research methods final exam

Why do people believe in conspiracies?

1. epistemic motives (fear of randomness)

2. existential motives (regain control)

3. social motives (to feel like they belong)

What is science?

1. based on observation (systematic empiricism)

2. testable questions

3. results must be shared

Testable ideas

• can be supported/opposed with data

• can be falsified (pseudoscience like ESP)

Qualitative methods

produce qualitative data

ex. written text, photos, interviews, videos

Scientific method

1. find research topic

2. generate good research questions

3. develop hypothesis

4. identify variables and population

5. analyse data

6. draw conclusions, report results

Intervention studies are particularly useful for

Applied questions

important basic research advancement occurs when

1. a key idea is challenged

2. a new theoretical model is developed that parsimoniously explains a phenomenon (need to belong theory)

3. new method is uncovered that can tackle previously unexplored questions

basic research

solves theoretical issues

better understand psychological processes and behaviour

applied research

apply theory to solve practical problems

ex. policy implementation, interventions, persuasion

applied advancement occurs when

1. new treatment can be implemented to help with a problem (ex. loneliness interventions)

2. making a particular choice helps the problem (decision making)

3. people can be convinced something is a problem (persuasion)

4. the government can help solve the issue (policy implementation like organ donor opt-OUT)

primary sources

where the research was originally published

• written for other experts

• includes full methods and results of study

secondary sources

summarise info from primary sources

• written to be accessible to non-experts

• can be inaccurate and incomplete

Impact factor

average number of citations

>1? no one is citing the journal

higher # = more relevant and influential content

respect for persons

respect people’s autonomy with informed consent. can withdraw consent without consequence

violations: children, mental disorders, hiding risks

concern for welfare

benefits of the study outweigh the risks

violations: not doing everything possible to reduce risks

seek justice

participants must be compensated fairly, risks must be distributed equitably, researchers must act with integrity

violations: unreasonably low or unequal compensation, lying to participants

ex. Tuskgee study

deception

study must be low risk and requires debriefing

open data

PROS: improves trustworthiness of findings, learning value

CONS: confidentiality breach (de-identification is not 100% possible)

theory

coherent explanation or interpretation of one or more phenomena

→ answers WHY questions

ex. (phenomenon happens BECAUSE of theory)

phenomenon

repeatedly shown to be true, an established finding

Constructs

variables that can’t be directly observed

ex. memory, creativity, costs/rewards

Perspectives

overall lens you’re looking at something with

ex. how you look at attachment theory with a developmental lens, social lens, biologically

model

explanation of a specific phenomenon

→ narrower and more specific than a theory

theories vary in terms of

1. formality

2. scope (broadness)

3. theoretical approach

functional theories

why things happen the way they do

mechanistic theories

explains how things work (like neuroscience)

direct replication

repeating the study in the same manner

• increases confidence in the hypothesis

conceptual replication

using different methods to test the same research question

• increases confidence in the theory

Daryl J Bem article

said you can get rid of data that doesn’t agree with your hypothesis.

huge data ethics and validity problem!

Nosek Large Scale Replication attempt (2015)

tried to replicate a bunch of studies, only to find that most of them were invalid (lots of false positives)

Causes of false positives

1. incentives to publish and take shortcuts

2. questionable research practises

3. file drawer problem

P value

probability that you would get these results in the null hypothesis was true

Questionable research practises

selective data elimination, faking data like Diederick Stapel, add/drop covariants, adjust conditions as desired

Mitigating QRPs

open sharing, preregistration, journals publishing null results, publishing replications

test retest reliability

same test is given twice with some time in between

• good for stable qualities (personality) not temporary states (mood)

• issue: practise effects

parallel forms reliability

different forms of the same test used

internal consistency

• split half correlation

• test with chronbach’s alpha

Chronbach’s Alpha

tells us how internally consistent our results are

the higher the better

issue: you can make items similar and therefore redundant

testing the validity of a measure

face validity

content validity

criterion validity (convergent and predictive)

discriminant validity

convergent validity

does it correlate with similar variables

predictive validity

does it predict expected outcomes

content validity

does it capture all construct nuances

face validity

does it look like it measures what it’s supposed to

discriminant validity

measure shouldn’t correlate with theoretically different variables

categorical data

each value represents a discrete category

• pie charts, bar graphs

numerical data

each value represents a real number or place on a continuum

• scatterplots, histograms

Descriptive statistics

response rate, central tendency, variability

time series graphs

line graph that shows how something changes over time

x axis: time (discrete)

y axis: continuous variable

survey research

uses self report

tries to obtain generalisable samples (large and random)

interviews, phone surveys, questionnaires

Good survey items

Unambiguous, specific, relevant, brief

Likert Scale

rating scale used to assess degree of liking or agreement

• typically 7 points, sometimes 5 or 9

Simple random sampling

everyone in the population has an equal chance of participating

• sample should look similar to population (representative)

stratified random sampling

“some of all”

• important sub groups are identified

• random sample of each subgroup to mirror population

what do employers want?

skills in: interpersonal, teamwork, initiative, work ethic, communication

Strategies for academic success

1. reading for comprehension

2. studying effectively

3. time management

4. caring for yourself

Academic networking

office hours, class discussions, department events, joining research labs, attending conferences