Research Methods 1 revision

Experimental design

• ·        Between- groups (Independent measures) -  Compare different participants in different conditions.

+       No carry-over effects.

+       Process is quicker for participants.

-        Individual differences

-        Twice the number of participants needed.

• ·        Within-groups - Compare the same participants in different conditions.

+       Individual differences reduced.

+       Fewer participants needed.

-        Boredom.

-        Carry-over effects.

Validity

·        Validity = the extent to which your study measures what you intended it to measure.

-        Face = Do items on a test appear to measure what they said they would?

-        Predictive = Can test scores be used to predict events?

-        Concurrent = Do scores correlate with other measures taken at the same time?

-        Construct = Does the test measure the construct it was designed to measure?

-        Ecological = The extent to which the experiment reflects the real world.

•     Reliability = the extent to which a measure produces consistent results

-        External reliability (Relates to consistency of a test over time)

-   Test using test-retest reliability = assess the correlation between scores taken at 2 points in the time from the same sample.

-        Internal reliability (Is the test consistent within itself?)

-        Test using the split-half method or Cronbach’s alpha.

Variables (e.g. IV, DV, extraneous)

-         Independent variable = The variable which is manipulated by the researcher (causal variable).

-         Dependent variable = The variable which is measured.

-         Extraneous variable = variables that potentially affect the relationship of the IV and DV.

-         Confounding variable = variables that differ systematically with the IV.

• To prevent confounding variables we can do 3 things:

1.      Match conditions of key variables e.g age

2.      Standardise the procedure

3.      Randomise the sample

-         Control variable = The condition when the IV is absent.

Evaluating research

1.       Individual differences -  Randomly allocate.  

                                             -  Use a within-subjects design.

2.       Practice effects  - Use a between-subjects design.

3.       Demand characteristics - Use a between-subjects design.

                                                 - Deception.

4.       Situational variables  - Think more carefully when designing the study.

5.       Floor and ceiling effects  - Use a pilot study

6.       Boredom/tired  -  Counterbalance

                                   -  Rest breaks

                                   -  Between-subjects design

Hypothesis testing

·        Null hypothesis = no difference will occur

·        Alternate hypothesis = A difference will occur

-         Directional (one-tailed) = Prior research

-         Non-directional (two-tailed) = no prior research

Types of studies (e.g. experimental, cross-sectional)

·        Experiments explore causal relationships and manipulate the values of the independent variable to see if there is any effects in the dependent variable.

·        To control for chance differences, we can use random allocation of participants.

·        Quasi-experiments occur when the researcher cannot randomly allocate participants to the IV condition, as the change is naturally occurring.

Sampling

-         Random sampling (Leads to representative sample)

1.       Simple random sample – everyone has an equal chance of being picked.

2.      Stratified sample – The population is divided into groups and a simple sample is conducted on each.

-         Non-random sampling (Less representative)

1.      Voluntary sample – Members of the population self-select themselves to participate. It becomes a snowball effect if they ask their friends to participate.

2.      Convenience sample – Researchers ask members of the population who are easy to reach.

Effect size

·        A measurement of the magnitude of the effect.

-         If the effect is small, ensure the experiment is designed well enough to have a chance.

Effect size =        (Mean of cond. 1 – Mean of cond. 2)

Pooled Standard deviation of both

Referencing

·        Journal article – Surname, Initials. (year). Title. Title of the journal, Volume (Issue). DOI.

·        Important to acknowledge the author to avoid plagiarism and give credibility to work.

Structuring sections

• Title   
•  Abstract
• Introduction  - Overview of the topic , Relevant research, Gap in the literature, Aims and hypotheses.
• Method  - Design, Participants, Materials, Procedure
• Results - Report and describe results relevant to the research question, No raw data,  No rationale
• Discussion - Summarise key findings, Interpret results, Discuss limitations and wider implications.
• References
• Appendices

Extracting key information from an abstract

·        Searching for background Info = Introduction

·        Looking for the design of the study = Method

·        Future research = discussion

Selecting research design to suit hypotheses

·        Psychological investigations:    

-        Experiments (Cause + effect)

-        Observations (Initial investigations)

-        Qualitative (In-depth understanding)

-        Cross-sectional (Relationships)

• Observational studies

-        Ask several research questions

-        Unstructured – Researchers record behaviour they see and monitor aspects of the phenomenon relevant.

