Chapter 1 - Data Visualisation

Question 1

Remove observations (rows) with missing values

Answer 1

DF = DF %>% tidyr::drop_na()

Question 2

Dpylr::Glimpse()

Str()

Answer 2

Inspect structure of objects

Question 3

Matrix[1,2]

Answer 3

Access element in 1st row, 2nd column

Question 4

Matrix[1:3, 2:4]

Answer 4

Returns matrix with data from rows 1, 2, 3 and columns 2, 3, 4

Question 5

Matrix[ ,1]

Answer 5

All elements of the 1st column

Question 6

Literary digest poll (1936)

Answer 6

Roosevelt won a high no. Of votes fo US presidential election despite predicted victory for Landon.

10 million people were polled via mail survey of which 2.4 responded back.

• Americans suffered from Great Depression then

Question 7

Survey sampling

Answer 7

• selecting subset of observations from an entire population to draw conclusions about the whole population
• consider design, maximise profit, high reliability, optimisation with less resources
• type of survey design that’s appropriate for the study

Question 8

Selection bias

Answer 8

Taking a large sample thinking it would make a difference (repetition of error on a large scale)

• inaccurate representation of the population
• getting opinion based on certain qs asked e.g should a doctor be allowed to murder unborn kids who can’t defend themselves?

Question 9

Measurement bias

Answer 9

• recall bias
• misinterpret questions
• sensitive questions
• wording of question
• sampling method influences data obtained

Question 10

Non-response bias

Answer 10

• certain groups are under-represented due to lack of participation

E.g those that choose to not respond/ not participate in research experiment, those that don’t want to give a tip/fill out ‘customer satisfaction’ after eating at restaurant

Question 11

Randomised controlled double blind trials

Answer 11

Allocates subjects into a treatment + control group randomly

• neither subjects nor investigators know who’s in which group
• investigator compares responses of the 2 groups and if there is a difference in response it’s likely due to be caused by the treatment

Question 12

Observational studies

Answer 12

• doesn’t establish causation, only establishes association
• used in educational research
• if had confounding variable can lead to Simpson’s Paradox

Question 13

Confounding variables

Answer 13

When both treatment + control group differ by some 3rd variable that influences the response that is studied

• e.g if not all subjects keep taking the treatment/placebo, we can get confounding of adherers/non-adherers