Lecture 4

Formulating and Testing

Hypotheses

Philippe Chouinard
Department of Psychology, Counselling, and Therapy
School of Psychology and Public Health

Readings:
Chapter 3

August 22, 2023

Outline of lecture
• Zimbardo Stanford prison experiment
• Research ideas
• Literature review
➢ Functions of the literature review
➢ Scope and length of the literature review
➢ What does the literature review seeks to answer?
➢ Doing a search for the literature review
➢ Summarising a study
➢ Four assessment criteria to consider
➢ Critical evaluation
➢ Synthesis
➢ Further advice
• Testing hypotheses
➢ Formulating the problem
➢ Formulating a hypothesis
➢ Null hypothesis statistical testing (NHST)
➢ Complimentary statistics to NHST
Zimbardo Stanford prison experiment

• Examined how the prison environment can drive

people to harm others.
• A prison was simulated at Stanford University.
• Participants were either ‘prisoners’ or ‘guards’.
• Prof Zimbardo served as the prison superintendent.
• Guards became sadistic and used cruel punishments.
• Prof Zimbardo continued even after it became clear
that the ‘prisoners’ were suffering.
• The experiment ended on Day 6 after Prof Zimbardo’s
future wife made him realise that he lost sight of what
is important.
Zimbardo Stanford prison experiment

Zimbardo Stanford prison experiment and the

Belmont report:

1. Respect of persons

2. Beneficence
Research ideas

1. Everyday life: We often come into

contact with many questions in need
of a solution, which can lead to
researchable questions. Personal
experiences shape our ideas.
2. Practical issues: Societal and
industrial problems that are in need
of solutions to encourage researchers
to find answers.
Research ideas

3. Past research: Research tends to generate

more questions than it answers. Every study
tends to lead to a subsequent study. It is an
ongoing process.
4. Theory: An explanation of “how” and “why”
some part of nature works provides
predictions that can be tested. Theories are
constantly tested and revised as needed.
Research ideas

Scientists are humans and

therefore their thinking and
motivations will be directed by
their biases.

To understand nature, we can’t

overlook topics. Hence, potential
biases could be checked by the
scientists and by having them
come from different
backgrounds.
Research ideas

Not all ideas can be confirmed or

refuted, particularly issues related to
morality and religion.

Big Bang
Functions of the literature review

Literature review: A scholarly paper that presents

the current knowledge on a topic, including
substantive findings and theoretical and
methodological contributions.

Before embarking on any study, it is important to do

a literature review:
• Need to know knowledge gaps.
• Allows one not to repeat what has been done
before.
• Identify ways to test your idea.
• Identify and not repeat mistakes of the past.
Functions of the literature review

Focuses on previous research

Shows how your research will fill ‘gaps’

Explains the necessity for your research

Sets boundaries for your research

Scope and length of the literature review

Avoid verbosity

Be brief but remain focused in

synthesising the literature
What does the literature review seeks to answer?

What has been done before?

What have others said?

What is the relevance?

What is the difference?

Is there agreement / disagreement?

Are there flaws?

Doing a search for the literature review

1. Getting started: Get to know the library and its resources, including its online resources.
Come up with keywords that can narrow your search to your topic in a manageable
manner.

2. Define your objectives: To what end are you doing the search? Is it to know what others
have done or determine what methods you could use? Coming up with objectives will
allow you to search more strategically.

3. PsychInfo: An electronic bibliographic database of abstracts and citations to the

scholarly literature in psychology. The procedure is to enter a list of search terms and let
the computer search for books and articles focusing on these terms. Then, read their
abstracts and identify those that are of interest. Other databases, such as PubMed, exist.
Doing a search for the literature review

4. Obtaining resources: The library is likely to have the books you need and can provide
you with online access to the journal articles you need. If not, they can get what you
need from other libraries using “inter-library loans”.

5. Review articles and books: Summarise the literature up to the time of writing the
review article or book. Good place to start reading.

6. Journals articles: Medium where research experiments are typically published.

Basic components of the literature review

1. Introduction
• Introduce topic
• Describe scope and organisation of the review
• Clarify purpose of the review

2. Theory
• Review past and present literature
• All research has a precedent build upon theory
• Integrate key points of different theories and make appropriate inferences
• Relate previous research to your topic
Basic components of the literature review

3. Instruments (not always)

• Describe instruments used to examine your topic
• Describe reliability and validity estimates
• Describe samples examined

4. Summary
• Draw conclusions from your investigation
• Provide a concise account of knowledge (and instruments used, if applicable)
Summarising a study

How was the topic studied?

What was found?

Make sure you read the
original sources
Which issues were highlighted?

How have key terms / concepts been used?

How does it relate to your research?

Other considerations: Participant information, sample size, location of study, type of study,
nature of task.
Four assessment criteria to consider

RESARCH
SYNTHESIS
INQUIRY

CRITICAL Writing Style

EVALUATION
Research inquiry

A valid research question has been formulated

The context of the inquiry has been described and
justified
RESARCH
Excellent quality reference information has been
located
INQUIRY
Information sources have been acknowledged with
conventions appropriate to psychology
Information provided is relevant to the research
question
Original sources were read and evaluated
Critical evaluation

What is the main idea being put forward?

