Psychology: Research Methods Intro

Introduction to Research Methods in Psychology

• The course includes three lectures, with recommended reading material from textbook chapters 2 and 2S, and a handout available on Canvas
• While the textbook serves as good background material, the lectures and handout remain the primary resources for test and quiz content

Contact Information

• Prof Douglas Elliffe's office is 301.236
• The best way to contact by email at [email protected]
• Important course announcements will be made via Canvas

Importance of Research Methods

• Research methods are essential for answering questions, such as:
  • The impact of social media on mental health
  • The correlation between religious belief and happiness
  • Safety of vaccinations etc
• Systematic research methods are essential for experimental design
• The last lecture will explain how to analyse the resultant data

Scientific Research Simplified

• Merton's norms for science, simplified
  • Science and its methods are common to everyone
  • Scientific validity is independent of the researcher's status
  • Scientific claims should be critically scrutinized
• It is critical that science is inclusive and encourages open debate

Research Cautionary Tale

• An experiment tried to see if memory could be transferred between flatworms, or planaria
• This grew from interest in the physical representation of memory in the brain
• The experiment sought the so-called engram
• This experiment isn't examinable, it's just a true story

Flatworm Experiment

• Flatworms were trained to turn right for food
• They were then fed to untrained flatworms to see if memory could be transferred
• Flatworms that ate trained flatworms also turned right
• This allegedly proved the existence of the engram

Flaws of that Experiment

• The flatworm experiment hasn't been successfully replicated
• Potential issues with the original study
  • It could've been a fluke
  • It might have been badly conducted
• These lectures are about research design - designed to prevent bad practice

Basics of Good Research Design

• Research involves inquiry
• Good research design uses careful question asking to obtain meaningful results
• This course aims to cultivate critical, cautious, skeptical consumers of research
• Good research design is a formalised way of asking questions, ensuring meaningful answers

Ethical Considerations

• Conduct psychological research well
• Psychological research has to uphold ethical standards
• Subjects are usually people, people's time shouldn't be wasted
• How people are treated is impacted by psychological research
• Because of this duty better do it right

Basics of Research Design

• Various factors to deliberate when designing research:
  • Aim/research question
  • Variables of interest
  • What to measure + how
  • What to manipulate
  • Who participates and how to recruit
  • What participant info to collect
  • How to analyse data

The Stroop Effect

• Reading becomes automatic, making it hard to inhibit
• When the color conflicts with the color named, reaction time slows

Terminology: Dependent Variable (DV) and Independent Variable (IV)

• The dependent variable (DV) is the measure of behaviour being studied
  • Example is number of colours named in the Stroop effect
  • Appears on the y-axis of a graph
• The independent variable (IV) is manipulated to affect the DV
  • Example is the match between colour and name in the Stroop Effect being tested
  • Appears on the x-axis on a graph

Terminology: a Word of Warning

• Terms are used in psychology and common English, meaning technical terms might not always align with English usage of the words
• The word "reliable" isn't different in technical terms, however it carries a different undertone in everyday usage

Terminology: Reliability and Validity

• Dependent variables (what to measure) need to be reliable and valid
• Reliability means replication – will you get the same results if you measure the same variable again?
• Validity means that the dependent variable should measure what it intends to
  • Example, is IQ score a valid measure of real intelligence?

Check Your Understanding

• The IV is what the researcher measures
• One group drinks coffee (caffeine group) while the other drinks decaf (control group)
• Then both read word lists and write down all the words from the list that they remember 10 minutes later

Terminology: Population and Sample

• Population includes all the scores and events of interest and the sample includes the representative subgroup of the population
• Samples are used to infer general results about populations because it's usually impossible to accurately measure the entire population

Terminology: Sampling Error and Sampling Bias

• Two possible sampling errors: sampling error and sampling bias
• Sampling error means results from repeated samples from the same population differing, which is unavoidable but can be minimised by bigger samples
• Serious sampling bias invalidates the research but can be avoided by doing random sampling