+       Qualitative data

+       Rich and detailed

-         Subjective?

-        Structured – Researchers do not observe anything. Predetermined behaviours are observed.

+       Quantitative

+       Objective

+       Observer consistency

-         May miss details

• Cross-sectional research

-         Usually consist of questionnaires

-         Gives rise to correlational data

+       Naturalistic

+       Can recruit lots of participants

+       Ask lots of questions

+       Investigate many topics

+       Can measure many variables at once

+       Stronger if use validated scales

-         People may not tell the truth

-         Can not infer causality

•  Longitudinal studies

-         Repeated observations over an extended period of time

-         Usually, cross-sectional

•      Advanced experimental design

-         More than 1 Independent variable

-         Factorial

-         Explore how variables interact

Principles

1.      Respect

-         Informed consent

2.      Competence

3.      Responsibility

-         Deception

-         Protection from harm

-         Debriefing

-         Confidentiality

4.      Integrity

• Informed consent

-         Participant information screen

-         What the study’s aims are and what it will entail

-         Contact details

•    Deception

-         Only used when necessary

-         Explain as early as possible in the study

•     Protection

-         Put in place resources for support

-         Potential harm should be made explicit

•    Debriefing

-        Tell the true aims at the end of study

What is qualitative data?

·        Focuses on the smaller picture

·        Gains rich descriptive and possible explanations of peoples meaning making.

Advantages of qualitative research

+       Can answer the ’how’ and ‘why’ questions

+       Help make sense of patterns

+       Can help generate theories

Approaches to qualitative research

·        Insiders

-         Researchers who share similar backgrounds to the group being studied

+       Greater understanding of issues and more aware of ethical issues

-         Lack critical distance, biased observation

-         Participants may treat you as a friend

·        Outsiders

-         Researchers do not share similar backgrounds

+       Make conclusions insiders could not

-         Unable to understand experiences

-         Miss contextual aspects

Methods of data collection

•  Ethnography

-         Immersion in a particular group

-         Examine group for a period of time and observe behaviour

-         Issues: - Gaining access, Structured or unstructured?, Video record?

•    Meta data - Newspaper, Tv programmes, Radio programmes

•   Diaries and documentaries

-         Diaries are often used in health psychology

-         Documents = letter or autobiographies

•    Internet-mediated research -  Publicly available research

-         Issues: -   Ethics – Best to ask for consent

• ·        Naturally occurring data  - Produced without intervention of researcher

• Semi structured interviews

-         Most widely used in qualitative research

-         Issues to consider: -  What questions to use to get to research question?

                                         -  Who and how to recruit?

                                         -  Where to interview?

                                         -   How to record?

                                         -  How to transcribe?

•     Interview schedule

-         Guides the interview

-         Forces the researcher to explicitly think about what to cover

-         Potential difficulties

•   Types of questions

-         Descriptive -   Participants provide an account

-         Structural -  How the participant organises knowledge

-         Contrast -   Participants make comparisons between events and experience

-         Evaluative -   Feelings towards something

-         Probing -  Explain more

•      Questions to avoid: - Closed questions,  2 in 1 Questions,  Value laden questions, Jargon,  Questions that do not make sense

•  Focus groups

-         4-8 participants recruited under some remit

-         Informal group discussions that are focused

-         Facilitates personal discussions

-         Issues: -  How many pts?

