Statistics Assignment 1
Name: Hugh Richardson
ID Number: 24002972
Use the following R code to load and examine the structure of the dataset.
lambs <- read.csv("C:/Users/Naomi/Desktop/LambBehavour.csv", header=TRUE)
str(lambs)
## 'data.frame': 355 obs. of 6 variables:
## $ lamb_tag : int 321 322 315 316 559 162 163 405 406 336 ...
## $ ewe_tag : int 1 1 2 2 3 4 4 9 9 10 ...
## $ Lamb_sex : chr "Ewe" "Ewe" "Ram" "Ewe" ...
## $ time_to_contact: int 18 41 21 16 14 18 18 23 23 16 ...
## $ time_to_follow : int 108 107 110 110 295 60 60 131 131 258 ...
## $ contact : chr "Both moved" "Both moved" "Ewe to lamb" "Ewe to lamb" ...
EXERCISE 1: Data Summary [total: 10 Marks]
lamb_tag=numerical; Lamb_sex=character; time_to_follow=numerical; contact=character
table(lambs$Lamb_sex)
##
## Ewe Ram
## 180 175
mean=38.34085; median=25; standard deviation=39.266
mean(lambs$time_to_contact)
## [1] 38.34085
median(lambs$time_to_contact)
## [1] 25
sd(lambs$time_to_contact)
## [1] 39.266
hist(lambs$time_to_contact,
main="Histogram of 'time to contact'",
xlab="time to contact")
boxplot(lambs$time_to_contact,
main="Boxplot of 'time to contact'",
ylab="time to contact")
EXERCISE 2: Sex and time to contact [total: 13 Marks]
The distributions of both sexs are similar, with a right skew on time to contact. There is a slightly larger range in the “ewe” group.
boxplot(time_to_contact ~ Lamb_sex,
data = lambs,
main = "Time to Contact by Lamb Sex",
xlab = "Lamb Sex",
ylab = "Time to Contact")
Ram=37.289; Ewe=41.12225
sd(lambs$time_to_contact[lambs$Lamb_sex=="Ram"])
## [1] 37.289
sd(lambs$time_to_contact[lambs$Lamb_sex=="Ewe"])
## [1] 41.12225
Yes, the two population standard deviations could be considered equal. The ratio between the two standard deviations is 1.102798. This ratio is <2.0, so equal variances may be assumed for testing of the sample means.
sd(lambs$time_to_contact[lambs$Lamb_sex=="Ewe"])/
sd(lambs$time_to_contact[lambs$Lamb_sex=="Ram"])
## [1] 1.102798
Null hypothesis: There is no difference in the true mean time_to_contact between male and female lambs. Alternative hypothesis: There is a difference in the true mean time_to_contact between male and female lambs. * What is the p-value?
t.test(time_to_contact ~ Lamb_sex, data=lambs, var.equal=T, conf.level=0.95)
##
## Two Sample t-test
##
## data: time_to_contact by Lamb_sex
## t = 0.86931, df = 353, p-value = 0.3853
## alternative hypothesis: true difference in means between group Ewe and group Ram is not equal to 0
## 95 percent confidence interval:
## -4.576047 11.825888
## sample estimates:
## mean in group Ewe mean in group Ram
## 40.12778 36.50286
p-value=0.3853 * What is your decision? with justification.
A p-value of 0.3853 is insufficient to reject the null hypothesis.
The statistical test result suggests there is no evidence for a difference in the average time_to_contact of male and female lambs. The difference observed in mean time_to_contact between male and female lambs is likely a result of sampling from the population and does not reflect true differences in the population.
The 95% confidence interval has a range that contains “0”. This means that the 95% confidence of the true difference between both groups may be a 0 value, which aligns with the null hypothesis.
