Qualitative and Quantitative methods Flashcards
why use observational methods
questionnaires limited applicability
- apparatus limits generalisability
- context-dependent behaviour infeasible in controlled environments
steps of observational research
observe informally ask questions choose measures choose recording method design experiment run experiment analyse publish ask more questions
choosing measures in observation
operational hypothesis or ostensive definitions
events or states
what is a ostensive definition
provide examples through pictures or diagrams along with descriptions of behaviour of interest
what is an ethogram
- full list of behavioural repertoire
- sometimes denotes quantitative description of time animal spends doing activity
- refer to complete repertoires and coding schemes in specialised studies of subset of species’ or groups behaviour
types of measures in observation
- latency
- frequency
- rate
- duration
- proportion
scales of measurement
nominal
ordinal
interval
ratio-interval
recording methods, sampling rules
ad libitum
focal sampling
scan sampling
behaviour sampling
what is ad libitum sampling
- record what ever want, for as long as want
- potential bias: miss rare events of short duration and underestimate contribution to smaller, less conspicuous subjects
what is focal sampling
specific individual isolated
potential bias: can be large if focal subject seeks privacy for specific behaviours
what is scan sampling
number of individuals observed
potential bias: rare events of short duration underestimated, while common events overestimated
what is behaviour sampling
all-occurrences sampling
potential bias: overestimation of conspicuous events
recording methods, recording rules
time sampling: instantaneous, one-zero
continuous recording
potential biases of recording rules
time sampling: underestimate rare behaviours of short duration
continuous: high fidelity records, fewer categories coded, underestimate long duration behaviours as more likely truncated by end of session
coding schemes
applied to observations to produce data
data isn’t standardised so can refer to videos or numerical data
coding schemes: principles measurement
perfect doesn’t exist
measurements more/less accurate
measurements more/less precise
what does accurate refer to (coding schemes)
correct and valid measurement
what does precise refer to (coding schemes)
exactitude: reliable and replicable
coding schemes: principle design
mutually exclusive
exhaustive
use category ‘other’ for behaviours not interested in
reliability; coding scheme
- precision of coding scheme NOT precision of observer
- sloppy observer show low reliability but ambiguous from reliability estimate
- reliability only attributable to specific coding scheme applied by specific observer in specific spatiotemporal contexts
intra observer reliability
same observer codes same behavioural record at diff times
how consistent observer is
inter observer reliability
different observers independently code same behavioural record at same/different times
-how similar different observers code
consensus measures of inter observer reliability
- assumption that two or more coders can come to exact agreement
- percentage agreement: not correct for random choice
- cohens kappa: proportion of agreement after correct fro random choice
consistency measures of inter observer reliability
assumption that it’s unnecessary for identical interpretation
- correlation coefficient: doesn’t take into accoutn variance between coders
- cronbachs a: corrects for variance, estimates reliability for two+ coders
equation for percentage agreement
A / A + D
A is agreed
D is disagreed
equation for cohen’s kappa
Pa - Pc / 1 - Pc
Pc = P(yes) + P(no)
Pa is proportion of observed agreement
Pc is expected agreement
must be 0.6+ to be good enough
advantage of self report
own opinions
gather large amount of data
economic
time effective
sources of bias in self report
social desirability
acquiescent response styles
careless responding
how can you avoid bias in self report
- avoid presenting options as more desirable
- use indirect items
- confidentiality and anonymity
- change between positive and negative items
- reverse phrased items
- antonymic expression
- direct negation
what is a nomothetic or idiographic research question
nomothetic: relating to discovery of general laws, quantitative
idiographic: relating to study of particular facts or processes, qualitative
what are structured, semi-structured and unstructured interviews
structured: exact same for everyone, could be written
semi: set topics and some basic items for everyone
unstructured: general themes and issues that want to discuss
some principles for interviews
funnelling express ignorance and ask fro concrete examples avoid double barrelled no assumptions no complex/jargon words no double negatives
why should you conduct a pilot interview
helps identify problems in wording
what is a rapport and why is it need in interview
informal conversation to enhance disclosure
relaxed atmosphere of openness
giving space to allow unanticipated material to be brought up
analysis issues of interview
transcription: words? meaning? non verbal communication? what happens with transcript?
