representation of quantitative data, statistical testing, peer review, psychology research and the economy, which test and why

correlations, reliability, validity, features of a science

how to present and display quantitative data

graphs, tables, scattergrams, bar charts, histograms

types of distributions

normal and skewed

normal distribution

symmetrical, bell shaped curve, most items in middle area of curve with very few at extreme ends, mean, median, mode occupy same mid point of curve

skewed distributions

lean to one side or the other because most items are either at lower or upper end of distribution

2 types of skew distributions

positive, negative

positive skew

distribution concentrated towards left of graph, mode is highest point, followed by median, mean

example of positive skew

a very difficult test that people get low marks on

negative skew

distribution to right of graph, long tail on left, mode is highest peak, followed by median and mean

example of negative skew

very easy test where people get high marks

why do we use a statistical test

find out if difference or association is significant, difference or association between 2 data sets is greater than what would occur by chance

probability

how likely it is that an event will happen if null is true

what is the accepted level of probability (significance level)

0.05

what happens at the 0.05 level

researcher decided to accept or reject null

what can psychologists find

statistically significant differences or relationships

what cant psychologists find

statistical certainties

why might a researcher use a statistical test

produce a calculated value

example of statistical test

sign test

why is calculated value compared with critical value

decide whether result is significant or not

how to find critical value

1. know significant level- 0.05

2. number of ppts- n value or df

3. whether hypothesis is directional (one tailed) or non directional (two tailed)

what are the conditions of use in sign test

used to analyse difference in scores between related items, can be used with nominal data

how to calculate sign test

1. each person score subtract condition b from condition a to produce sign of difference

2. work out how many + and -

3. = score should be ignored

4. s value is total of less frequent sign

critical value

if s is = or less than critical value then its significant, null is rejected and alternative hypothesis accepted at 5% level of certainty

peer review

before publication all aspects of investigation are scrutinised by experts in the field, who should be objective and unknown to researcher

aims of peer review

• funding, allocate research funding

• validation, quality and relevance of research

• improvements to report re suggested

strength of peer review

protects quality of published research, minimises possibility of fraudulent research and means published research is high quality, preserves reputation of psych as a science

limitation of peer review

publication bias, tendency for journals to want to publish headline grabbing findings, means some research is ignored, file draw problem

how can eyewitness research benefit the economy

improves accuracy of eyewitness info, reduces wrongful convictions and costly retrials

how has eyewitness research improved police interviews

research on memory led to development of cognitive interview

what is the cognitive interview designed to improve

accuracy and amount of info recalled by eyewitnesses

why are leading questions important in eyewitness research

research shows they can influence recall, improves understanding of memory

what is post event discussion

witnesses discuss event afterwards, can influence later recall

how can eyewitnesses research reduce economic costs

fewer mistaken accusations can mean fewer retrials and compensation payments

how can psychological research benefit mental health treatments

research identifies causes and effective treatments for mental health treatments

how can mental health research benefit the economy

effective treatments can help people return to work and contribute economically

why do mental health absences affect the economy

half of absences from work are linked to mental health conditions such as depression

how can treatments such as ssri benefit the economy

can help people manage their condition effectively and return to work

which test and why table acronym

super

cats

will

meow

silently

racing

up

pipes

unrelated design

study uses different groups of people for each condition

related design

study uses same people in both conditions

correlation/association

study doesnt split people into groups, measures 2 things to see if they change together

ordinal data

can be ranked or scored on a scale, but gaps between numbers are subjective, e.g rating happiness from 1-10

nominal data

data fits into categories or boxes

interval data

data measured on precise scale with equal fixed gaps between units

chi square

used to see if there is a pattern or link between categories of data

is chi square correlation or association

a

wilcoxon

compares ranks or scores of same group tested twice to find a difference

spearman’s rho

checks for a relationship between 2 sets of ranked scores

is spearmans rho correlation or association

c

unrelated t test

highly precise test comparing exact averages of 2 separate groups

related t test

highly precise test comparing exact averages of same group tested twice using interval data

pearsons r

most precise test for finding linear relationship between 2 exact interval scales

