Research Methods Flashcards
Features of Science
Objectivity
Repeatability
Measurable and testable concepts
Trial and error
Hypothesis testing
Paradigm shifts
Aim
A general expression of what the researcher intends to investigate
Hypothesis
A predictive statement of what the research believes they will find
Directional (one tailed) - states whether changes will be greater or lesser
Non-directional (two tailed) - predicts difference / correlation
Independent Variable
The thing you change to see the effect on the DV
Dependent Variable
The thing you are measuring
Extraneous Variables
Other things that could affect the DV that you are not measuring
Confounding Variable
Other variable that have affected your results
Demand Characteristics
Any cue from the research situation or researcher that may reveal the aim of the the study and ptps react accordingly
= research issue
Pilot Study
A trial run of the research to work out any problems
Randomisation
The use of chance when designing an investigation to control for the effects of bias
= research techniques
Standardisation
Using exactly the same formalised procedure for all ptps in the same study
= research technique
Control Groups
Used for the purpose of setting a comparison
Act as a baseline to help establish causation
Single Blind
Ptp doesn’t know the aim of the study
Double Blind
Ptp and researcher don’t know the aim of the study
Independent Groups
One group condition A and the second group to condition B
Ptps should be randomly allocated to groups
No order effects
Less likely to guess aim
Ptp variables
More ptps used
Repeated Measures
Same ptps in all conditions
The order should be counterbalanced to avoid order effects
Ptps variables
Fewer ptps
Order effects
Ptps may guess aim
Matched Pairs
Two groups of ptps are used but are related to each other by being paired in Ptp variables that matter to the experiment
Ptp variables = matched
No order effects
Matching not perfect
More ptps
Lab Experiment
Controlled environment when EVs and CVs can be regulated
Ptps go to researcher
IV = manipulated and the effect on the DV recorded
EVs & CVs = controlled
Can be easily replicated
May lack gen
DCs
Field Experiment
Natural setting
Researcher goes to ptps
IV = manipulated and effect on DV = recorded
More natural environment
Greater external validity
More difficult to control CVs
Ethical issues
Natural experiment
Doesn’t manipulate IV - would have varied even if the experimenter not interested
DV may = naturally occurring
Ethical
Greater external validity
Natural event may only occur rarely
Ptps not randomly allocated
Quasi Experiment
IV based on pre-existing difference e.g., age / gender
No one has manipulated the variable
DV may be naturally occurring or be measured by the experimenter
High control
Comparison can be made between ppl
Ptps not randomly allocated
Causal relas not established
Population Vs Sample
Population = large group of ppl the researcher is interested in investigating
Sample = usually not possible to include all population in study so smaller group is selected
Opportunity Sample
Ppl who are the most available and willing at the time
= quick but biased
Volunteer Sample
Ptps select selves for study
Ptps = willing but is likely to = biased
Random Sample
Every person in the target population has an equal chance of being selected e.g., names in hat
= unbiased but not necessarily representative
Systematic Sample
Ptps are selected using set pattern e.g., every ninth person from list of target population
Is unbiased but = time and effort
Stratified Sample
Ptps are selected according to frequency in the target population
E.g., a strata (sub-group) is identified and then a random sample selected from each
= representative but strata won’t reflect all personal differences
Informed Consent
Ptps should be able to make informed judgement about whether to take part
Too much info may affect behaviour so alternative forms of consent = ask similar group they consent to being deceived / consent after
Deception
Deliberately misleading or withholding information so consent is not informed
Ptps should be debriefed to be told true aim, details that were not given and what their data will be used for and their right to withdraw
Protection from Harm
Ptps should be at no more risk that they would be in everyday life
Should be given the right to withdraw at any time and reassured their behaviour was typical / normal when being debriefed
Counselling should be provided if have been distressed
Privacy / Confidentiality
Ptps have right to control info about them is that is evaded confidentiality should be rightly respected
If personal details are held they must be protected - usually not taken - refer to using numbers, initials or false names
Ptps personal data cannot be shared with other researchers
Correlation Vs Experiment
Experiment = researcher manipulates the IV and records the effect on the DV
Correlation = no manipulation of variables and so cause and effect cannot be demonstrated - EVs not controlled
Association
Correlations illustrate the strength and direction of an association between two co variables
Plotted in scattergram
Types of Correlation
Positive = as one variable increase the other increases
Negative = as one variable increases the other decreases
Zero = no rela between two variables
Correlation Evaluation
= useful starting point for research
Is relatively economical
No cause and effect