Sampling Bias Example

• 1948 US election where polls predicted a win for Thomas Dewey, but Harry Truman won easily
• This was because richer people were more likely to have phones and were thus overrepresented in the sampling, but also more likely to vote Republican

Examples of Sampling Bias

• Psychological testing lacks diversity because most research is done on WEIRD (Western, educated, industrial, rich, and democratic) populations
• Confirmed COVID-19 cases in the U.S. don't capture the full picture because testing was primarily restricted to those with severe symptoms
• Consumers with extreme experiences review products more than people with moderate ones

Things to Remember About Sampling Bias

• Misrepresentation must be systematic for a sample to be considered biased
• Being more likely to favor someone or something doesn't necessarily mean the sample is biased

Now That There's a Bit of Knowledge...

• Independent and dependent variables
• Important things to consider when choosing your dependent variable (reliability and validity)
• Important things to consider when recruiting participants/choosing a sample
• Take a look at some general kinds of research designs

Two General Kinds of Research Design

• Observational Design measures is when you measure two dependent variables and look for a relationship
• This examines relationships but does not constitute manipulation
• This method is most useful when only applicable for ethical or practical reasons

Weakness of Observational Designs

• Correlation doesn't imply causation
• The findings would also be impacted by the third variable problem

Correlation vs Causation

• It is important to remember from these lectures that correlation does not imply causation

Two Kinds of Experimental Design

• Experimental Design manipulates IV and look to see DVs on effect, but does imply causation
• Experiment design should be used when it's ethical and practical
• Experimental designs are more powerful than observational designs, that have implications that should be proved

The Fundamental Principle of Research Design

• Eliminate all explanations of results except one
• In an experiment, ensure a change in behaviour to know it was caused by your IV
• Alternative explanations of results are called confounds or confounding variables - and these are bad

General Approaches to Eliminating Confounds

• Hold the confounding variable constant, especially good for environmental or external confounds
• Randomize the confound, especially good for subject-based or internal confounds

Thought Process for Research

• Think about possible confounds
• Have the confounds been controlled
• If the confounds haven't been controlled then the research shouldn't be trusted

Minimising Confounds in Research

• Usage of design depends on research question
• Within-subjects design: each participant goes through all experimental conditions
• Between-groups design: each participant is exposed to one experimental condition

Within-Subjects Design

• Each subject tries to learn a list of words in both "sober" and "drunk" conditions
• Then compare the number of words remembered when "sober" vs "drunk" for each
• Helpful for internal confounds because subjects act like their own "control"
• Risky because of environmental confounds

Between-Groups Design

• Separate groups of subjects in "sober" and "drunk" conditions
• Then compare average number of words for "sober" vs "drunk" groups
• Then see if subjects respond differently in different scenarios

Between-Groups and Within-Subjects

• "Sober drunk" means individual tries both, whereas "sober / drunk" means different groups participating
• The terms imply the groups participating through the usage of a slash
• Helpful for environmental confounds because only participating in one condition, therefore subjects allocated to the group randomly can prevent this

Naming Differences

• Between-groups design can also be dubbed between Subjects, independent samples, and/or independent groups
• Within-subjects design can be referred to as repeated measures for the conditions

Matched-Pairs Design

• A way to get the best of both worlds
• First give a pre-test on memory ability before the experiment
• Second, find pairs of subjects with similar memory ability
• Randomly allocate one member of each pair to each condition ("sober" and "drunk")

Why Use Matched-Pairs Design?

• It keeps environmental confounds constant, and essentially keeps subject confounds constant too
• It's a good design, however there's a labor-intensive component that prevents usage as often

Standard of Comparison

• Compares an experimental group to a control group
• The 2 groups are both treated the same except the experimental group receives the treatment while the control group doesn't

Single and Double Blind

• Single-blind refers one aspect kept from the subject
• Experimenter expectation impacts results, therefore a double-blind is best to where it's unknown if the subject the recipient of the true variable or not
• Double-blind design is the gold standard