STAT MOD 1: Chapter 1

Question 1

What is population?

Answer 1

the entire group of units that we want information about

Question 2

What is a sample?

Answer 2

a part or subset of population that is selected for study and observed to gather information
- Usually has a sampling design behind it
- (SRS, stratification, cluster, systemic, voluntary, convenience)

Question 3

What is a census?

Answer 3

collects information from all units in the population

Conducting a census might not be reasonable:
- measurements require destroying the item
- difficult to find entire population
- limited resources

Question 4

What is a statistic?

Answer 4

a number or numeric value that describes a SAMPLE
- value computed based on data from the sample
- using our statistic to learn about population

Question 5

What is population characteristic/parameter?

Answer 5

a number that describes the POPULATION
- did they use the population to get that number?
- a value computed based on data from the entire population (based on census)

Question 6

When can sample data be used to make inferences/conclusions or generalize about a larger group?

Answer 6

if the data can be considered to be representative with regard to the questions of interest

Question 7

What are the two main methods of collecting data?

Answer 7

1) observational study
2) experiment

Question 8

What is an observational study?

Answer 8

in which researcher observes characteristics of a sample

• DO NOT determine the groups or influence anything
• Participants tells the researcher, and researcher only records
• COLLECT data about relationship between two variables

Question 9

What is an experiment?

Answer 9

in which researcher tries to determine how a response behaves under different experimental conditions

• Different experimental treatments and how the response varies with each treatment
• MANIPULATES and IMPOSES experimental groups on the subjcts
• DETERMINE causal relationship between two variables

Question 10

What do conclusions drawn from a statistical study depend on?

Answer 10

the way in which the data are collected:

1) observational studies can show association, not causation
2) causal relationships need a well-designed experiment
- other variables might impact the response variable, so must use randomized design

3) Group studied and recruitment method can affect to whom results of the study generalize

Question 11

What is the key difference between observational study and experiment?

Answer 11

researchers IMPOSE experimental conditions on the subjects in an experiment while

researchers RECORD, do not determine who will be in what groups in an observational study

Question 12

What is random selection?

Answer 12

good sampling plan in an observational study (think SRS)

Question 13

What is random assignment?

Answer 13

set of subjects for an experiment are randomly assigned to treatment groups

Question 14

How do you evaluate whether conclusions drawn from the study are appropriate (given study description)?

Answer 14

• whether the study was observational or experimental for conclusions
• consider presence of other variables (confounding, lurking)
• the group studied and recruitment method to whom results can be generalized

Question 15

What is a simple random sample?

Answer 15

designed so that every group of n (you and another in the group) has the same probability of being selected

Question 16

What are the procedures for drawing a simple random sample?

Answer 16

Requires a list of units in the population and source of random numbers

Ex: drawing names from hat, random number generator, rand command or randint

Question 17

What are characteristics of simple random sample?

Answer 17

All units have:
- same probability of being selected
- same probability of being selected with every other person

As opposed to systematic sampling, the starting point is random but a unit would never make it into the same sample as person above or below them due to systematic nature

Question 18

Evaluating design of an observational study:

What is simple random sampling?

Answer 18

every group of n has the same probability of being selected
- every unit has a chance of being selected and with every other unit

Question 19

Evaluating design of observational study:

What is stratified random sampling?

Answer 19

population is first divided into subgroups (strata) then separate SRS are selected from each subgroup

• Divide population into strata then randomly select from each strata

Question 20

When do you use stratified random sampling?

Answer 20

Used when you know something about the units ahead of time (ex: grade level, homeroom) and ensure a more representative sample

Question 21

Evaluating design of observational study:

What is cluster sampling?

Answer 21

Divide population into groups of units that are naturally clustered together (clusters) then randomly select clusters and include all individuals in the into the sample

Ex: homerooms are clusters—randomly select two homerooms then include all students within the homeroom into the sample

Question 22

What is the difference between stratified random sampling and cluster sampling?

Answer 22

Stratification
- divide population into strata then RANDOMLY select from the strata (not all individuals in the strata are included)

Cluster
- divide population into natural clusters then randomly select the clusters themselves (include all individuals of a cluster in the sample)

Question 23

What is systematic sampling?