What points develop the idea?
What evidence is provided? CRITICAL
Do you agree with the conclusions? EVALUATION
Is it a response to another text?
What is the underlying theory?
Model?
Strengths, weaknesses, limitations?
Other considerations: Sample size, research design, measures used, biases, confounding
variables, reliability, validity.
Synthesis
Not just a list of the
work of others

Identify issues
Identify consensus
among studies Synthesis highlighted by different
studies

Highlight differences
and similarities
between studies

Your voice guides the discussion …

Synthesis

Further considerations …
• Compare and contrast between studies
• Draw similar conclusions from studies
• Criticise the methodology of studies SYNTHESIS
• Draw areas of disagreement between studies
• Highlight exemplary studies
• Identify gaps in the literature
• Explain how your research relates to previous studies
• Evaluate what the literature says as a whole
Writing style

The work has been formatted correctly (i.e., APA)

Correct discipline-specific terminology has been used and,
where appropriate, care has been taken to explain
terminology
Writing Style
The content is very clear and easy to understand.
The information is very well organised and presented in a
logical order.
No repetitive information is present
The writing is fluent, engaging, and interesting to read
The word choice, sentence variety, and expression are
excellent
Further advice

This notion also applies to many other facets of writing

Formulating the problem

An exact definition of the problem is important

because it guides the research process.
Research problem: An interrogative statement that
asks what is the relationship between two variables.
The problem …
i. includes variables that can express a
relationship.
ii. is phrased as a question so it is directly
addressed.
iii. is capable of being empirically tested.
iv. has some degree of specificity and not be too
broad.
Formulating the problem

Examples of problems that need improvement:

“Do space creatures influence the behaviour of La Trobe students?”

→ Not feasible.

“What effect does the environment have on learning ability?”

→ Meets the first three criteria but not the last one (i.e., requires more specificity).

Example of a good problem for investigation:

“What effect does the amount of exposure to words have on the speed with which they are
learned?”
Formulating a hypothesis

Hypothesis: The best prediction or a tentative

solution to the research problem.
A hypothesis must fulfill the following criterion: It
must be stated so that it is capable of being either
refuted or confirmed.
Conventional statistical testing requires a null
hypothesis because these methods are designed to
determine if you can reject it.
Null hypothesis: A statement of no relationship
among the variables being examined.
Null hypothesis statisticalNoise
testingProblem
(NHST)

• In NHST, a hypothesis cannot be entirely proven, ever, because it deals

with probabilities.

• Experiments are carried out to determine how probable the null

hypothesis is true.

• The p weighs the strength of the evidence in support of the null

hypothesis.

• A small p (less than 5% by convention) is suggestive that we can consider

the alternative hypothesis instead.
Null hypothesis statistical testing (NHST)

NHST corresponds to the assumption in court that

the accused is innocent (null hypothesis) until
proven guilty (alternative hypothesis).

Type 1 error: Rejecting the null hypothesis when

the null hypothesis is TRUE.

Type 2 error: Failing to reject the null hypothesis

when the null hypothesis is FALSE.

Rejection of the null = Guilty verdict.

Type 1 error = Conviction of an innocent.
Type 2 error = Letting a criminal go free.

NHST convicts an innocent 1 in 20 times.

Null hypothesis statistical testing (NHST)

Example: A vaccine study produces a p of 0.05.

This p indicates that if the vaccine had no effect,
you’d obtain a significant result in 5% of studies
due to random sampling error.

Replication crisis:
• Twenty studies are carried out hoping to find
remarkable finding X.
• One study confirms remarkable finding X and
publishes in prestigious journal (e.g. Nature or
Science).
• People now believe remarkable finding X is true
when it is not.
Null hypothesis statistical testing (NHST)

Open science collaboration study:

• Reran 100 psychology studies published in 2008

in good journals.
• On average, effects were half as large as what
was originally reported.
Open Science Collaboration (2015). Estimating the
• Only 36% of the replications produced reproducibility of psychological science. Science,
statistically significant results. 349(6251), aac4716. doi:10.1126/science.aac4716

• Psychology has a replication crisis.

Complimentary statistics to NHST

Effect sizes: Provides a measure of magnitude.

• Cohen’s d: A measure of how large a difference exists
between two conditions (small: 0.2; medium: 0.5; large:
0.8).
• Clinical versus statistical significance.
• Clinically unimportant effects may be statistically
significant if a study is large.
Are the extra
minutes of exercise
convincing?

Can we recommend
drinking to get
people to exercise
more?
Complimentary statistics to NHST

Bayesian statistics: A different way of doing things that

weighs both the alternative (H1) and null (H0) hypotheses.
• Bayes factor (BF): A ratio between support for H1 and
H0. BF = Support for H1/ Support for H0.
• Bayes factor >3: Substantial support for H1.
• Bayes factor < 0.33 (or 1/3): Substantial support for H0.
Complimentary statistics to NHST

Watch the following YouTube video on Bayesian statistics from 0:00 to 5:30:

https://www.youtube.com/watch?v=9TDjifpGj-k

Try to understand Bayesian statistics in broad strokes. There is no need to understand

the details, including the formulas.