There are three basic types of interviews

1. Structured
2. Non structured
3. Semi structured
They depend on the questions they ask, and the way we word questions that will ultimately
decide the response the subject is giving us

Have to be careful wording questions as things can be inferred from early questions and
influence responses in the long term

Structured interviews
- Uses interview guide
- Researcher poses the same question to all interviewees and same wording for every
question asked to every person
- This enhances comparability
- When you interview 30 people and ask the same question, same order, same wording,
these responses are therefore comparable and are not affected by different ordering or
wording
- Questions are typically close ended, with predefined response categories
● Yes or no, completely agree or disagree

Semi structured interviews

- Researcher uses an interview guide but does not necessarily pose the same questions
to all interviewees using the same wording or in the same order
- Have medium to high comparability
- Efficient because of the improvisation of questions to interviewees
- They allow follow up questions and contain high quality of info because of the
improvisation of question wording
- Most scholars that conduct interviews, are semi structured interviews
- Political elites are interviewed under this genre

Unstructured interviews
- More common in participant observation in subjects such as anthropology
- They do not use an interview guide and improvise questions based on context
- Questions come up when you observe the subjects
- They have low comparability because it is not systematic
- Requires greater time investment
- However, they provide high quality information based on the rawness of an individuals
responses

Interviews are best used for hypothesis BUILDING, using INDUCTIVE reasoning
- You as the researcher can gain insight through interviews by getting information about
certain people and their beliefs
- Use all this to provide a better understanding of human building
 - Interviews allow us to learn new information that we do not know, such as in
unstructured and semi structured environments

Interviews are NOT USEFUL for hypothesis TESTING

- Because interviewees are not representative to the population drawn
- Individuals are usually referred and recruited through snowball processes
- Not randomly selected, not representative of the populations as a whole
- Responses are not comparable especially for semi structured and unstructured
interviews
- Data from interviews are largely subjective and open to interpretation

Interview 5 step process

1. Identify the target population, depending on the research question and the objective of
the interview. For example; if you are interested in how people in canada vote, then the
target population would be Canadian voters
2. Design the interview: think about the structure, format and wording/order of the
questions, think about the mode of the interview. Basically, setting up an interview guide
3. Recruit participants through random or non random sampling. Using interviews as a
qualitative tool may require more non random sampling to think about target populations
for the target subject.
4. Conduct the interview, create a trial run of the interview and gain consent of the person
being interviewed. Should you record the interview, or take notes on it? This stage is to
go through a trial run that will ultimately lead to the actual interview
5. Write the follow up and analysis for your interviewees by providing contact information to
participants. Tell them how you are going to keep the notes and recordings in a safe
place.

3 common interview modes

1. In person
- Gold standard
- Not always possible because of time or distance
- They produce elaborate and spontaneous responses, best used in unstructured
and semi structured environments because they can improvise and create a
good environment that allow the interviewer and interviewee to feed off
eachother
- They allow body language to be observed
- Allow for follow up questions, you can add questions about specific themes that
are commonly discussed

2. Phone interviews
- Also produce spontaneous responses
- When on video call, can observe body language
- Here, we can also ask follow up questions
 - Adequate for open ended questions and unstructured/semi-structured
environments
- This is a good fall back for in person interviews

3. Email interviews
- Adequate for factual information to be presented
- They encourage brief and guarded responses
- They are less costly to conduct, and are very convenient from interviewees
- They do not require a recording or transcription
- They are best used in structured interviews and have close ended questions
being asked

Analyzing interview data through INDUCTIVE or DEDUCTIVE approach

Inductive ( go from data to the theory )

- Researchers can look at transcripts of interviews to identify common themes or
categories in response to each question being asked
- They can also develop a typology of interviewees based on demographic or other
characteristics that they possess
- With this information, researchers can draw connections among respondents and their
responses

Deductive approach (before collecting data, devise a coding scheme (things you expect to find
in responses of the interviewees)
- As well , prior to collecting data, use analytic categories about the types of individuals
they expect to have differing views on an issue, then code the interviewees response
against these categories (this is a hypothesis testing tool)

Chapter 11

Focus Groups
- A semi structured group interview designed to gain information from participants on a
specific topic of interest
- Small groups of 6-12 people
- Has an interview guide, but the moderator provokes discussion among participants
- One can observe this discussion through responses
- Purpose is to gain heterogeneity or diversity of responses and see how initial
interventions can start other discussion
- Encourage people to develop the answers they provide to your questions
- Each focus groups has a difficult dynamic
Focus groups help HYPOTHESIS BUILDING
- This generates inductively
- Focus groups can provide a lot of insight to theories
- May look at data and think back about how things may work in ones cognitions, and then
derive hypothesis that you could better test through the data provided by focus groups

