Import the Data
df <- read_xlsx(
"data/Blastexposure.xlsx"
)Summarize the data set to get an idea of the variables and their values:
summary(df)## PartNum Rank HighRank Ref
## Min. : 1.0 Length:1008 Length:1008 Length:1008
## 1st Qu.:215.8 Class :character Class :character Class :character
## Median :425.0 Mode :character Mode :character Mode :character
## Mean :434.8
## 3rd Qu.:649.2
## Max. :892.0
##
## NumBlasts DiagA
## Min. :0.000 Length:1008
## 1st Qu.:1.000 Class :character
## Median :2.000 Mode :character
## Mean :1.637
## 3rd Qu.:2.000
## Max. :2.000
## NA's :309That didn’t tell me much. I should figure out what’s in DiagA:
df %>%
pull(DiagA) %>%
unique() %>%
sort()## [1] "OTHER" "PTSD"Seems clear enough!
Some Tables
OK, let’s see if there is a relationship between exposure to blasts and PTSD:
tab2 <- xtabs(~ NumBlasts + DiagA, data = df)
tab2## DiagA
## NumBlasts OTHER PTSD
## 0 41 15
## 1 47 69
## 2 108 331Holy moly! Looks like the more blasts you get, the greater the likelihood of PTSD.
Let’s rename and recode a bit:
df2 <-
df %>%
rename(diagnosis = DiagA) %>%
mutate(blast_exposure = case_when(
NumBlasts == 0 ~ "0 blasts",
NumBlasts == 1 ~ "1 blast",
NumBlasts == 2 ~ "2+ blasts"
)) %>%
mutate(blast_exposure = factor(
blast_exposure,
levels = c("0 blasts", "1 blast", "2+ blasts")
))Another table:
df2 %>%
drop_na(blast_exposure, diagnosis) %>%
group_by(blast_exposure) %>%
summarize(
PTSD = sum(diagnosis == "PTSD"),
total = n(),
percent = PTSD / total
)## # A tibble: 3 x 4
## blast_exposure PTSD total percent
## * <fct> <int> <int> <dbl>
## 1 0 blasts 15 56 0.268
## 2 1 blast 69 116 0.595
## 3 2+ blasts 331 439 0.754Some Graphs
Go for a bar graphs:
df2 %>%
xtabs(~ blast_exposure + diagnosis, data = .) %>%
mosaicplot(
main = "Mosic Plot of Diagnosi vs. Blast Exposure",
xlab = "Number of blasts",
ylab = "Diagnosis"
)
nice caption
Some Inferential Statistics
A \(\chi^2\)-test is appropriate here:
chisqtestGC(
~ blast_exposure + diagnosis,
data = df2
)## Pearson's Chi-squared test
##
## Observed Counts:
## diagnosis
## blast_exposure OTHER PTSD
## 0 blasts 41 15
## 1 blast 47 69
## 2+ blasts 108 331
##
## Counts Expected by Null:
## diagnosis
## blast_exposure OTHER PTSD
## 0 blasts 17.96 38.04
## 1 blast 37.21 78.79
## 2+ blasts 140.82 298.18
##
## Contributions to the chi-square statistic:
## diagnosis
## blast_exposure OTHER PTSD
## 0 blasts 29.54 13.95
## 1 blast 2.58 1.22
## 2+ blasts 7.65 3.61
##
##
## Chi-Square Statistic = 58.5475
## Degrees of Freedom of the table = 2
## P-Value = 0We have overwhelmingly strong evidence that folks exposed to more blasts are more likely to be diagnosed with PTSD.