Use LEFT and RIGHT arrow keys to navigate between flashcards;
Use UP and DOWN arrow keys to flip the card;
H to show hint;
A reads text to speech;
185 Cards in this Set
- Front
- Back
- 3rd side (hint)
What is the observational stream steps in research |
Ask Observe informally Choose measures Choose recording method Collect and analyse data Publish Ask |
|
|
What's the experimental stream steps in research |
Ask Hypothesise Predict Design Experiment Analyse Interpret (Hypothesise/publish) |
8 |
|
What are the 2 definitions of measurements |
Operational Ostensive |
2 os |
|
What are operational definitions |
Specify the physical requirement for coding a behaviour (e.g. press a lever) |
|
|
What are ostensive definitions of measures |
Provide examples through pics or diagrams, along with written descriptions of the behaviour of interest (e.g. coordinated play vs solitary play) |
|
|
What's an ethnogram |
Complete behavioural repertoire and Coding schemes used in specialised studies of a subset of a species or groups behaviour |
|
|
5 types of measures |
Latency Duration Frequency Rate Proportion |
LADFRAP |
|
What are the 4 scales of measurement |
Nominal (categorical) Ordinal (ranking) Interval (0 is arbitrary) Ratio-interval (continuous) |
|
|
What are the recording methods 4 sampling rules (names) |
Ab libitum Focal sampling Scan sampling Behaviour sampling |
|
|
What is ab libitum sampling |
Writing down any important events from sample (rare) |
|
|
What is the potential bias of ab libitum sampling |
Tend to miss events of short duration and underestimate the contribution of smaller, less conspicuous subjects |
|
|
What is focal sampling |
A specific individual is isolated for observation |
|
|
What is a potential bias of focal sampling |
Can be large if focal subject seeks privacy for some kinds of behaviour |
|
|
What is scan sampling |
A number of individuals (typically big group) is sampled often in rapid succession |
|
|
What is the potential bias of scan sampling |
Rare events of short duration tend to be underestimated |
|
|
What is behaviour sampling |
All occurrences sampling |
|
|
What is the potential bias of behaviour sampling |
Overestimate conspicuous events |
|
|
Which sampling method is preferred |
Ab libitum |
|
|
What are the 2 recording methods recording rules |
Time sampling Continuous recording |
|
|
2 types of time sampling |
Instantaneous sampling (e.g. record every 10 mins) One-zero sampling |
|
|
Potential bias of time sampling |
Can underestimate rare behaviours of short duration |
|
|
What is a potential bias of continuous recording |
Underestimate long duration behaviours as these are more likely to be truncated by the end of the recording session |
|
|
Principles of coding scheme designs |
Mutually exclusive Exhaustive |
|
|
2 measures for consensus of inter-observer reliability |
Percent agreement Cohens kappa |
|
|
2 measurements for consistency of intra-observer reliability |
Correlation coefficient Cronbachs alpha |
|
|
Equation of cohens kappa |
Probability (observed) - probability (expected) / 1 - Probability (expected) |
|
|
Problem with percent agreement |
Does not correct for agreement by random chance |
|
|
How to work out expected proportion of agreement |
P(both say yes) = prob 1 says yes × prob 2 says yes P(both say no) = prob 1 says no × prob 2 says no P(expected) = p(yes) + p(no) |
|
|
What are the cut offs for kappa (Landis and Koch) |
<0 = no agreement 0-.2 = slight .2-.4 = fair .4-.6 = moderate .6-.8 = substantial .8-1 = near perfect |
|
|
Issue with correlation coefficients for intra rater reliability |
Does not take into account variance between coders |
|
|
What are the 3 things that maks a good questionnaire |
Discrimination Validity Reliability |
|
|
What is discrimination in a questionnaire |
Whether the questionnaire can successfully tell people apart on the construct |
|
|
What are the 3 types of validity needed for a good quesionnaire |
Content Criterion Factorial |
|
|
What is content validity |
Items relate to the contruct of interest Are comprehensive (cover entire construct) Do not overlap |
|
|
What is criterion validity |
Items do in fact measure the contruct of interest in a meaningful way |
|
|
What is factorial validity |
Does the factorial structure you hypothesized correspond to reality |
|
|
What are the numerical response scales used in questionnaires |
Ranking scales Rating scales Semantic differentials |
|
|
What are semantic differentials |
Where each response on the scale has a different semantic meaning E.g. For me, binge drinking would be... Pleasant 1-7 unpleasant Wise 1-7 unwise |
|
|