Focus groups DO NOT HELP HYPOTHESIS TESTING

- Participants are not fully representative of the general population
- Participants are not selected randomly
- Responses are not easily comparable because they are drawing from different people
who have different experiences and may interpret the questions in different ways
- Whatever is being said in the focus groups is open to discussion in a subjective manner
- Information that may be given to researchers are inaccurate because of sensitive issues

Focus groups produce greater openness through others, and save time and money
- They can produce more honest and complete responses compared to interviews
because of the comfortability of the environment in focus groups
- The group dynamic can also hide someone's true views of a subject because of the fact
that people hide their true opinions through others
- Focus groups can also create an environment where people feel more comfortable,
when they share the same experiences

5 Steps to Conducting Focus Groups

- Identify target population: depending on the research question and the objective of the
interview
- Design the focus group: the size, the setting, audio records
- Recruit participants: through random or non random sampling, see if they will receive
compensation, saturation point:
- Then conducting the focus group, however going through practice runs first, find a
moderator to hire, start with easy questions then move to specifics
- Analyzing a follow up: prvovide contact info to participants and keep notes, recordings in
a safe place

Analyzing
- Can either analyze inductively (from data to theory and ellaboraiton of hypothesis) or
deductively (hypothesis to data and empirical evaluation)
- Focus groups used to build hypothesis
- Focus group is a good place to observe dyanmics to eventually develop hypotheses

Interviews
- Are methods of collecting qualitative information from individuals through a series of
questions posed either face to face, by phone or by email
Structured Interviews
- The researcher uses an interview guide that poses the same questions to all
interviewees without changing the wording of questions and order of which they are
asked.

Interviews allow researchers to…

- Collect factual information and collect less observable information about people’s mental
process

Interview Guides
- Are written documents containing questions that researchers ask interviewees and in the
exact order they will ask them.
- They are used in structured and semi structured interviews

Close ended questions

- In structured interviews, these type of questions are asked
- They are questions with defined response categories, such as yes or no

Semi-structured interviews
- The researcher will use an interview guide, but does not necessarily pose the same
questions to all interviewees using the same wording and same order, has more
freedom

Unstructured interviews
- The researcher does not use an interview guide, and improvises questions based on the
context of discussion
- This occurs typically in participant observation where the researcher immerses
themselves in the research

Interviews are used for HYPOTHESIS BUILDING

Interviews are NOT used for HYPOTHESIS TESTING

The five step interview process

1. Identify target population
2. Design interview
3. Recruit participants
4. Conduct interview
5. Follow up analysis

Three common interviewing modes

1. In person
2. Email
 3. Phone

In person interviews
- Allow body language to be observed
- Are the gold standard
- Allows for follow up questions
- Work best in semi structured or unstructured environments

Phone interviews
- Allow for follow up questions and for body language to be observed if its via a video call
- They work best in unstructured or semi structured environments

Email interviews
- Work best for structured environments
- Close ended questions
- Are less costly and more convenient

Interview data can be analyzed

- Deductively or inductively

Inductive approach
- Data to hypothesis
- Researchers pursue transcripts to identify common themes or analyze categories in
response to each question
- Then developing a typology of interviewees based on demographics or characteristics
they possess

Deductively
- Hypothesis to data
- Devising a coding scheme prior to any interview about what they expect to find in
response to each question, then after interviewing, elaborate the responses against the
coding scheme

Focus groups
- A semi structured group interview designed to elicit information from participants on a
defined topic of interest
- Uses an interview guide but does not strictly adhere to the guide to allow the flow of
conversation and questioning

Focus groups are used for HYPOTHESIS BUILDING

Focus groups are NOT used for HYPOTHESIS TESTING

- Because participants are not random
 - Responses are not comparable
- Data is NOT open to interpretation

Conducting a focus group undergoes 5 steps

1. Identify the target population
2. Design the focus group
3. Recruit the participants
4. Conduct the focus group
5. Follow up and analyze

Analyzing focus group data either INDUCTIVELY or DEDUCTIVELY

Quantitative Measures
- Measures are quantitative representations of concepts used as a basis or standard of
comparison: include ordinal, nominal, ratio and interval

Nominal measures
- Are discrete measures in which there is no order or hierarchy among categories
- The categories are mutually exclusive and exhaustive, and can be dichotomous or
multichotomous

