.rnorm <- function(n, m, sd) {
  x <- rnorm(n)
  x <- (x - mean(x)) / sd(x)        # standardize to mean 0, sd 1
  x <- x * sd + m                   # rescale to target mean and sd
  return(x)
}

# Mean and SD of height and weight from Table 1

sTeamTraen

# Compare multiple correlations to see if they are similar.
# See http://home.ubalt.edu/ntsbarsh/business-stat/opre504.htm#rmulticorr
# By Nick Brown (nicholasjlbrown@gmail.com), September 2022. Licence: CC-0.

r_to_z <- function (r) {
  (log(1 + r) - log(1 - r)) / 2
}

many_corr <- function (vec.z, vec.n) {
  tot1 <- 0

sTeamTraen

# Demonstration of chi-square distribution from chi-square tests under the null
# By Nick Brown, February 2022.
# Licence: CC-0

set.seed(2)

df.range <- 1:6      # range of DFs for which to generate plots
loops <- 10000       # number of cases to generate (more gives better precision)
N <- 5000            # range of numbers in tables

sTeamTraen

# Matta et al., 10.1001/jamainternmed.2021.6454

# See Altman, 2011, 10.1136/bmj.d2304 for calculation of z-scores
# See https://en.wikipedia.org/wiki/Fisher%27s_method#Relation_to_Stouffer's_Z-score_method for Stouffer's method

eTable7 <- data.frame(name="", N=0, OR.m6=0.0, CIlo.m6=0.00, CIhi.m6=0.00, OR.m7=0, CIlo.m7=0.00, CIhi.m7=0.00)
eTable7 <- rbind(eTable7, c("Sleep problems", 54, 0.76, 0.42, 1.36, 0.92, 0.51, 1.67))
eTable7 <- rbind(eTable7, c("Joint pain", 35, 1.38, 0.69, 2.73, 1.40, 0.64, 3.03))
eTable7 <- rbind(eTable7, c("Back pain", 38, 1.58, 0.80, 3.09, 1.24, 0.60, 2.55))
eTable7 <- rbind(eTable7, c("Digestive track problems", 29, 0.61, 0.27, 1.35, 1.16, 0.50, 2.49))

sTeamTraen

# Matta et al., 10.1001/jamainternmed.2021.6454

# Table 2

n.s0b0 <- 25271
n.s0b1 <- 461
n.s1b0 <- 638
n.s1b1 <- 453

symptoms <- list(

sTeamTraen

# Code to analyse Table 1 from some articles about the use of anti-androgens to treat Covid-19.
# By Nick Brown, October 2021.
# Licence: CC-0.
# See https://steamtraen.blogspot.com/2021/10/applying-john-carlisles-table-1.html

# Compute the chi-square p value for a given analysis modality (see the blog post).
mychi <- function (s1, s2, n1, n2, analysis)
{
  mat <- matrix(c(s1, (n1 - s1), s2, (n2 - s2)), ncol=2)

sTeamTraen

# Simple demonstration of mean absolute correlation with random data.
# By Nick Brown, February 2020. License: CC-0.

library(ggplot2)

set.seed(1)
iter <- 1000

# Create data frame to contain correlation results, excluding CI (which is a list)
ct.names <- names(cor.test(1:3, 3:1))    # little hack: correlation is perfect, so no CI

sTeamTraen

# Reanalysis of Laird et al. (2020) https://pubmed.ncbi.nlm.nih.gov/32603576/
# By Nick Brown, January 2021.
# Licence: CC-0.

# Laird et al. did not state how they went from four (Italy) or three (Spain) measurements
#  of Vitamin D levels to a single figure.
# When I calculate the unweighted mean of their samples, this seems to correspond to the
#  Y-axis values in their figure 1, but I have included the relevant numbers from their
#  Table 1 here to allow a weighted average to be calculated.