is Pearsons r correlation or association

c

type 1 error

null rejected alternative hypothesis accepted when null is actually true

type 2 error

null accepted but alternative hypothesis is true

what makes a type 1 error more likely

if significance level is too lenient

what makes a type 2 error more likely

if significance level is too stringent

what did Kuhn 1962 say about paradigm

what distinguishes scientific disciplines from non specific disciplines is shared set of assumptions and methods

what did Kuhn also argue

social sciences lack a uni accepted paradigm and are best seen as pre science, unlike natural sciences

paradigm shift

occurs when there is a scientific revolution , a handful of researchers question the accepted paradigm when there is too much contradictory evidence to ignore

theory

set of general laws or principles that have ability to explain particular events or behaviours

what does testing a theory depend on

being able to make clear and precise statements on basis of theory

correlation

measures relationship or association between 2 continuous variables

what is the numerical value statistical tests of correlation produce

-1 and +1

what is -1 and +1

correlation coefficient, tells us the strength of relationship between 2 variables

what does it mean if coefficient is closer to +1 or -1

strong relationship, weak if closer to 0

what are correlation coefficients calculated using

statistical test

reliability

when a measurement is repeated and same result is obtained, consistency

2 ways to test reliability

test-retest, inter-observer

test-retest reliability

testing same person twice, test should be similar or the same depending on previous results

inter-observer reliability

compare observations from different observers, conduct a pilot study, observers should watch the same event but record data separately

how is reliability established using a correlation

if correlation coefficient isnt 0.8 its not great

4 ways to improve reliability

questionnaires, interviews, observations, experiments

questionnaires improving reliability

rewrite questions when a questionnaire produces a lo test retest reliability, replace open questions with closed ones

interviews improving reliability

improve training by using same interviewer each time, interviewers must be trained to avoid leading questions

observations improving reliability

operationalise behavioural categories so they are measurable and self evident, categories shouldnt overlap, all possible behaviours included

what happens if reliability is low in observations

observers need more training in using behavioural categories and or discuss decisions with each other so they can apply categories more consistently

experiments improving reliablity

standardise procedures to compare performance of different ppts, procedures must be consistent every time

validity

if an observed effect is genuine and represents what is out there in the real world, is result legit

give an example of why data can be reliable but not valid

an iq test may produce same result every time when same people tested but not measure what its designed to measure, low validity

4 types of validity

internal, external, ecological, temporal

internal validity

if researcher measures what was actually intended to be measured

external validity

generalising findings to other settings, populations or eras

ecological validity

do findings from lab settings generalise to everyday life, do research tasks have realism

temporal validity

do findings remain true over time

2 ways of measuring validity

face, concurrent

face validity

does the test measure what its supposed to on the face of it, achieved by eyeballing the measuring instrument or passing it top an expert to check

concurrent validity

whether findings match a well established test, new intelligence test may be used to a group of ppts, scores compared with performance on a well established test

4 ways to improve validity

experiments, questionnaires, observations, qualitative research

experiments improving validity

control groups used so researcher is more sure of changes in dv where due to iv and standardise used to reduce ppts reactivity and investigator effects

questionnaires improving validity

lie scales control effects of social desirability bias, respondents assured that all data submitted is anonymous

observations improving validity

covert observation means authentic behaviour, behavioural categories well defined, thoroughly operationalised, and not ambiguous or overlapping

qualitative research imropving validity

interpretive validity shown through coherence of narrative and direct quotes from ppts, triangulation uses different sources as evidence

5 features of science

paradigms and shifts, theory construction and hypothesis testing, falsifiability, replicability, objectivity, empirical method

theory construction

theory is as set of general laws that have ability to explain particular events or behaviours, testing a theory depends on being able to make clear and precise statements on basis of theory

hypothesis testing

hypothesis can be tested using scientific methods to determine if it will be supported or not

what is deduction in hypothesis testing

process of deriving new hypothesis from existing theory

falsifiability

proof is not possible, genuine scientific theories should hold themselves up for hypothesis testing and possibility of being proved false

what did popper 1934 argue about falsifiability

its the key criterion of a scientific theory, she distinguished between theories that can be tested and falsified, pseudosciences do not allow possibility of being shown to be false

replicability

testing validity of research results, if scientific theory is going to be truetd, findings must be shown to be repeatable across a number of different contexts

what does repeating a study mean

can see extent to which findings can be generalised