What is the likert scale an example of |
Rating scale with a midpoint |
|
|
What did Schwarz 1996 show about response scales |
The way your response scale is scaled has an effect "How successful would you say ypu have been in life?" 0-10 scale, 34% gave value 0-5 -5-5 scale, 13% gave value -5-0 |
|
|
5 wording effects to take into account when creating a questionnaire |
Ambiguity Leading questions Double negatives Double barrelled questions Acquiescence bias (tendency to say yes) - reverse statements |
|
|
What is nomothetic |
Relating to the discovery of general laws |
|
|
What is idiographic |
Relating to the study of particular facts or processes |
|
|
Semi-structured interview structure/principles |
Into/explanation Questions in funnelling (gen to specific) Responses, follow ups and prompts Reflection |
|
|
What are 2 techniques to encourage disclosure |
Express ignorance - encourages state the obvious Ask for concrete details |
|
|
4 things to avoid in interviews |
Double-barrelled questions Introducing assumptions (leading qs) Complex or jargon words Include double negatives |
|
|
4 features for good rapport |
Relaxed Atmosphere of openness Trust Giving space |
|
|
Before analysing interviews, what must you do |
Transcribe it |
|
|
What are the 2 types of coding |
Selective coding (identify relevant material) Complete coding (line by line) |
|
|
What is verbatim transcription |
Written speech exactly the way its spoken (e.g. with um's) |
|
|
2 methods of selective coding analysis |
Top down (a priori categories - content anaysis) Bottom up (thematic analysis) |
|
|
What are the 5 standard critiques to interviews |
Over reliance of self report for behaviours Makes comparability across cases difficult Practical: time consuming Interviewer effects Humanistic interviews take data at face value (window onto mind) - might differ between people |
|
|
What are the 2 interviewer effects |
Features of the interviewer Actions of the interviewer (responses to what the interviewee says - response bias) |
|
|
Who gave the "humanistic" interview critique |
Potter and Hepburn 2005 |
|
|
3 other research designs that interviews can be applied to |
Focus groups Ppt observation (ethnography) - framework for being an insider Piloting for survey questionnaire studies |
|
|
5 examples of types of qualitative analysis |
Thematic analysis Interpretative Phenomenological Analysis (IPA) Ground Theory Discourse analysis Conversation analysis |
|
|
What is a theme |
A pattern of meaning Captures something important about the material Represents some level of patterned response within the dataset Emphasis on meaning not prevalence |
|
|
Who wrote the most famous article of thematic analysis |
Braun and Clark |
|
|
What are the 2 thematic analysis drives |
Theory Data |
|
|
What are Braun and Clarks 6 steps for thematic analysis |
1. Familiarise self with material 2. Initial coding 3. Searching for themes 4. Reviewing themes 5. Defining and naming themes 6. Provide commentary |
|
|
What are the 2 types of coding |
Selective coding - identify relevant material Complete coding - line by line |
|
|
What are the 3 things you may do to themes when reviewing them |
Drop Merge Split |
|
|
What did Howitt suggest to be an issue of thematic analysis |
At worst, the analyst "see's" 5 or 6 themes, then finds examples for them Implies themes are there without researcher input, no justification/explanation, no effort/criteria |
|
|
What does IPA stand for |
Interpretative Phenomenological Analysis |
|
|
Main way Interpretative Phenomenological Analysis differs to thematic analysis |
A type of thematic analysis that makes a number of psychological aasumptioms |
|
|
What are the 2 assumptions in IPA |
1st = people interpret the world, studies = people's interpretation of their world 2nd = researchers interpet world too, so interpret people's interpretations - REFLEXIVITY (self-awareness in our activity) |
|
|
What is the sample sizes for IPA |
6-8 standard 4-5 acceptable 1 is allowable but not advisable |
|
|
How is the data gathered usually for IPA |
Semi structured interviews Diaries can be used too |
|
|
Is sample type or sample size more important in IPA |
Type |
|
|
After carrying out the interview, what does IPA always start with |
A detailed reading and analysis of a single case |
|
|
5 steps of IPA |
1: Read through transcript 1 (several times) 2: Identify key words/phrases 3: identify themes 4: clustering of themes (connections between themes) 5: integration of cases (if you have >1 case), use these themes as hypotheses for other interviews |
|
|
How to keep validity in an IPA |