Ordinal Measures
- Are discrete measures in which there is an order among the categories of this measure
- Categories are mutually exclusive and use a LIKERT SCALE
- Examples: polity index

Interval measures
- Are continuous, and can take on any value or any value within a range of values
- An example of this can be age in days, or government surpluses

Ratio measures
- Similar to interval measures, are continuous, however they have zero value that is
meaningful
- example : elected officials from a specific party

Operationalization
- Process in which concepts are transformed into measures

Good measures have DISCRIMINATORY POWER, and are VALID and RELIABLE
Measurement Validity
- Validity refers to the extent to which a measure captures the concept it is intended to
represent
- Is a matter of degrees, with some measures exhibiting a higher degree of validity than
others
Measurement Reliability
- Reliability refers to the degree to which a measure produces consistent and dependable
results , yields the same data each time it is used

Measurement Discriminatory Power

- Refers to the degree to which a measure is able to distinguish between two different
concepts
- Discriminatory power enables researchers to differentiate not only between two
concepts, but also two competing theories

Measurement Error
- The difference between the true value of an object and the observed value, the greater
the difference the greater the amount of error

Random measurement error

- Error unrelated to the real value being measured and is unlikely to bias results

Systematic measurement error

- A consistent, repeatable error associated with faulty equipment or a flawed research
design and is caused by measurement instruments that are used incorrectly. These
errors are likely to bias results.

Ecological inference problem

- Arise when researchers are unable to collect data directly on their subjects and then
make inferences about individuals based on characteristics of the group to which the
subjects belong

Quantitative Data
- Any type of data in a numeric form and is not limited to obvious forms of numeric data
measured in terms of frequency, weights, lengths
- Any data that is assigned to a numeric value

Observational Data
- Is data collected without researchers interacting with their subjects of their environments
- This data is subjective in nature and is derived from the content analysis of written, and
spoken material
- This type of data is not subject to research ethics boards, guinea pig effect or bias
Observer Bias
- Results from researchers unwittingly influencing the responses of their subjects through
their actions

Guinea Pig Effect

 - Results from subjects changing their behavior independent of any particular behaviors of
the researchers but simply because they’re being observed

Non Observational Data

- Collected through researchers either interacting with their subjects or intervening in
subjects environments
- Surveys and experimental data are examples

Five Data quality standards

1. Accuracy
2. Validity
3. Consistency
4. Completeness
5. Precision

Accuracy
- Refers to the extent to which the values of the data are correct

Validity
- Describes the extent to which the data depicts the measures they claim to represent
- Indicates the degree to which a measure represents a concept represents an idea

Precision
- Refers to the specificity of the data, data are more precise the smaller the units/intervals
in which they are measured

Completeness
- The extent to which the data includes the values of the whole universe of relevant cases,
analysis may be biased as a result of incompleteness

Consistency
- Refers to the absence of contradictions in the data, data may be inconsistent when the
data for all cases are not coded according to the same rules/procedures, or the data are
not collected using the same types of sources

Source bias
- Is the tendency to select information sources to support a confirmation bias or negativity
bias on a set of beliefs or values

Coding procedures
- Is an examination of the coding procedures used to assign quantitative values to
observations

Inspecting data
 - Checking values for individuals cases to conform with expectations as to how these
observations should be coded according to the codebook

Triangulating data
- Checking the data against itself but to also compare data to similar datasets

Surveys
- Are a method of collecting information from a sample of individuals by asking questions
in order to construct quantitative descriptors of attributes of the larger populations from
which the sample is drawn,

Surveys are categorized by the mode in which they are applied, their breath or scope OR the
population that they survey.

Expert Surveys
- Involve surveys of individuals with a specialized knowledge on a subject related to the
theme of the survey
- Chosen though non random sampling, and include educators such as doctors or lawyers

Non expert surveys

- Surveys for individuals that do not have a specialized knowledge on a particular subject
- Chosen through random surveys, and include the attitudes, desires and objective
information

Surveys provide comparable and generalizable data about the cognition, emotions and
behaviors of individuals.

Comparability
- Surveys facilitated comparisons due to the fact that every respondent is asked the same
questions in the same order, and most are close ended questions

Surveys have a high potential to yield GENERALIZABLE results, because they are often
involved in large case studies and random sampling

Threats to generalizability in surveys includes:

1. Low response rates
2. Survey attention
3. Observer or interviewer bias
4. Social desirability or guinea pig effect

Survey questions need to be and have

1. Clarity
2. Conciseness
3. The right terminology
 4. Demand
5. Efficiency

Researchers should avoid questions that have a high demand

Researchers should also avoid

1. Double negative questions
Ex: Please identify which of the following statements is not inaccurate
2. Double barreled questions
Ex: do you support increased spending on education and health?

High Demands LOWER the accuracy of responses

When dealing with sensitive research questions

1. Extend anonymity
2. Extend confidentiality
3. Create a safe environment
4. Avoid judgmental questions
5. Phrase questions in broad terms

List Experiments
- Respondents will be randomly assigned to a control or treatment list
- This is in order to facilitate indirect questioning that limits untruthful answers caused by
social desirability bias, shame or fear.
- The principle is to allocate respondents randomly to two different groups

List experiments work to help work around sensitive issues

Survey mode
- Means by which a survey is administered

Surveys can be conducted

1. By phone
2. By the internet
3. On paper
4. By person

Researchers will focus on three things when conducting surveys

1. The cost
2. Response rates
3. Potential for selection bias

A high response rate is necessary for a survey to be representative of the population from which
it is drawn.
Sample weights
- Used to create more representative sample which can occur when response rates are
low in some groups more than in others
- They adjust for the over and underrepresentation of groups by adjusting weight in the
sample

Margin of sampling error

- Indicates how much the results of a survey may differ compared to what would be found
if the entire population was surveyed

The smaller the margin of error, the better!

Margin of sampling error is based on sample sizes and is typically smaller for larger samples

Experiments
- Are a quantitative method of research
- Researchers will randomly assign individuals to experimental conditions and the
variation of only the explanatory variable across those conditions
- Therefore isolating the effect of the explanatory variable and helping to eliminate
alternative explanations for the outcome of interest

Experiments are randomly assigned

Labratory experiments
- Are conducted in a common location where the researcher has full control over the
environment
- They are useful due to the fact experimental manipulations do not exist in reality and
cannot be studied through observational approaches as a result.

Negatives of lab experiments include

- They may not represent real world effects because activities in the experiment are
artificial
- They may not represent a target population because they are conducted on non
representative populations

Field experiments or lab in the field

- Conducted on theoretically relevant populations
- Researchers do not have control over the environment they are in
- They involve changing an aspect of the real world and measuring the effect of this
change

There is a greater chance that randomization won’t be complete in field experiments

 - This is because participants may not comply with treatment
- And treatment and control conditions can be cross contaminated

Survey experiments
- Researchers do not have full control over the environment the survey is in
- They recruit samples representative to the population of interest

Natural experiments
- Include observational studies but with the added characteristic that the treatment is
external to the outcome as in experiments
- The assignment to control/treatment groups is a function of the real world not the
researcher

Construct validity
- Refers to the degree to which an experimental set up tests the theory it claims to
represent

Internal validity
- Refers to the degree to which the relationship between the explanatory variable and
outcome variable are causal
- It is the ability to attribute changes in the outcome variable exclusively to changes in the
explanatory variable

Internal validity is compromised when

- There is not a randomized selection of the treatment and control groups
- Subjects do not comply with treatment
- Observer and experimenter bias
- Cross contamination between treatment and control groups

External validity
- Refers to the extent to which the inferences from an experiment are generalizable

External validity is compromised when

1. Experiment participants are not representative of a theoretically relevant population
2. Attrition bias
3. Setting and time is unrepresentative of a population

Observational Study
- Quantitative research method that is used to test arguments in which researchers DO
NOT randomly assign subjects to treatment and control conditions
- Researchers do not interact with subjects or manipulate environments

Observational Studies are a LARGE NUMBER OF CASES STUDY that is analyzed through
statistics
A challenge of observational studies is causal inference

Causal inference
- Is the process of drawing a conclusion about a causal connection based on the
conditions of the occurrence of an effect

Strengths of observational studies

1. They do not pose ethical issues
2. They are more realistic
3. Contain a systematic and rigorous comparison of observation
4. Conclusions are objective
5. High generalizability

Limitations of observational studies

- Measurement can simplify concepts due to the lack of quantitative data
- Causal inference makes it impossible to observe outcome under treatment and control
conditions
- And the issue of reverse and simultaneous causation

Simultaneous Causation
- refers to when the explanatory variable influences the outcome variable, and the
outcome variable influences the explanatory variable

Reverse causation
- Explanatory variable does not cause the outcome variable, but the outcome variable
causes the explanatory variable 3