Answer 23

choose a starting point at random from the first k individuals then every kth individual after that

Ex: first randomly select a faculty from the first 20 faculty members (k = 20) then every 20th after that

Question 24

What are some good methods of sampling?

Answer 24

1) SRS
2) Stratified random sampling
3) Cluster sampling
4) Systematic sampling

Question 25

What are biased sampling methods?

Answer 25

1) Convenience sampling 2) voluntary sampling

Question 26

What is convenience sample?

Answer 26

selecting individuals or objects that are easy or convenient to sample - First ones encountered, most approachable-looking, etc.

Question 27

What is voluntary response sample? What is its con?

Answer 27

type of convenience sampling that relies on individuals volunteering to be part of the study Con: volunteers often hold strong opinions–might not be representative of the population

Question 28

Why is random selection an important component of a sampling plan

Answer 28

Because it helps be representative of the population and reduces the potential of excluding parts of the population, as opposed to convenience or voluntary response

Question 29

Why are volunteer response samples unlikely to produce reliable information about a population?

Answer 29

Volunteer response samples have volunteers who hold strong opinions, so might not be representative of the population

Question 30

Why are convenience samples unlikely to produce reliable information about a population?

Answer 30

Convenience samples only includes people who are easy to access, so might exclude parts of the population

Question 31

What is a sampling error?

Answer 31

error that comes from drawing a sample instead of taking a census

Question 32

What are the three sources of bias in a survey?

Answer 32

1) Selection (how you select) 2) Measurement/response (people lie!) 3) Nonresponse (no answer)

Question 33

What is selection bias?

Answer 33

when the method in which sample is selected systematically excludes a part of the population - Also occurs if volunteers are used in a study Ex: phone survey excludes those without access to a phone

Question 34

What is measurement (response) bias?

Answer 34

when the method of observation produces values that differ from true value in some way - You sampled somebody and collected a variable, but the value was wrong Ex: lying about a variable, have leading questions that influence a response

Question 35

What is nonresponse bias?

Answer 35

when responses are not obtained from all individuals selected for inclusion in the sample - Selected a unit but the unit did not respond - The people who did not respond differ in a systematic way from those that did respond Ex: giving a survey when there are kids absent

Question 36

Evaluating design of comparative experiment: What is completely randomized design?

Answer 36

Units are randomly assigned to treatments Subjects → randomly assigned to groups → measure response

Question 37

Evaluating design of comparative experiment: What is blocking or block design?

Answer 37

divides the units into blocks of similar units then randomly assigns the individual within each block to treatments Subjects → blocking into girls or boys → randomly assigned to groups or mini experiment for the blocks → compare overall or within blocks

Question 38

Evaluating design of comparative experiment: What is matched pairs designed?

Answer 38

type of block design in which each pair is its own block - One unit or pair of units in a block (ex: twins, pretest/post-test), applying all conditions to one unit across time Ex: twins or doing a before-and-after with the same subjects

Question 39

What is a hint term for matched pairs design?

Answer 39

fractional factorial experiment

Question 40

What are the limitations of using volunteers as subjects in an experiment?

Answer 40

Volunteers might differ from the population in some way, so limits ability to generalize to population (selection bias)

Question 41

What is a control group?

Answer 41

a baseline of no-treatment or the standard treatment group - Baseline against which the treatment groups can be compared to determine if the treatment had an effect - Tells how the response behaves in the absence of treatment (to see if treatment had an effect)

Question 42

What is a placebo group?

Answer 42

something identical to the treatment, except it does not contain active ingredients - reduce psychosomatic effects that might influence the response (eliminates confounding variables)

Question 43

What is blinding?

Answer 43

best if subject does not know which treatment they’re receiving (single blind) and if both subject and evaluator do not know which treatment they are receiving (double bind) - prevents placebo effect and bias from evaluator

Question 44

Why is random assignment important when collecting data in an experiment?

Answer 44

- Important to split or spread variables that might impact the response evenly across groups (produce chance-like variability) - Isolate treatment effects

Question 45

What are the procedures for randomly assigning experimental units to treatment groups?

Answer 45

Put them into a list, flip a coin for two groups, random number generator