As you do the analysis stick to the data, ensuring it fits In the write up present illustrative quotes from each of your themes |
|
|
3 of Smith's criteria for quality in qualitative analysis |
Skilled interviewing Well evidenced Commentary/narrative shows both convergence and divergence - not too big leaps beyond data |
|
|
What is Willig's 2 critical conclusions about IPA |
Assunes that language is essential an unproblematic tool for access into cognitions How the world is experienced is not necessarily what the world is really like |
|
|
Other differences between IPA and thematic analysis |
IPA - suggested sample size IPA - when interviewing, TA can be social media use IPA builds up codes and themes from the single case |
|
|
5 parts of DA |
1. Formulate research qs 2. Gather material 3. Reading 4. Coding 5. Analysis |
|
|
What materials can be used for discourse analysis |
Any textual matetial |
|
|
What is coding like in discourse analysis |
Selecting and organising data Not the analysis itself, done prior to the actual analysis |
|
|
What do you qs do you ask while analysing in discourse analysis |
What kinds of things does the language serve to define or construct? What kind of people does the language serve to define or construct? What are the functions of talking about these things/people in these ways in this context? |
|
|
7 rhetorical devices |
Disclaimer Stake innoculation Extreme case formulation Categorical entitlement Passive voice 3 part list Identity claims |
|
|
What is a disclaimer |
An explicit disavowal of the very stance or opinion a speaker subsequently advocates |
|
|
What is stake inoculation |
A speaker rebuts the potential claim that they have a prior interest even before they are challenged on it "At first, I was skeptical about the new cream. But after I tried it..." |
|
|
What is extreme case formulation |
Semantically extreme word/phrase used to defend or justify a description or assessment, especially in ease of challenge, often involves exaggeration |
|
|
What is categorical entitlement |
Based on the fact we accept that certain categories of people (e.g. experts) are entitled to make specific knowledge claims, give them special credence "The psychologist told me my child is gifted" |
|
|
How does a passive voice have an effect in discourse |
A way to downplay the responsibility of the actor in relation to the verb "Police killed rioters" vs "the rioters were killed (by police)" |
|
|
What are identity claims |
Identity as something done in talk to achieve various objectives |
|
|
What is Potter and Wetherell's validity criteria |
Ppt orientation: if ppts see the constructions as different, then analysts can have confidence in doing so |
|
|
Gilbert and Mulkays study showing ppt orientation |
Scientists contradictory repertoire of science Labelled own practice of discovery Labelled others' as bias Kept them separate |
|
|
What device did scientist use when confronted on their contradictory repertoires of science |
'The truth will out' |
|
|
What is Liz stokoe's type of discourse analysis called |
Conversation analysis |
|
|
What 3 things did Liz Stokoe practically use conversation analysis for |
Helped mediators resolve disputes between neighbours Led people to become clients in the initial encounter Crisis negotiators - getting people to talk = lets 'speak' not 'talk' |
|
|
Discourse is the study of talk as ? |
Action |
|
|
3 Stages of factor analysis |
Identify variables and design Check data and assumptions Interpret the results |
|
|
2 parts of checking data and assumptions in a factor analysis |
Before analysis: normality, SDs During analysis: correlation matrix, sphericity and KMO |
|
|
3 parts of interpreting the results in a factor analysis |
How many factors are there? Which items load onto each factor? What trait/characteristic/idea does each group of items capture? |
|
|
What must SDs be between |
0.5 and 1.5 |
|
|
What 4 things do you check on a correlation matrix |
Looks for items with: r<.3 r>.9 p>.05 Check the determinant is greater than 0.00001 |
|
|
What does r>.3 show on a correlation matrix |
Item doesn't correlate with anything else |
|
|
What does p>.05 show on a correlation matrix |
Item doesn't correlate with anything else |
|
|
What does r>.9 show on a correlation matrix |
Item correlates too highly - mutlicollinearity |
|
|
What does a determinant > 0.00001 show |
No problems with multicollinearity (nothing correlated too high) |
|
|
What tests checks for suitability on a factor analysis |
Kaiser-Meyer-Olkin Measure (KMO) |
|
|
What are the KMO cut offs |
Marvellous = >.9 Meritorious = >.8 Middling = >.7 Mediocre = >.6 Miserable = >.5 Merde = <.5 |
