Research Methods Flashcards
What is an aim
A statement of a study’s purpose
3 types of hypothesis
One tailed/directional, two tailed/non-directional and null
Directional Hypothesis
States the difference between conditions
Non directional hypothesis
States there will be a difference but doesnt say what the difference will be
Null hypothesis
There will be NO significant difference between the conditions
Independent Variable
The variable we’re changing/ manipulating
Dependent Variable
The variable we’re measuring
Control
The extent to which any variable is held constant or regulated by a researcher.
Random allocation
Everyone has an equal chance of doing either condition
Counterbalancing
Half the participants participate in condition A before condition B and vice versa. (overcomes order effects)
Randomisation
Materials are presented in a random order
Standardisation
Everything should be as similar as possible for all participants
Extraneous variables
Variables other than the IV that could influence your results
Confounding variables
A type of extraneous variable that is related to both the independent and dependent variables
Ethical guidelines
Standards of behaviour, promoting fairness, protecting rights, and minimising harm.
Informed consent
Knowing aims and giving your permission to take part in the study
Deception
Deliberately misleading or withholding information
Right to withdraw
Being able to leave when desired
Confidentiality
Details should be kept private
Protection from harm
No more harm than daily life
3 types of experimental design
Repeated Measures
Independent groups
Matched pairs
Independent Groups
There are 2 separate groups of participants. One takes park in Condition A and the other in B
Independent Groups A+W
Fewer demand characteristics
No order effects
But more participants needed
Individual differences
Repeated Measures
One group that takes part in both conditions
Repeated Measures A+W
No individual differences as the same person does both conditions
Demand characteristics
Order effects
Matched Pairs
Two groups and they are matched into pairs for certain qualities such as age or intelligence. One does Condition A and the other B
Matched Pairs A+W
No order effects
Controls for individuals differences
Difficult to match people perfectly
Costly and time consuming
Types of experiments
Laboratory
Field
Natural
Quasi
Field Experiment
Take place outside of the lab but still manipulates IV
Field Experiment A+W
Less Artificial
Avoids participant effects producing more natural behaviour
Less easy to control extraneous variables
Ethical Issues (pps unlikely to know they are being studied)
Laboratory Experiment
Controlled artificial environment where IV is manipulated
Laboratory Experiment A+W
Controlled environment
Minimises extraneous variables
Artificial environment
Pps may behave differently due to environment
What is meant by the term ‘double blind’?
Neither the participants or the researchers are aware of the aims of the investigation
What is meant by the term ‘single blind’?
Participants aren’t aware of the condition they are in
Attempts to control for the confounding effects of demand characteristics
Natural Experiment
Natural Environment. IV manipulated taking advantage of a naturally occuring event
Natural Experiment A+W
High ecological validity
Few ethical issues
Many extraneous variables
Naturally occuring events are infrequent limiting research opportunity
Quasi Experiment
The IV is a naturally existing characteristic between people and hasn’t been changed by anyone or anything
Quasi Experiment A+W
Done in labs so high in control
‘Real’ problems can be studied
Pps can’t be randomly allocated to conditions so there may be confounding variables. Meaning we can’t say cause and effects
What are behavioural categories?
Categories defined by the researcher to observe during the experiment
2 types of sampling
Event
Time
Event Sampling
Counting the number of times a certain behaviour occurs
Time Sampling
Recording behaviours in a given time frame
Controlled Observation
When the researcher has some measure of control over the environment
Controlled Observation S+W
Control over extraneous variables
Easy to replicate
Can’t be applied to real life setting
May be subjective towards what the researcher wants to see
Naturalistic Observation
Studying behaviour in a natural setting where everything has been left as it is normally
Naturalistic Observation S+W
High ecological validity
Natural Environment- generalised to everyday life
Replication is difficult
Uncontrolled extraneous variables
Covert Observation
The participants aren’t aware that they are being observed
Covert Observation S+W
No demand characteristics
Ethical Issues
Overt Observation
The participants are aware that they are being observed
Overt Observation S+W
Less ethical issues
Might be demand characteristics as they know they are being watched
Participant Observation
The observer acts a part of the group being watched
Participant Observation S+W
Experience situation and increases validity
Lose objectivity
Difficulty in recording observations
Non-participant Observation
The experimenter does not become part of the group being observed
Non-participant Observation S+W
More ethical, more objective
Less insight
Not experiencing the same things
Structured Observation
The researcher determines precisely what behaviours are to be observed and uses a standardised checklist to record the frequency with which they are observed within a specific time frame
Structured Observation S+W
Easy to gather relevant data because you know what you are looking for
Interesting behaviours could go unrecorded because they weren’t pre-defined as important
Unstructured Observation
The observer recalls all relevant behaviours but has no system
Unstructured Observation S+W
Interesting behaviours don’t go unnoticed
Difficult to gather relevant data because you don’t know what you are looking for
What is inter-rater reliability?
