content analysis
a form of observational research where people are studied indirectly via communications they have produced (media, written, spoken)
aim is to systematically summarise so that conclusions can be drawn
case studies
detailed study of a single individual, institution or event
strengths and weaknesses of case studies
- diff to generalise to rest of the target population
+ produces rich detailed data
strengths and weaknesses of content analysis
- may lack objectivity as it may impose meaning of behaviour from preconceptions
- reductionist
- detailed qualitative data gets reduced to numerical figure
+ allows for statistical procedures
+ more objective and replicable
thematic analysis
identifying/analysing patterns of meaning within qualitative data to organise into pre-identified themes to analyse an existing theory
strengths and weaknesses of thematic analysis
- time consuming
- subjective
+ allows systematic study and exploration of qualitative data
procedure of content analysis
Sampling: researcher decides what material to use, what they will analyse and how they will collect the info
Coding: researcher must decide how to categorise the analysed materials
Instances of each theme are gathered and categorised
Frequencies counted numerically
Reliability is tested (+0.8 positive correlation)
Conclusions are drawn
Reliability
a measure of consistency
Reliability ain’t great unless it’s +0.8
Ways of assessing reliability
Test-retest
Inter-observer reliability
Test-Retest method
administer the same test/ questionnaire to the same person/people on diff occasions
used for questionnaires, interviews, tests (like IQ etc)
reliable test should give similar results
Inter-Observer reliability
observers should work in teams of 2, watch the same event and record separately
results should be correlated at the end to assess reliability
Ways to improve reliability
Questionnaires
use test retest method, low reliability = some items may need to be rewritten due to ambiguity or too complex
Interviews
use same interviewer each time, structured interviews, no leading questions
Experiments
lab for high control and standardisation
Observations
behavioural categories = properly operationalised, measurable and self-evident
Validity
a measure of legitimacy
internal validity: whether the effects observed are due to manipulation of IVs or not
external validity: factors outside of the investigation such as generalising to other populations, settings etc
Ways of assessing validity
face validity: does it look good?
concurrent validity: compare current research to previously established research and see if they produce similar results
Ways to improve validity
Experimental research
single/double blinds, standardisation, control groups
Questionnaires
lie scales to control social desirability bias, anonymity
Observations
covert observations as behaviour more likely to be authentic, behaviour categories = properly operationalised
Qualitative
interpretive validity = research must be reported coherently, triangulation = multiple diff sources as evidence
hypothesis
precise and predictable statement about the world which can be tested
null hypothesis
alternative hypothesis
significance
indicating that the research findings are sufficiently strong to enable a researcher to reject the null hypothesis
95% + = IV caused change
higher for drug trials = 99%
type 1 error
rejecting a null hypothesis which is true
likely if significance level is too high (10%)
type 2 error
accepting null hypothesis that is false
likely to happen when significance level is too low (1%)
Levels of measurement
nominal
ordinal
interval
Nominal
frequency data that consists of the # of pps falling into categories
eg. 7 ppl passed the test whilst 5 ppl didn’t
Ordinal
data presented in rank order
eg. attractiveness ratings
Interval
data measured in fixed units with equal distance between points on the scale
eg. temperature
descriptive statistics
describe, show differences and pattern in the data set
includes measures of central tendency and measures of dispersion
inferential statistics
find out whether results are significant so they can be used to generalise to the rest of the population
critical value
calculated value needs to be less than or equal to critical value to be significant
parametric tests
better able to detect a significant affect
only used on interval data
homogeneity of variance between conditions is needed
unrelated t-test
related t-test
Pearson’s R
non-parametric tests
used for ordinal and norminal data
less powerful
no homogeneity of variance
Spearman’s Rho
Wilcoxon
Mann-Whitney U-Test
Chi-squared
Mann Whitney U test
used when
hypothesis predicts a diff between 2 sets of data
independent groups design
ordinal data
Wilcoxon
used when
hypothesis predicts difference
repeated measures or matched pairs
ordinal data
Unrelated t-test
used when
independent groups design
interval data
Related t-test
used when
repeated measures or matched pairs design
interval data
Chi-squared
used when
test of difference
independent groups design
nominal data
Spearman’s Rho
used when
test of relationship
ordinal data
Pearson’s R
used when
test of correlation
interval data
Features of science
objectivity
empirical method
replicability
falsifiability
theory construction: evidence needs to be collected before and knowledge cannot be based on beliefs
hypothesis
paradigms and paradigm shift
structure of psychological studies
title
abstract
introduction
aim and hypothesis
method
design of the investigation
procedure
use of participants
resources used
results
discussion
explanation of findings
implications of study
limitations and modifications of the study
relationship to background research
suggestions for further researcher
references
appendices