Assumption 1 - Sample mean is approximately normally distributed. Assumption 2 - Sample is representative of the population of interest. The sample size was sufficiently large (n>30),meaning the central limit theorem holds and the sampling distribution of the mean may be considered to be approximately normal. We may consider the second assumption to be valid given that each data point was sampled randomly from the population of interest.
EXERCISE 3: time_to_contact, time_to_follow and contact [total: 5 Marks]
In this exercise you will explore the relationship between time_to_contact, time_to_follow and contact.
plot(time_to_contact~time_to_follow, data=lambs,pch=5)
points(lambs$time_to_follow[lambs$contact == "Both moved"],
lambs$time_to_contact[lambs$contact == "Both moved"],
col = "blue", pch = 5)
points(lambs$time_to_follow[lambs$contact == "Lamb to ewe"],
lambs$time_to_contact[lambs$contact == "Lamb to ewe"],
col = "red", pch = 5)
points(lambs$time_to_follow[lambs$contact == "Ewe to lamb"],
lambs$time_to_contact[lambs$contact == "Ewe to lamb"],
col = "green", pch = 5)
legend(
"topleft",
legend=c("Both moved","Lamb to ewe", "Ewe to lamb"),
col=c("blue","red","green"),
pch=c(5,5,5),
title="contact")
title("scatter plot of 'time to contact' and 'time to follow' in lambs ")
time_to_follow has a positive correlation with time_to_contact. This positive relationship is similarly observed when broken down into the different “contact” variables. However, the strength of the correlation between time_to_contact and time_to_follow is quite weak (0.43). When looking at this correlation at the different levels of “contact”, the “Ewe to lamb” variable has the weakest positive correlation of 0.40. This increases for “Both moved”, which has a correlation coefficient of 0.68, but is highest in “Lamb to ewe”, where the correlation coefficient is 0.81.
cor(
lambs$time_to_contact,
lambs$time_to_follow)
## [1] 0.4344611
cor(
lambs$time_to_contact[lambs$contact=="Both moved"],
lambs$time_to_follow[lambs$contact=="Both moved"]
)
## [1] 0.6767942
cor(
lambs$time_to_contact[lambs$contact=="Ewe to lamb"],
lambs$time_to_follow[lambs$contact=="Ewe to lamb"]
)
## [1] 0.3961282
cor(
lambs$time_to_contact[lambs$contact=="Lamb to ewe"],
lambs$time_to_follow[lambs$contact=="Lamb to ewe"]
)
## [1] 0.8058358
EXERCISE 4: Sex of the lamb and direction of contact [total: 8 Marks]
my.table1 <- table(lambs$Lamb_sex,lambs$contact)
my.table1
##
## Both moved Ewe to lamb Lamb to ewe
## Ewe 40 132 8
## Ram 42 123 10
barplot(my.table1,beside=TRUE)
?barplot
## starting httpd help server ... done
The table and barplot suggest that there is no effect of the sex of the lamb on the direction of contact as there appears to be negligible differences between the sex at all levels of “contact”.
my.ChiSq1 <- chisq.test(lambs$Lamb_sex,lambs$contact)
my.ChiSq1
##
## Pearson's Chi-squared test
##
## data: lambs$Lamb_sex and lambs$contact
## X-squared = 0.51833, df = 2, p-value = 0.7717
Null hypothesis: direction of contact is independent of the sex of the lamb Alternative hypothesis: direction of contact is dependent on sex of the lamb
0.7717
## [1] 0.7717
There is insufficient evidence to reject the null hypothesis, given a p-value>0.05 (0.7717).
There is no impact of sex of lamb on the direction of contact. As a result, sex of the lamb need not be a consideration when evaluating the direction of contact exhibited by lambs.
my.ChiSq1$expected
## lambs$contact
## lambs$Lamb_sex Both moved Ewe to lamb Lamb to ewe
## Ewe 41.57746 129.2958 9.126761
## Ram 40.42254 125.7042 8.873239
We can assume the first assumption to be met, as the lowest expected value is 8.9.