Chapter 10 - Surveys and Official Statistics

Question 1

Non-Singular Question

Answer 1

When you ask more than one thing with only one answer.

What do you think of x and y?

Question 2

2 Types of Survey Mode/Administration

Answer 2

Interviewer administered: face-to-face, telephone

Self-administered

Question 3

How could interviewer administered surveys affect results?

Answer 3

Honesty, social desirability, interviewer effect

Question 4

How could self-administered survey modes affect results?

Answer 4

Non-representative - not everyone has internet, mail, and/or language skills.
Doesn’t allow for clarification.

Question 5

Non-Response Bias

Answer 5

Some people are more likely to respond to certain (or all) survey modes than others.
Tends to skew towards overrepresentation of old people and females.

Question 6

Reactivity

Answer 6

The interviewer effect, basically. Reacting to people different based on gender and race.
Especially applies to face-to-face interviews.

Question 7

Survey Research and Causality

Answer 7

Survey research has a tough time dealing with spurious relationships.

Question 8

Cross-Sectional v Longitudinal

Answer 8

Cross-sectional: at one point in time

Longitudinal: as times goes on

Question 9

Pros of Longitudinal Studies

Answer 9

Can increase internal validity
Can look at change over time
Can better understand temporal order

Question 10

Attrition Problem

Answer 10

Longitudinal studies suffer from this because subjects can die, move, change phone numbers, etc.

Question 11

Experimental Surveys, Cross-Sectional Surveys, and Longitudinal Studies as they all relate to confounding variables and temporal order

Answer 11

Experimental surveys: 
- good for temporal order
- good for confounding
Cross Sectional Surveys 
- bad for temporal order
- bad for confounding
Longitudinal:
- good for temporal order
- not good for confounding

Question 12

4 Types of Close-Ended Questions

Answer 12

Scales (ordinal)
Binary (nominal)
Feeling thermometer
Forced choice

Question 13

1 Pro and 1 Con for Open-Ended Questions

Answer 13

Pro: more information and their own words
Con: more time consuming for both parties

Question 14

7 Criteria for Assessing Survey Questions / Strategies to Decrease Measurement Error

Answer 14

Single-barreled questions
Balanced response options
Exhaustive Response Categories 
Mutually exclusive categories
Non-response option
Neutral language
Clear language

Question 15

Balanced Response Options

Answer 15

Making sure there is an equal amount of positive and negative answers

Question 16

Assessing Question Properties. Also, why would we want to?

Answer 16

To guess to what extent the interviewer effect could be affecting things.

Is your question (non)-sensitive?
Is your question (non)-factual
Is your question open or close-ended
Is your question easy or difficult to answer?

Question 17

Response Instability

Answer 17

People have contradicting opinions so they will have inconsistent answers.

Question 18

Two Issues with Question Order

Answer 18

We don’t see the order often, and early questions could be priming for you to answer a later one in a certain way.

Question 19

5 Ways to Critically Read Survey Research

Answer 19

Margin of Error
Authors/sources
Other literature
Equivalence in measures
Appropriate for survey research

Question 20

Official Statistics

Answer 20

Elections Canada, OECD, etc.

Administrative data

Question 21

Secondary Data

Answer 21

Made available to the public. Not OC.

Question 22

Pros and Cons of Secondary Data and Official Statistics

Answer 22

Pros: good quality because of high budgets, identifies subgroups, breadth, cost-saving to not do your own, longitudinal often.
Cons: data often aggregated - can’t tie to demographic factors, not tailored to your research.

Question 23

Challenges with International Data

Answer 23

• Less trustworthy
• challenges with reporting illegal activities
• fraudulent data
• equivalencies

Question 24

Ecological Fallacy

Answer 24

The assumption that group-level patterns imply individual level patterns

Question 25

Ethics with Data

Answer 25

Make sure to read technical information. For example, policy based on an exploding Metis population would need to know that it’s self-identification that’s causing that number to change not just reproduction.