STA1010

Question 1

Explanatory Variable

Answer 1

may explain or cause a change in another variable, may
be manipulated or set at a value

Question 2

Response Variable

Answer 2

the variable measured to see if it changes in response to
another variable

Question 3

Confounding Variable

Answer 3

A variable that is thought/known to influence both the
explanatory and response and so confuses the
interpretation of any relationship between them.

Question 4

Lurking Variable

Answer 4

As confounding but NOT known beforehand - used during analysis/after experiment

Question 5

Simpson’s Paradox

Answer 5

Rare cases of misleading information
e.g. consider heterogeneity

Question 6

Observational Studies - key features

Answer 6

• No manipulation of the factors under investigation
• No random assignment of units to any specific treatment
• Can be very informative and is the only way possible if:
• Ethical considerations preclude manipulation
• Want to see what happens in a natural settings, without contrived involvement.

Question 7

Observational Studies - Cautions

Answer 7

Causation is difficult to establish - too many uncontrolled confounding, lurking
variables
* Special type of observational study : case-control study
* Often seen in medical studies where cases of a disease are compared with
others who are known not to have the disease (control group)

Question 8

Observational Study Schematic Layout

Answer 8

(ADD PICTURE)

Question 9

Experiment

Answer 9

• Do have active imposition of a treatment level on the subject -
  different values of explanatory can be set or controlled
• Do randomly assign units to a specific treatment group
• Measure the responses for each treatment group.
• “Cause and effect link” more likely to be established

Question 10

Experiment - general schematic layout

Answer 10

(ADD PICTURE)
MAY BE ASKED TO DRAW IN EXAM!!!!

Question 11

Experiment - major features

Answer 11

CONTROL - “absence”, placebo, blinding
* RANDOMISATION → all treatment groups have similar background,
minimises bias
* REPLICATION - sample size sufficient, each treatment group has >1
unit, minimises random error
* BLOCKING - according to a confounding variable to “control”

Question 12

BLOCKED Random Experiment

Answer 12

ADD PICTURE

Question 13

Factorial Experiments

Answer 13

Vary more than 1 factor at a time – time and cost considerations
* Can see the interactions between different explanatory variables

Question 14

Repeated-measures designs = before + after

Answer 14

blocks = individuals , and
* units = repeated time periods in which receive varying treatments.

Question 15

Matched-Pair Designs

Answer 15

either two matched individuals or same individual receives each of two
treatments.
* Important to randomize order of two treatments and use blinding if possible.

Question 16

Types of Study

Answer 16

Observational Study
* Experiment

Question 16

Types of Variables

Answer 16

• Explanatory,
• Response,
• Confounding (lurking)

Question 17

Design Features of Experiments

Answer 17

Control (e.g. Placebo),
* Randomisation,
* Replication,
* Blocking

Question 18

Define population

Answer 18

→ parameters
the larger group of units about which inferences are to be made

Question 19

Define sample + what they calculate

Answer 19

→ statistics
the smaller group of units actually measured

Question 20

List the advantages of a Sample survey over a census

Answer 20

• Sometimes census isn’t possible, when measurements destroy units
• Speed, especially is population is large
• Accuracy, devote resources to getting accurate sample results
• Cost, less costly + less time then census

Question 21

Define population

Question 22

Define units

Question 23

Define sample

Question 24

Describe the sampling design process stages: