METHODOLOGY
LECTURE 1:
How does scientific research start:
Fundability
Scientific worldview
Advisor’s topic
Need in society
Methodology and technique
Current events
Profitability
Social or political beliefs
Personal experiences
Knowledge of the existing research
What makes a good research topic:
Original
Contributes
Convincing
What makes a good research question:
Specificity (generic-precise)
Connection in theory
Grounded in research
No assumptions
Scope (broad-narrow)
Open-ended
Clarity
What makes a good research question wording:
- How does x influence...
- What factors shape...
- How does x perceive...
- … in the context of…
- ...for (x population)…
,LECTURE 2
Worldviews: different takes of what the purpose of research should be
1: Nomothetic approach
Human behavior is generalizable, predictable, and motivated by events, personality, and people.
Researchers aim to make generalizations about human communication that will hold true across
space and time.
Understanding behavior best done: by isolating factors
Example: letter of more difficult topic has fewer errors
- Hypothesis-based
- Systematic manipulation
- Clear causal relation
- Generalizable result
- Quantitative
2: Idiographic approach:
Each person is unique, unpredictable, and self-motivated
This view assumes that knowledge is socially constructed out of interaction between people and is
subjective.
Understanding behavior best done: from the participant’s perspective and by considering the whole
situation
Example: Researcher joined a class to watch children play and how they make groups with friends
- Data-driven approach (no hypothesis)
- Real communicative setting (classroom)
- Diverse set of observations (language, physical, etc.)
- Qualitative
Creswell and Creswell (2018) identify 4 worldviews:
- Postpositive: challenges the notion of truth but emphasizes cause and effect and the idea
that the world is governed by laws or theories that can be tested or verified. Big ideas,
hypothesis testing, quantitative methods, objective observation, measurement.
- Constructivist: individuals seek understanding and construct own views. Interpretive,
qualitative, from observation to theory development.
- Transformative: change oriented and argues for mixing research with politics to confront
social oppression and change lives. Indivudal people without rights, variety of research
interests, including action and critical analyses.
- Pragmatism: focuses on solutions and using all possible approaches to understanding these.
Mixed-method research, real world, practice oriented, focus on problem rather than
method.
,Research traditions:
- Rhetorical: words; considers discourse, debate, or discussion
- Semiotic: signs and symbols
- Phenomenological: experience of others
- Cybernetic: flow of information; communication in information processing and feedback
- Sociopsychological: interaction of individuals; attitudes, perceptions, influencing each other
- Sociocultural: considers production and reproduction of social order; shared meanings,
social structures, conflict, alienation, and the individual as products of society.
- Critical: power, oppression, emancipation
Quantitative versus qualitative
Quantitative methods (numbers)
- Experiments (manipulating something)
- Surveys
- Content analysis
- (Meta-analysis)
- (Computational modeling)
Qualitative methods (observing and interviewing people):
- Interviews
- Focus groups
- Case studies
- Observational studies
- Diary studies
Triangulation: combining research methods to neutralize the weakness of single research methods
Induction and deduction
Induction: from observations to a theory that might explain your observations. Specific > general
(observation > pattern > hypothesis > theory)
Deduction: from theory to defining observations you will make to test the theory; general > specific.
(theory > hypothesis > observations)
Empirical cycle: observation > analysis > interpretation > theory > theory > hypothesis > ob…
Abduction: from effect to possible causes.
Deduction versus induction: deduction more efficient than induction because it leads to a specific
observation that will test your hypothesis. With induction, you have a further step: finding a way to
decide which of the many possible theories you induced from your observations are correct.
Induction requires the confidence that you have enough observations to support your conclusion
and that you can rule out all the other conclusions that might also be derived from your
observations.
,Starting points for research:
- Look at examples of the setting/ product etc that you are interested in (insurance)
- Consider own personal experiences with assumptions
- Map the different stakeholders and relevant factors (location, friends, older people)
- Read the existing research
- Define your research purpose
o Exploration: curiosity-based research
o Description: compelling reading
o Explanation: attempt to answer the why question (greater credibility when they
have a prediction)
o Prediction
o Control: researching with a view to being able to predict and manipulate physical
processes
o Interpretation: attempts to understand human communication from the point of
view of the people doing it (in the other person’s shoes)
o Criticism: understand and explain the way in which communication is used to
exercise and maintain power in groups, organizations, and societies.
Other dimensions:
Wide scope - narrow scope
Dispassionate - involved
Objective - subjective
Your questions - their answers
Large sample - small sample
Quantitative - qualitative
BOOK NOTES
Noise: extraneous information or distractions that can disrupt an interaction
PSAs (public service advertisements): targeted communications designed specifically to promote
positive attitudes and behaviors. They focus on public interest topics such as health, education,
safety, the environment, and other social causes.
