Chapter 3

Question 1

Sample

Answer 1

the data you
collectthe data you
collect

Question 2

Population

Answer 2

the true

underlying situation

Question 3

Standard statistical methods assume that

Answer 3

the

population is much larger than the sample

Question 4

We usually assume
that the data you
collect is __________ from the
population.

Answer 4

randomly sampled

Question 5

But values (e.g. mean)
computed from a
sample can differ
from…

Answer 5

the true population values

Question 6

Random sampling error

Answer 6

by chance, a random
sample can have different values from the true pop.
-no one is ‘at fault’.

Question 7

Bias is not

Answer 7

random but
produces errors
in a consistent
direction.

Question 8

Selection bias:

Answer 8

the way the sample is collected is
not random. Subjects with particular values are
preferentially sampled.

Question 9

In statistics, ‘sampling error’ does not usually

refer to

Answer 9

sloppy or incorrect measurements

Question 10

Statistics usually
assumes that
sampling error

Answer 10

does not exist

Question 11

If the measurements are just sloppy, then…

Answer 11

patterns

might be missed and/or variation over-estimated.

Question 12

If the measurements are systematically

Answer 12

too low or

too high, then this is an additional source of bias.

Question 13

New perspective

Model

Answer 13

mathematical
description of a
simplified view of
the world.
- Consists of general
description and
parameters.

Question 14

Example New perspective

Model:

Answer 14

a Gaussian bell-shaped distribution is a
model with the parameters: mean and standard
deviation. Image shows effect of the parameters.

Question 15

Goals of Statistics

Answer 15

1. Make inferences
about the values
of parameters
for a model.
2. Compare
alternative
models.

Question 16

Pseudoreplication

Answer 16

a sampling design
that is inappropriate
for describing the
population of
interest.

Question 17

Pseudoreplication dont’s:

Answer 17

Do not take many samples from one individual person
and extrapolate to all people.
Do not treat hierarchical samples as independent
samples (e.g. 5 randomly chosen countries with 5
randomly chosen individuals from each. This is not the
same as 25 individuals chosen randomly worldwide.)