coding: complete line by line, top down or bottom up
what is top down and bottom up coding
top down: know what looking for
bottom up: no preconceptions, looking at concepts/categories/language
why do interviews
- function theoretical orientation
- exploration of topic
- give participants voice; own language, concepts, categories
- higher in validity
- subjective experiences
problems with interviews
over reliant on self report
need observational and behavioural measures to verify what they say is true
unstandardised questions limit comparability
interviewer effects
what are interviewer effects
impact of interviewer gender, ethnicity, appearance
actions of interviewer; response biases
inhibiting some answers but not others
inconsistent training
Types of qualitative analysis
thematic IPA grounded theory discourse analysis conversation analysis
what is thematic analysis
approach which identifies themes in textual material
underpins many forms of quali analysis
what is a theme
pattern of meaning
captures something important about material
6 steps of thematic analysis (Braun and Clarke 2006)
- familiarise self with material
- initial coding: codes = basic units of meaning
- searching for themes: play with grouping codes
- review themes: use judgement, drop, split or merge themes
- define and name themes; add narrative to explain what is interesting and why
problem of thematic analysis
- analyst sees five or six themes and looks for examples of it
- no justification or explanation
- no criteria
advantage of thematic analyssi
can be used on any quali data-set
what is IPA
form of qualitative analysis that makes number of psychological assumptions
analysis of subjective experiences of people’s lifeworlds
IPA- assumptions about knowledge
people interpret world of phenomena; what we study is their interpretation of world
researchers interpret world too; interpret people’s interpretations, reflexivity is built into IPA
designing IPA study
always idiographic as looking at particular cases always about meanings it's quality over quantity sample size; 6-8 standard homogenous sample
rule of transcribing
for verbatim interviews, one hour coding = 6 hour transcription
IPA alwasy begin with detailed reading and analysis of one case before move to next
stages of IPA analysis
- read through transcript
- identify key words or phrases; make judgment abou twhat reflects speakers experience
- identify themes; key words may be indicative
- clustering of themes; superordinate themes, connections between them
- integration of cases; use themes from first as hypotheses for rest
validation of IPA
validity of analytic claims is matter of their plausibility
- iterative reading; stick to data and return to it
- ask self are these instances similar/different?
- present illustrative quotes for each themes
thematic vs IPA
- TA for analysisng social media, IPA for interview
- TA atheoretical, IPA psychological assumptions
- IPA recommends specific sample size
- IPA builds codes and themes from single case
characteristics of thematic analysis
flexible
transparent (clearly explain decisions made at each step)
recursive
types of comments in IPA coding
descriptive: describing content of what participant says, subject of talk
linguistic: exploring specific use of language by participant
conceptual: engaging at more interrogative and conceptual level
ways to cluster themes (Smith 2009)
Abstraction: basic form of identifying patterns between emergent themes and developing ‘super-ordinate theme’
subsumption: the emergent theme itself becomes the superordinate
polarisation: opposing themes
contextualisation: consider wider context
numeration: taking account of frequency of theme
function: function in transcript
what is discourse analysis
- analysis of talk and text in their own right
- distinctive way of thinking about textual material; construction of reality
stages of DA
formulate research Q gather material read text 2-3 times coding; select and organise data analysis; what language constructs, what kind of people defines, functions of talking about things in that context, suggest other words or phrases that might have been used instead
rhetorical devices in DA
disclaimer stake inoculation extreme case formulation category entitlement passive voice three part list identity claim
what is a disclaimer
Explicit disavowal of very stance or opinion speaker advocates
what is stake inoculation
Claim they have prior interest before they are challenged on it
what is extreme case formulation
Semantically extreme word or phrase to defend or justify description/assessment
what is category entitlement
Give special credence to certain categories of people that are entitled to make specific knowledge claims
what is passive voice
Downplays responsibility of actor in relation to verb
what is identity claims
persona along with the degree or range of power a particular person can claim in a specific discourse
validity of DA
participant orientation: if speaker treats constructs differently then we can too
written presentation allows readers to judge
reliability fo DA
if same material examined more than once, do get same answer?
rhetorical devices are specific and identifiable that can relate to for consistency
advantages of DA
gives tools to study how people do things with words and how language constructs reality
study power
study any topic
wide variety of data types
disadvantages of DA
doesn’t allow for usual generalisations
relativism: only comment on the language not what it represents
what is a latent variable
can’t be observed directly
label given to collection of experiences or events or representations of things
what is moderation
alludes to combined effect of two variables on outcome
a third variable influences the relationship between outcome and predictor
what is mediation
when the relationship between a predictor variable and an outcome variable can be explained by their relationship to a third variable
model of mediation
total effect = indirect effect + direct effect c' = ab + c a = X influence on M b = M influence on Y c = X influence on Y
when to use mediation
if you can use a linear model, you can use mediation
correlational : investigate process through which objective measures are related to outcomes (underlying psychological process)
experimental: investigate how experimental manipulation influences DV
steps of mediation analysis
-simple regression of total effect of X on Y
-indirect effect: simple regression of X -> M and M-> Y controlling for X
-direct effect: multiple regression controlling for M
ALL UNSTANDARDISED B’S
how to analyse mediation
bootstrapped confidence intervals of indirect effect
if 95 CI not include zero, then indirect effect is treated as significant
statistical model of moderation
three predictors in linear model, X, W and XW
grand mean centring
- for deviations around fixed point
- transform variable so it has mean of zero
- subtract mean from each participant’s score, do for both predictor and moderator
interpreting significant interactions
use simple slope analysis or johnson-neyman interval
simple: at what mean of the moderator is the direction of effect and if its sig
J-N: plots line when relationship is zero, anything above is positive and anything below is negative