,LECTURE 3
Theory requirements: broad area, can be disproved, generate hypotheses, by evidence
Constructs: variables that reflect abstract concepts (ideas, attitude, motivation, appreciation)
Operationalizing constructs: define them in such a way that they can be measured.
Variables levels:
- Nominal: categories (gender, favorite food, nationality)
- Ordinal: ranks (education level, top 5 favorite food)
- Interval: equal intervals (temperature, 7 point likers scale)
- Ratio: also zero (calorie intake) (2x zoveel)
Variables roles:
- Dependent: measured outcome, changes as a result of changes in another variable
(perceived attractiveness, brand attitude, user satisfaction, reading time, test performance)
- Independent: cause or correlate with outcome, often manipulated, usually nominal
(language errors, profile picture, age, gender, cultural background)
- Moderating: when the s the relation between IV and DV depends on the level of some other
variable (opinions about correct language use > negative effect)
- Mediating: the causal effect of IV and DV is indirect, the IV influences the mediating variable,
which in turn influences the DV (language errors > impression of intelligence > perceived
attractiveness/ Nutri-score A > perceived healthiness > purchase intention)
Interaction effect: where the combination of levels in two or more independent variables interact to
cause differences in variation in the dependent variable, when you can’t state what the effect of IV
is without mentioning the other IV
Moderating different than interaction effect:
- Clear idea what moderate /the cause is
- Not interested in the effect of the moderating on the DV, only its moderating role
Variables to avoid:
- Third variable: explains the correlation between IV and DV, because the third variable causes
changes in both variables (high statistics scores > high methodology scores > smartness)
- Confounding variable: unknown variables that co-vary with the levels of your IV
Hypotheses:
- Must be testable
- Must specify the relation we expect to find
- Not said to be true or false, only supported or not
Correlation:
- Two tailed: there is a relation between two variables (without direction)
- One-tailed: specify the relation between two variables (predicting the direction)
Correlation DOES NOT IMPLY causation
(high statistics scores > high methodology scores).
,LECTURE 4
Experimental design: expose participants to controlled conditions:
- Can help determine whether there is a causal relationship between variables.
- Can isolate the effect of different variables on a variable of interest.
- Requires random assignment
We do experiments to determine if A (IV) causes B (DV), A causes B only if:
- A takes place before B
- A and B covary: if A changes, B changes
- It is certain that A causes the change in B (make sure that all else is equal)
Control the correlation by: sampling > random assignment > pretest > treatment > posttest
Weakness experimental design:
- Experimental conditions rarely resemble real-life situations.
- Online experiment threats
o High attrition rates (dropouts)
o Inattentive participants (people can take a break)
o Multiple submissions
o Non-representative participants
o Noisy measurement of time responses
Inconsistency of question lay-out across devices (part of scale hidden on phone)
Solomon four-group design: pre-test, post-test, skipped pre-test and without manipulation
(sophisticated experimental designs)
Pretests are not always necessary, especially when participants are randomly assigned
Experimental manipulation often not something that is present of absent, but takes different forms
,Factorial designs: two or more IV’s which are crossed with each other, 2x2 (interaction effects)
- Between-subject designs: participant is exposed to only one level of each IV
- Within-subject designs: participant is exposed to multiple levels of each IV
- Mixed design: a given participant is exposed to only one level of at least one IV, and to
multiple levels of at least one other independent variable
Internal validity: causal relationship is not explained by other factors:
- Spurious relationships: similar to the third variable problem, except that there isn’t any
meaningful relationship between two variables (cheese consumptions/ died people in bed)
- Selection bias: groups are different (not randomly assigned)
- Attrition: are participants in all groups equally likely to complete the study (bored in second)
- Diffusion: experimental groups interact and exchange information outside the experiment
- Maturation: when people’s knowledge, skills, beliefs, attitudes change over the course of
your study (children get better at reading, after reading books but they also went to school)
> use control groups
- Repeated testing: people can get better at some tasks, questions as they have seen them
more often
- Experimenter bias: researcher causing the effect themselves > make sure the person
collecting data/ running the experiment is blind to the research purpose and conditions
External validity:
- If our experiment resembles all other situations/ the outside
- Hawthorne effect: observation changes behavior
Construct validity: Are u measuring what you supposed to measure?
, LECTURE 5
Literature review:
- To summarize what we know
- What we do not know (gap)
- To engage with theory
- To design our research question and hypothesis
Writing literature review:
- Individual studies: questions, theory, method, conclusions
- Themes across studies: shared conclusions and ongoing questions
- Gaps and opportunities: emerging questions and overlooked points
Literature review as a study:
- Lack of coherence between different approaches in fields, methods, concepts and theories
Review literature as study is the growing body of research with many approaches and a lack of
coherence
Systematic review: explicit statement of objectives, materials and methods and has been conducted
according to explicit and reproducible methodoly.
Systematic review: PRISMA:
- Eligibility criteria: specify conditions for including or excluding a study from a systematic
review
Disadvantages:
- Not up to date
- Not available, require login