The test should give consistent results regardless of who administers it
This can be assessed by correlating the scores each researcher provides and compare. There should be an 80% agreement
How can inter-rater reliability be improved?
Offer a chance to discuss difficult issues or problems and monitor the quality of the data collection over time
Structured Interview
When the questions are decided in advance
Structured Interview S+W
Can be easily repeated (standardised questions)
Requires less skill than unstructured
Can be interviewer bias
Data collection will be restricted
Unstructured interview
When the interviewees answers to questions guide subsequent questions
Unstructured Interview S+W
Detailed and in depth information obtained
Insight into feeling and thoughts
Affected by interviewer bias
Hard to analyse answers
Semi-Structured Interview
Combination of structured and unstructured
Qualitative data
In-depth Information in a written form - words, texts, ideas
Quantative Data
Information that can be reduced to number and quantities
Reliability
Overall consistency of a measure
Internal Reliability
The extent to which a test is consistent within itself
Split-Half Method
Methods ensuring reliability
Compare an individual’s performance on two halves of a test
Test-retest method
Methods ensuring reliability
A person repeats a test a month or so after doing the test the first time
Concurrent Validity
Results from a new test can be compared to a previously well-established test
Predictive Validity
If diagnosis leads to successful treatment then the diagnosis is seen as valid
Temporal Validity
Assesses to what degree research findings remain over time
Content Validity
Involves asking experts in the field to check the content of the study
Ecological Validity
Generalisable to real life settings- generalising findings from one setting to other settings
Population Validity
Whether you can reasonably generalise the findings from your sample to a larger group of people
Bar Chart and why its different to a histogram
Used to present discrete data that are placed into categories
Columns do not touch and have equal width and spacing
Histogram
Used to represent data on a continuous scale
What is correlation analysis?
When two or more variables are measured in order to see if there is a relationship
(positive, negative or no correlation)
Correlation coefficient
Number between 1 and -1 telling us how strong the correlation is
Types of Correlation
Positive - both variables increase
Negative - One increases the other decreases
No correlation- no relationship
Positive and Negative Skew
Positive -more scores on the lower end of the data set
Negative - more scores on the higher end of the data set
What is an experimental group?
The participants are the experiment who the researcher is testing
What is a control group?
The other condition where participants are taking part in the experiment, but no manipulation is used
What are demand characteristics?
Participants may have determined the aims of the study (may act deliberately to please the researcher
How can demand characteristics be controlled?
Counterbalancing / randomisation
Name 2 self report methods
Questionnaires and Interviews
Interview
Used to gather qualitative data
Advantages and disadvantages of Interviews
Can get rich and detailed data
Time consuming and impractical
What are investigator effects?
Anything the researcher does which can effect how the participant behaves
What is researcher bias and how can it be avoided?
Researchers expectations can influence how they design their study.
Research assistant conducts the research using standardised procedures
What is content analysis?
Research analysing secondary data and data you’ve collected
Content analysis S+W
Inexpensive
Ethics - participants not directly involved
Subjectivity
Data analysis is time consuming
What is thematic analysis?
Making summaries of data and identifying key themes and categories
Mean
How its calculated and S+W
Adding up all the numbers and divide by the number of data items
Represents all the data
Effected by extreme values
Median
How its calculated and S+W
Middle value in an ordered list
Not affected by extreme values
Exact values not represented
Mode
How its calculated and S+W
Most common data item
Not affected by extreme values
Sometimes there are too many modes
Range
How its calculated and S+W
Difference between top and bottom values
Easy to calculate
Affected by extreme values
Standard Deviation
How its calculated and S+W
Measure of the average distance between each data item above and below the mean
Precise measure of dispersion
Affected by extreme values
Quantitative Data
Numbers
Primary Data
Data collected first hand by the researcher
Secondary Data
Data collected from another source
Pilot Study
A small scale investigation that takes place before the real investigation is conducted
Questionnaires
Set of questions used to assess a person’s thoughts and experiences
Questionnaires S+W
Produces quantitative data
Easily repeatable
Answers maybe chosen that don’t represent real thoughts
Poor/vague questions lead to incorrect results
Case study Definition + S+W
Intense description of a single individual case
Rich data, unique cases studied in detail
Can’t be generalised
What is a sample?