Method used to measure the variables may be flawed
Naturalistic Observation
Takes place where the target behaviour would normally occur
High external validity
Low control
Controlled Observation
Some control/ manipulation of variables including control of EVs
Can be replicated
May have low external validity
Covert Observation
Ptps are unaware they are being studied
Reduced DCs
Ethically questionable
Overt Observation
Ptps aware are being studied
More ethically acceptable
DCs
Participant Observation
Researcher becomes part of the group they are studying
Leads to greater insight
Loss of objectivity
Non-Participant Observation
Researcher remains separate from the group
More objective
Loss of insight
Behavioural Categories
The target behaviour to be observed should be broken up into a set of observable categories
Similar to operationalisation
But difficult to make clear
Dumped behaviours go unrecorded
Time Sampling
Observations made at regular intervals eg., once every 15 seconds
Reduced number of observations
May be unrepresentative
Event Sampling
A target behaviour/ event is recorded every time it occurs
May record infrequent behaviour
Complex behaviour is oversimplified
Questionnaires
Made up of pre-set list of questions (or items) the Ptp responds to
Can be used as part of an experiment to assess the DV
Can be distributed to lots of ppl
Respondents may be willing to open up
Responses may not always be truthful
Response bias
Designing Questionnaires
Write good questions, avoid jargon, avoid double-barrelled questions and avoid leading questions
Closed questions = respondent has limited options = easier to analyse but = restrictive
Open questions = respondent provide own answers expressed in words, respondents not restricted, difficult to analyse
Interviews
Face to face interview between interviewer and interviewee
Structured interview
List of pre-determined questions asked in a fixed order
Easy to replicate
No elaboration
Unstructured interview
There are no set questions
= general topic to be discussed but the interaction is free-flowing and elaboration is encouraged
Is greater flexibility
Difficult to replicate
Semi-structured interviews
List of questions that have been worked out in advance but interviewers are free to ask follow up questions when appropriate
Designing an interview
Have a schedule
Have a quiet room
Build rapport and make sure you abide to ethics
Qualitative Vs Quantitative
Qualitative: non-numerical expressed in words e.g., extract from diary = represents complexities, less easy to analyse
Quantitative: numerical e.g., reaction time / number = easier to analyse but oversimplifies behaviour
Primary Data
First hand data collected for the purpose of the investigation
Fits the job
Requires time and effort
Secondary data
Collected by someone other than the person who is conducting the study
= inexpensive
Quality may be poor
Meta-Analysis
Secondary data that includes combing data from a large number of studies
= increases validity of conclusions
Publication bias
Measures of Central Tendency
Mean: arithmetic average = sensitive, but may be unrepresentative
Median: middle value (if two middle calculate the mean) = unaffected by extreme scores, less sensitive
Mode: most frequent/common value = relevant to categorical data, an overly simple measure
Measures of Dispersion
Range: difference between highest and lowest value = easy to calculate, doesn’t account for the distribution of scores
Standard Deviation: measure of average spread around the mean, larger = m spread out, more precise than the range but may be misleading
Tables
Raw scores are displayed in columns and rows
Summary paragraph beneath explains the results
Bar Chart
Categories along x axis and frequency on y axis
Height of each column represents the frequency for the item
Histogram
Bars touch - data = continuous not discrete
Is a true zero
Line Graph
Frequency on one axis and categories on the other
Often shows how something changes over time
Scattergram
Used for correlational analysis
Each dot = one pair of related data
Both sets must be continuous
Normal Distribution
Symmetrical bell-shaped curve
Most ppl are in middle
Mean, mode and median = in same area
Skewed Distribution
Lean to one side because most ppl are at lower/higher end of distribution
Positive Skew: most if distribution concentrated towards left of graph leading to long tail on the right
Negative Skew: most of the distribution is concentrated towards right meaning long tail on the left
Significance
Difference/association between two sets of data is greater than what would occur by chance
Probability
A numerical measure of the likelihood that a certain event will occur
Acceptance level of probability is p_<0.05 = 5% chance that the results are down to coincidence
Calculated value
Value calculated through stats test
Compare to critical value to decide whether is significant or not
Critical value
Given in table of critical values
Sign Test Calculation
The score for condition B is subtracted from Condition A to produce a sign of difference (+ or -).
Do this for each Ptp
The total number of pluses and the total number of minuses should be calculated
Ptps who achieved the same score in both conditions should be disregarded and deducted from the N value
The S value is the total of the less frequent sign
Critical Value: If the critical value is equal to or less than you S value then it is significant.
