Website Data

Website Data R Codes & Analyses

# 1) Load
path <- "C:/Users/skyep/OneDrive/Desktop/websitedata.csv"
dat  <- read.csv(path, stringsAsFactors = FALSE)

# 2) Pull weighted columns
mission <- dat$wsubmission
ebp     <- dat$wsubebp
access  <- dat$wsubaccess
fs      <- dat$wsubfam

# 3) Normalize to 0–1 (edit maxima if custom weights)
mission_max <- 1 * 4
ebp_max     <- 2 * 4
access_max  <- 3 * 4
fs_max      <- 1 * 4

scale01 <- function(x, M) pmin(pmax(x / M, 0), 1)
mission <- scale01(mission, mission_max)
ebp     <- scale01(ebp,     ebp_max)
access  <- scale01(access,  access_max)
fs      <- scale01(fs,      fs_max)

# 4) Paired t-test: EBP > Mission
i1 <- complete.cases(ebp, mission)
tt1 <- t.test(ebp[i1], mission[i1], paired = TRUE, alternative = "greater")
dz1 <- as.numeric(tt1$statistic) / sqrt(sum(i1))
cat(sprintf("EBP > Mission: t(%d)=%.3f, p=%.4f, dz=%.3f\n",
            tt1$parameter, tt1$statistic, tt1$p.value, dz1))

EBP > Mission: t(99)=-13.713, p=1.0000, dz=-1.371

# 5) Paired t-test: Family Support > Accessibility
i2 <- complete.cases(fs, access)
tt2 <- t.test(fs[i2], access[i2], paired = TRUE, alternative = "greater")
dz2 <- as.numeric(tt2$statistic) / sqrt(sum(i2))
cat(sprintf("FS > Access:  t(%d)=%.3f, p=%.4f, dz=%.3f\n",
            tt2$parameter, tt2$statistic, tt2$p.value, dz2))

FS > Access:  t(99)=7.986, p=0.0000, dz=0.799

# 6) Proportion test: >= 50% recommended
# Use existing binary if present; else rule = overall mean >= cutoff
cutoff <- 0.70
rec_col <- intersect(c("recommended","is_recommended","recommend_flag","meets_recommendation"),
                     names(dat))[1]
if (!is.na(rec_col)) {
  rec <- as.logical(dat[[rec_col]])
} else {
  overall <- rowMeans(cbind(mission, ebp, access, fs), na.rm = TRUE)
  rec <- overall >= cutoff
}
rec <- rec[!is.na(rec)]
k <- sum(rec); n <- length(rec)

pt <- binom.test(k, n, p = 0.50, alternative = "greater")
cat(sprintf("Recommended: %d/%d (%.1f%%), one-sided exact p=%.4f\n",
            k, n, 100*k/n, pt$p.value))

Recommended: 0/100 (0.0%), one-sided exact p=1.0000

EBP > Mission: t(99)=-13.713, p=1.0000, dz=-1.371

• The negative t means that mission scores are actually higher than EBP scores, not the other way around.

• The one-sided test was set up as “EBP > Mission,” but the data went in the opposite direction.

• p = 1.000 means there is no statistical support for your hypothesis

• Cohen’s dz = -1.37 → a huge effect size (mission much higher than EBP).

FS > Access: t(99)=7.986, p=0.0000, dz=0.799

• Family support scores are significantly higher than digital accessibility scores.

• Very strong evidence (p < 0.001).

• Cohen’s dz = 0.80 → a large effect size.

Recommended: 0/100 (0.0%), one-sided exact p=1.0000

• (≥0.70 average across categories), none of the websites met the recommendation threshold.

• The test asks if at least half (≥50%) of sites are recommended. With 0%, the result is the opposite: evidence that the true proportion is much lower than 50%.

• The one-sided p = 1.0 simply reflects that there is no support for the hypothesis “≥50%.”