The people the researcher actively use in the research
5 sampling techniques
Opportunity
Volunteer
Random
Stratified
Systematic
Random Sampling
Each person has an equal chance of being selected. Chosen by a computer random generator
Random Sampling S+W
Fair
More likely to be representative
Can be biased if the sample is too small
Volunteer
Sampling method
People who are interested apply to be in the research
Volunteer S+W
Sampling method
Convenient and ethical
Sample is biased because the participants are likely to be more motivated (volunteer bias)
Opportunity
Sampling Method
The participants available at the time to take part in the research
Opportunity S+W
Sampling Method
Easy and quick method because you just use the first participants you find
Biased as the sample is drawn from a small part of the target population
Stratified
Sampling method
Selective people from every portion of your populations - in the same proportions
Stratified S+W
Sampling method
More representative
Time consuming as all participants need to be assessed and categorised
Systematic Sampling
Selecting every nth name from a list
Systematic Sampling S+W
Avoids bias as there is no control over who is being selected
Not necessarily representative if the pattern used for the sample coincides with a pattern in the population
Nominal
Levels of measurement
Data represented in the form of categories
Ordinal
Levels of measurement
Data which is ordered in some way
Interval / Ratio
Levels of measurement
Numerical scales that include units of equal precisely defined size
What part of the research report should include the psychologists hypothesis
The introduction
What are the 5 ethical guidelines?
Informed consent
Deception
Right to withdraw
Confidentiality
Protection from harm
External Validity
The extent the results of the study can be generalised to others
Internal Validity
The study measures or examines what it claimed to measure or examine
4 Aims of Peer review
- Assess the appropriateness of the research to the research topic/aim
- Check the validity of the findings
- Judge the significance of the research
- Check that the research is original and has not been plagiarised
Role of Peer Review
It helps to determine if the research can be deemed scientifically acceptable
Peer review is an independent assessment carried out before the research is published by other experts in the field
It is completed independently and usually anonymously
3 Types of Peer review
- Single Blind (researcher’s name is not revealed to the reviewers)
- Double Blind (researcher and the reviewers are anonymous to each other)
- Open Review (researcher and reviewer known to each other)
Evaluation of Peer review
- Reviewers are especially critical of research that contradicts their own
- Reviewers tend to be established scientists and are more likely to publish research that ‘fits’ with current opinions
- This could slow down the rate of change
Implications of research for the economy
Psychological research involves real people, investigating real behaviours which have real consequences
Implications of research for the economy evaluation
Absences at work cost the economy an estimated £15 billion per year
-a third caused by depression, anxiety and stress (the telegraph 2014)
-CBT/SSRIs and anti anxiety drugs have allowed people with mild mental health disorders to return to work and access medical treatment
Title
Should say what the study is about and include the independent and dependent variables
Abstract
Short summary that includes all the major elements: the aims and hypotheses, method/procedure, results, and conclusions
Introduction
- Literature review of the general area of investigation
- Look at the relevant theories, concepts, and studies
- Should follow a logical progression, beginning broadly then becoming more specific until aims & hypotheses are presented
Method
Footnote ##
DS APE
Should include sufficient detail so that other researchers are able to replicate the study
* Design - (eg independent groups, naturalistic observation etc) and reasons/justification given for choice
* Sample - amount of p’s, biographical/demographical info (avoiding compromising anonymity), sampling method, target pop.