Peer Review
B4 publication all aspects if investigation are scrutinised by experts in the field
Should = objective and unknown to researcher
Aims = allocate research funding, validate quality and relevance of research, suggest improvements and amendments
Protects quality of published research
May be used to criticise rival research
Publication bias
Ground breaking research may be buried
Psychology and the Economy
Findings can benefit economic prosperity
Research into attachment and the role of the father - promote more flexible working arrangements in the family meaning modern parents are better equipped to contribute more effectively to the economy
Mental Illness Treatment- 1/3 of all days off are caused by MI - research means patients can have quicker diagnosis and = therapies and drugs to help manage effectively and return to work
Open Questions used when … (questionnaires)
Need opinions and attitudes
Not time bound
Don’t know expected answer
Literate ptps
Close questions used when … (questionnaires)
Quick
Generates number data for stats
Opportunity samples
Public places - ppl = rushed
Structured questions used when … (interviews)
Time is short
Opportunity samples
Need number data
Uncontroversial topic
Adults
Unstructured questions used when … (interviews)
Children and vulnerable groups
Secretive/ unknown behaviour
Traumatic/ upsetting topics
Personal information
Disadvantages of Self Report
Validity issues: not proper interval scale as spaces between numbers = not known/ individually devised, only as good as the questions
Investigator bias
Social desirability
DCs
Content Analysis
An indirect way of studying behaviour by looking at what humans create e.g., posters etc
Involves generating themes of codes to analyse the behaviour behind the material
Normally qualitative data
Case Studies
Single Ptp or small group = single place / organisation = in-depth
Describes but doesn’t explain
Not just a snapshot = more detail
Relies on less scientific methods e.g., interviews and self reports - bias = possible
Not easily generalisable
Controls
Keep the DV uncontaminated so is only changed by IV
Environment - stop distractions and interruptions
Procedures - anonymised data, standardisation of materials and actions, counter balance effects
Design - randomisation of selection to conditions
Investigator bias / effect
Any influence the investigator / experimental situation has on ptps - might alter the results & is not intended
Ptps might not act naturally if know they are being watched
Investigators mood / demeanour may encourage / discourage ptps
Single blind / standardisation can counter
Demand Characteristics
Can be dealt with using independent groups, or not telling true aim but = ethical concerns
Social Desirability
Changing behaviour for fear of being judged
Not all behaviour is socially sensitive e.g., memory digit spans
Includes topics such as: sexual behaviour, criminal behaviour, parenting and relationships
Can be dealt with/ by reassuring info is confidential and anonymous
Issues w/ research
Validity, internal and external
Reliability, internal, external & inter-rater
DCs and investigator effects
Ethics’s including the British Psychological Society’s code of ethics, ethical issues in the design and conduct and dealing w/ ethical issues
Internal Reliability
Split half method used to ensure reliability of written ability tests such as IQ
Makes sure = consistent throughout
External Reliability
To improve need to repeat tests over and over again to see what happens to results - using same ptps = test re test
Observers collecting data need to be reliable and so have to = trained and use checklists and coding systems to held = inter-rater reliability
Internal Validity
Do the tools used to measure the IV & DV / co variables accurately measure them
E.g., thermometer accurately measures temperature and nothing else
External Validity
A tool is also not valid if it only measures things in labs and not in real life settings = lack of external validity
Checking validity
Compare results with previous theories = construct validity
Comparing results with previous established tests/ research = concurrent validity
Asking informed people if it looks right = face / content validity
Type I Error
A false positive and occurs more when a lot of scope for chance - 10% ( make it too easy to show a significant result)
Type II Error
A false negative and occurs when you don’t allow much chance - 1% ( make it too hard to show a significant result)
Implications of Research for the Economy
Treatments could = waste of money if not effective / good quality (valid)
NHS is funded by taxpayer money
Anything leading to a loss of productivity in the workplace = loss of money
Good research into wellbeing could save money
Nominal Data
Simple frequency headcount (number of times something occurred)
Found in discrete categories (something can only be in one category)
Most simple data
E.g., number of people who did / didn’t help in an emergency
Ordinal Data
Measurements that can be put in order, rank / position
Intervals between are unknown e.g., how far away 2nd was from 1st in a race
Interval Data
Measurements on a scale - intervals are known and equal
Can go into negative values
= most precise
E.g., temperature
Ratio Data
Measurements in a scale
Intervals are known and equal
Has true zero - can’t go into negatives most precise
E.g., length or time
Inferential Stats Test
Used to find out chance of chance - is the result significant?
Used w/ repeated measure design and nominal data
Result is the sum of the least frequent sign
Theory Construction
Scientific process starts with observation
Induction model states scientists develop a testable hypothesis
Hypothesis = tested leading to new questions & new theories e.g., newtons laws
Deduction model involves reasoning & theory 1st & finding instances to support e.g., Darwin
Generation of Laws & Principles
Popper (1969) argued theories should = falsifiable
Abstract theories = impossible to prove well though empirical research
Claimed theory = scientific if falsifiable - can be proved wrong Freud can’t be
Kuhn (1962) argued science should have a paradigm - should have key assumptions - a paradigm
= reductionist idea m complex can = reduced to smaller level - bio approach
Paradigm Shift
Kuhn (1962) argued = 3 stages to science:
Pre-science - isn’t science bc = lots of competing approaches
Normal science - an overall paradigm = estab to which = gen agreement & appropriate research method
Scientific rev - research evidence that challenges the current paradigm so returns to normal science stage
So according to Kuhn psych was not a science in the 60s but is now
Sign test
N - any that scored same score so = 0
Chi-squared
Df = (rows-1) x (column -1) = usually 1
Wilcoxen
N - any scores that = same in dif condition
Mann Whitney
Na is number of across ptps in condition A and down is Nb is the number of ptps in condition B
Related T test
N-1
Unrelated T test
NA + NB (number of scores in 1st condition + number of scores in 2nd condition)
Spearman’s
Number of sets of scores
Pearson’s
N - 2