                      -   Session prep

                      -    Ethics (Permission)

•  Ethical considerations

-         Access to people

-         Vulnerable pts

-         Skills to handle difficulties

-         Withdrawal

Thematic analysis

•  Simplest method of data analysis - Rich and detailed data
•  ‘Method for identifying, analysing, & reporting patterns in data.’
•  Themes -  Prevalent across the data set

                    -  Captures something important in relation to research question

TA steps:

1.      Transcription of data

2.      Make notes of 1^st impressions

3.      Initial coding of data

4.      Gradually move from descriptive to conceptual themes

•     Inductive

-         Data driven

-         ‘Bottom up’

-         No pre-existing coding frame

•     Theoretical

-         Driven by interest

-         ‘Top down’

-         Rich descriptive data and more detailed analysis

•       Reaction time

-         Stimulus, Perception, Processing, Produce response, Response

-         Provides an indirect measure of mental processes

-         Types of RT: -   Simple choice RT – 1 type of stimulus; 1 type of response

                               -   Choice RT – 1+ stimulus; each with its own response

                               -   Discriminant reaction time – 1+ stimulus; pts only responds to 1

•        Positron emission tomography (PET)

-         Uptake of glucose is greater in active brain areas

-         A radioactive substance is injected into the bloodstream

-         Active cells take up the substance readily

-         Use the subtraction method

•       Evaluation:

+        Creates high spatial res. Images

-         Poor temporal res.

-         Individual exposed to radioactive substances, so limits number of scans

•      Functional Magnetic resonance imaging (fMRI)

-         Active brain regions require more oxygen so blood flow increases

-         Oxygenated blood produces a different signal to deoxygenated blood

•    Evaluation:

+       Nothing is injected to pts

+       Better spatial res.

-         Poor temporal res.

•      Transcranial magnetic stimulation (TMS)

-         Used to establish cause and effect

-         Magnetic field applied to scalp, inhibiting neurons below

•       Electroencephalography (EEG)

-         Measures brain’s electrical activity

-         Electrodes attracted to scalp

-         Different waveforms produced reflect different functional states of the brain

•       Event related potentials

-         Interested in EEG waves in response to psychological events

-         Peaks observed in ERP

-         Researchers interested in how the task conditions affect the amplitude and latency

•    Evalution:

+       Excellent temporal res.

-         Poor spatial res.

•     Test items

-    Items take the form of closed questions or statement

-   Psychological scales provide quantitative data

-   Derived from theory and research

•   Response biases

-   Acquiescence bias = Tendency to respond positively to all items

-   Response bias = Tendency to give the same response to all items

Philosophy of Science

•      Science = a systematic enterprise that builds knowledge in testable explanations

-   Foundations in empiricism

•     The scientific method:

1.      Create hypotheses

2.      Test hypotheses with observations

3.      Accept/Refute hypotheses

4.      Integrate findings into theory

·        Carefully design the study and eliminate bias

Is Psychology a Science?

·        Psychology is made up of many different approaches e.g cognitive neuroscience and psychoanalysis

·        Quantitative and qualitative methodologies

·        Popper – science as deduction

Falsification

•  Falsification = you can only prove that a prediction of a theory is false, not true
•    For a theory to be scientific: -  Logically coherent, Testable via experience, Parsimonious, Components of theory must reflect real experience

* Why do we need theories?:

-   Can’t describe a notion unless a theory of underlying process is present

-   A theory that explains certain data is unsolvable

-   Can have alternative models that describe data

-   Theories allow us to make predictions or account for data

Parsimony & Paradigm shifts

·        Parsimony = Preference for the simplest theory

·        If the theory doesn’t capture data, there are infinite numbers of ‘add-on’.

·        Paradigm = A typical example/pattern of something

-   In a field of enquiry, a dominant way of doing + thinking about what is being explored occurs

-   Over time, we experience a paradigm shift

Replication

·        Attempt to replicate a study and produce the same finding

·        Fundamental principle of scientific developments

·        Replications studies relatively rare as harder to publish

Open Science

·        Many labs (2014)

·        Many labs 2 (2018)

·        Open science collab (2015)

-   Replications of 100 experimental and correlational studies

-   Designed a protocol to follow

-   Registered protocol of design, participant numbers and analysis

·        Open science is a movement to make parts of the research openly available

·        Sharing prevents making up theory to explain data, and allows comparisons of data with other sets

Publication bias

·        A lot of publishers have a bias for positive results

-   Exaggeration

-   Post-hoc

-   P-value fishing

-   Outliers

-   Non-publication

Confirmatory vs exploratory analysis

·        Exploratory

-   Not sure what you are doing or to expect

·        Confirmatory

-   Rationale and hypotheses known in advance