|
|
What is a test of sphericity for factor analysis |
Bartlett's test |
|
|
What does Barlett's test show |
Whether the correlations are too small for factor analysis |
|
|
Correlations are large enough for FA, what will be shown |
Bartletts test will be significant |
|
|
What 2 things can be used for factor extraction |
Kaiser's criterion Scree plot |
|
|
What does Kaiser's criterion lead to |
Retaining factors with eigenvalue > 1 |
|
|
What is an eigenvalue |
The variance in all the variables accounted for by a particular factor |
|
|
When are Kaiser's criterion reliable? |
There are <30 variables and all communalities are >.7 OR There are >250 partipants and average communalities >= .6 |
|
|
What is communality |
The percent of variance in a variable explained by all of the factors together (ie communally) |
|
|
When can you use a scree plog |
If Kaiser's criterion unreliable and you have > 200 ppts |
|
|
What does factor rotation do |
Optimises how the items load onto a factor (equalising the relative importance of each factor) Aiding and clarifying the interpretation Spread oyt the variabce more evenly Same total variance explained, but eigenvalues change and become better distributed |
|
|
What are the 2 types of factor rotation |
Orthogonal Oblique |
|
|
When should you use orthogonal rotation |
If factors unrelated/independent/uncorrelated |
|
|
When should you use oblique rotation |
Factors are intercorrelated |
|
|
Type of orthogonal rotation |
Varimax |
|
|
Type of oblique rotation |
Direct oblimin |
|
|
In an oblique rotation, how does the axis move? |
Move so that they are more correlated with each other |
|
|
In an orthoganol rotation, how is the axis moved? |
Axis kept separate and rotated together |
|
|
What is a correlation |
A standardised measure of to what degree 2 variables covary |
|
|
5 types of reliability |
Test-retest reliability Alternative forms reliability Split-half reliability Chronbachs coefficient alpha Inter-scorer reliability |
|
|
Test-retest reliability and alternative forms reliability measures what |
Reliability across time |
|
|
Split half reliability, chronbachs coefficient alpha and inter-scorer reliability shows what |
Internal consistency |
|
|
What is alternatice forms reliability |
Change the wording/order of the qs, seeing correlation between results |
|
|
What is split-half reliability |
Group qs in a questionnaire that measure the same concept Calculate the correlation between those 2 groups |
|
|
What is reliability based on in split half reliability? |
How you split the data |
|
|
What does chronbachs alpha do |
Split the questions on your scale every possible way and computes correlation values for all splits, average of these = a |
|
|
What chronbachs alpha shows acceptable reliability |
>= .7 |
|
|
What chronbachs alpha shows good reliability |
>= .8 |
|
|
What does IRI stand for |
Item reliability index |
|
|
At what value of IRI should you consider removing the item |
<.3 |
|
|
What value of chronbachs alpha is acceptable between 2 items |
Impossible to get alpha because 2 items can only be split one way |
|
|
What to use to check internal consistency between 2 items |
Pearsons correlation |
|
|
2 types of construct validity |
Convergent validity Discriminant validity |
|
|
What is convergent validiry |
Measures of constructs that should be related in theory are in fact related in reality |
|
|
What is discriminant validity |
Measures of constructs that should NOT be related in theory are in fact NOT related in reality |
|
|
What does mediation investigate |
How X => Y |
|
|
What does moderation investigate |
When X => Y |
|
|
Letter for mediation |
M |
|
|
Letter for moderation |
W |
|
|
What effect does a mediator show on Y |
Indirect |
|
|
When can you use mediation analysis |
With regression (or ANOVAs) Self-report questionnaires Experimental research |
|
|
In terms of relationship strength between X and Y, when is mediation said to have occurred |
If the strength is reduced by including a mediator (c' < c) |
|
|
What does each letter mean in a mediation triangle |
a = predictor - mediator b = mediator - outcome c = predictor - outcome (without mediator) c' = predictor - outcome (with mediator) |
|
|
Equation for C in mediation |
C= ab + c' |
|
|
What are the 3.5 steps to mediation analysis |
1 - c = simple regression X-Y 2 - a = simple regression X-M 3a - b = multiple regression M-Y controlling for X 3b - c' = multiple regression X-Y controlling for M |
|
|
If you don't control for X when calculating M-Y, what could this do |
Y and M could be correlated simply because the X predicted them both without a direct relationship between M and Y |