* Apparatus/materials - detail of any assessment instruments used & other relevant materials
* Procedure - list of everything that happened, including briefing, standardised instructions, and debrief
* Ethics - how these were addressed within the study
Results
Summary of key findings from investigation
* Descriptive statistics eg tables, graphs, charts, measures of central tendency, and measures of dispersion
* Inferential statistics eg reference to choice of statistical test, calculated & critical values, level of significance, final outcome (which hypothesis rejected)
* Any raw data & calculations appear in appendix rather than main body of report
Conclusions/Discussion
Summary of findings in verbal form
* Relationship of results to previous research (mentioned in intro)
* Limitations of study & suggestions of how these might be addressed in a future study
* Wider real-world implications of research
When to use the Sign test
The assumptions of the sign test:
• we are looking for a difference not an association
• we need to use a repeated measures design
• we need data that is organised into categories- nominal data
Using the Sign Test
- Calculate difference between the two sets of data (just state +/-/0)
- Add up positives and negatives and ignore zeros (eg -=6 and +=2)
- The less frequent sign is your S value (2 in this case)
- The N is number of participants minus the zeros (eg 10 participant but 2 had no significance and 2 were positive so N=6)
- Add S+N (2+6)
- Compare S with the critical value using N=8 (level of significance assumed to be 0.05)
- Critical value needs to be greater than S for it to be considered significant
Levels of significance and probability
• all statistical tests employ a significance level
• you can reject the null hypothesis and claim you have found a significant difference/correlation
• However: there is a 5% probability that the observed effect occurred by chance
The Rule of R
Statistical tests (eg Spearman’s Rho) with an R in the name are those whose calculated value must be equal to or more than the critical value to be significant
(Significance and probability) How to use the table
-one-tailed or two-tailed: probability levels double when two-tailed tests are being used for a more conservative prediction
-the number of participants: n value in the table/ however sometimes use the degrees of freedom (df)
-the level of significance: p value
Type I errors
When a true null hypothesis is rejected, meaning a conclusion is drawn that there is a significant effect when there isn’t one
Type II errors
When a false null hypothesis is not rejected, meaning a significant effect is missed when it actually exists
Difference between Type l and Type ll errors
Type l is a false positive
Type ll is a false negative
When to use Mann Whitney test
- a test of difference
- testing between independent groups
- using ordinal data - scores
When to use Wilcoxon
a test of difference for related data that is ordinal
experimental design can be repeated measures or matched pairs
calculated value is the sum of the numbers of the less frequent sign (+ or -)
Chi Squared
-used when
• test of difference or association/correlation
• the data is nominal and recorded as a frequency
• independent groups design- unrelated data
uses contingency table
looks for observed frequency
Paradigm shifts
A paradigm shift, as identified by Kuhn (1962), is an important change in the basic concepts and experimental practices of a scientific discipline. It is a change from one way of thinking to another
Paradigms
The basic assumptions, ways of thinking, and methods of study that are commonly accepted by members of a discipline or group.
Theory constructs and hypothesis testing
• What is a theory?
- A set of general laws or principles that have the ability to explain
particular events or behaviours
- Theory construction occurs through gathering evidence via direct
observations (the empirical method)
• What is a hypothesis?
- Prediction based on theory
- Scientifically tested
Falsifiability
- Popper (1934) ‘genuine scientific theories should hold themselves up
for hypothesis testing and the possibility of being proven false’ - Even ‘proven’ research is not true, it has just not yet been proven false!
Replicability
- Trusted findings should be repeatable across a number of contexts
and circumstances - Validity and reliability
Objectivity and the empirical method
• ‘critical distance’
• Controlled laboratory studies
• Experience- knowledge is determined only by experience and sensory
perception (Locke)
The case for psychology being a science
• Key findings in psychology are counter-intuitive and not predictable
• Psychology gained credibility by adopting scientific methods of
enquiry
• Practical application and challenged/ modified human behaviour
The case against psychology being a science
• Methods can be subjective, non-standardised and unscientific
• Universality??
• Based on inference rather than objective measurement
Spearman’s Rho
• test of correlation
• used for two sets of values at an ordinal or one interval one ordinal
Pearson’s R
• Interval data and a test of correlation
• the closer r is to -1 or +1 the stronger the relationship
Related T-test
• repeated measures design or matched pairs
• test of difference with interval or ratio data
• e.g. weight in kg
N-1
Unrelated T-test
• independent groups
• interval data is needed
• test of difference
Degree of Freedom= N1+N2-2
Statistical Tests table
3 Types of Data (NOI)
Nominal - can only be categorised
Ordinal - can be categorised and ranked
Interval - can be categorised, ranked, and evenly spaced
3 Statistical Tests that use Repeated Measures/ matched pairs design
Sign, Wilcoxon, Related T tests
3 Statistical tests that use independent groups
Chi Squared, Mann Whitney, Unrelated T tests
3 Statistical Tests that use correlation or association
Chi Squared, Spearmans Rho, Pearson R
Statistical Tests that use Nominal data
Chi Squared and Sign
Statistical Tests that use Ordinal data
Mann Whitney, Wilcoxon and Spearmans Rho
Statistical Tests that use Interval
Unrelated, related and pearson r