|
|
What was the old way of telling if there's a mediation |
Using p values: Full mediation if direct effect reduced to non significant Partial mediation if direct effect reduced but still significant- both an indirect and direct effect |
|
|
What test gives a p value for the estimate or the indirect effect (the ab estimate) |
Sobel test |
|
|
Why is the old way of telling id there's mediation no longer used |
All-or-nothing conclusions drawn from significance tests (change from p = .049 to p=.051 doesn't imply full mediation |
|
|
What is the new method of determining if a mediation has occured |
Bootstrapped confidence intervals of indirect effect |
|
|
How does confidence intervals show if a mediation has occurred |
If 95% BCa CIs does not include 0, then the indirect effect is treated as significant |
|
|
How to report a mediation |
X predicts higher/lower Y because X predict higher/lower M which predicts higher/lower Y Or M mediations the relationship between X and Y Or There is a significant indirect effect from X to Y via M WITH Indirect effect = , 95% boostrapped (BCa) CIs [LB, UP] |
|
|
What does b represent if mediating in categorical variables |
The difference in the outcome between the 2 categories |
|
|
Moderation is the same as what |
Interactions |
I |
|
In a moderation, what are X and W labelled as |
Main effects |
|
|
What is evidence of a moderation |
If the interaction is significant |
|
|
What value of the moderator do we want to estimate the coefficient for the predictor |
The mean |
|
|
What ia adjusting the variable so that is has a mean of 0 called |
Grand mean centring |
|
|
How do you grand mean centre |
Subtract the mean from each ppts score for predictor and moderator (not interaction) |
|
|
How to report the predictors b parameter |
At the mean of the moderator, there was a significant +ve/-ve relationship between the predictor and the outcome |
|
|
We can interpret a significant interaction using what |
Simple slope analysis |
|
|
What does a simple slope analysis tell us |
What the coefficient for the predictor is at different values of the moderator |
|
|
How to report mediation |
The relationship between X and Y is conditional upon W When W is high (+1SD from mean), X is strongly +/-vely related to Y At the mean of W (?), X is +/-vely related to Y When W is low (-1SD from mean), X is non-significantly? related to Y |
|
|
When can you use moderation analysis |
Same as mediation Regression Survey research Experimental research |
|
|
What are the parts of the cake analogy in variance |
SST = total variance SSM = varience explained by model SSR = residual variance |
|
|
Linear model equation |
Y = b1x + b0 + e |
|
|
How to work out b1 from a graph (gradient) |
Change in y/change in x |
|
|
What is hierarchical data |
Data when there are existing sub-groups or structures within it |
|
|
When there are groups within data (hierarchical) what don't we have? |
Independence of errors |
Which assumption |
|
What model accounts for the structure of the data (hierarchical) |
Multilevel models |
|
|
If we were to use the linear model for hierarchical data, what would entail and do |
Would have to summarise/average across levels Would lose a lot of the original detail |
|
|
How is the multilevel model different to the linear model |
Has some new 'random' elements |
|
|
What is the predictor called in multilevel models |
Fixed effect |
|
|
What is the heirarchical group called in multilevel models |
Random effects |
|
|
Random effects basically means what |
A different slope/intercept for each group |
|
|
3 benefits of multilevel models |
Captures the existing relationships between people or groups that would otherwise be unexplained variance Explicitly models/accounts for the structure of hierarchical data You can add random intercepts and/or slopes to any type of linear model |
|
|
A pattern matrix is used for which rotation |
Oblique |
|
|
A rotated factor matrix is used for which rotation |
Orthogonal rotation |
|
|
What are the two versions if alpha |
Normal (used when items are summed to produce a single score on a scale) Standardised (used when items on a scale are standardised before being summed) |
|
|
How to report reliability of individual items in a scale |
All items correlated with the total scale to a good degree (lower r =) |
|
|
How to report a total effect (C) before mediation |
X positively predicts Y b= ,t= , p = , 95% BCa CI [LB,UP] This explains?% of the variance in Y |
|
|
Are confidence intervals bootstrapped for moderation |
No |
|
|
Are CIs bootstrapped for mediation |
Yes |
|