research methods Flashcards
what are the experimental methods?
lab, field, quasi
what is a lab experiment?
highly controlled in an artificial environment.
what is a field experiment?
controlled in a natural environment.
what is a quasi experiment?
no control, the IV is naturally occurring.
what is the iv?
the cause, what you change.
what is the dv?
the effect, what you measure.
what are controls?
what stays the same.
what are confounding variables?
something that affects the DV and its validity.
what are extraneous variables?
a variable that could affect the DV but has been controlled for.
what are the experimental designs?
independent, repeated, matched pairs
what is an independent measures design?
participants are in only one condition of the experiment.
benefits of independent measures
Only experiences one condition so unable to guess the aim of the study; reduces situational variables; easy to replace.
limitations of independent measures
Twice as manty participants needed; increases participant variables; some may be ‘naturally better’- confounding variable.
what is a repeated measures design?
in both conditions of the experiment.
benefits of repeated measures
Controls participant variables; only half the number of participants needed.
limitations of repeated measures
Situational variables (use counterbalancing); demand characteristics; two tasks of the same difficulty must be made; if someone drops out two data sets are lost.
what is a matched pairs design?
matched on certain characteristics.
benefits of matched pairs
Controls participant variables and situational variables.
limitations of matched pairs
If someone drops out, you must find a new match or risk losing two sets of data; requires hard work to match; some may drop out early.
what are individual differences?
demand characteristics, fatigue effects and order effects.
alternate hypothesis
“there will be a significant difference between…”
null hypothesis
“there will be no significant difference between…”
participant variables
age, sex, mood, gender, culture, ethnicity.
situational variables
environment, time of day, order effects. Controlled by standardisation and counterbalancing.
standardisation
uses the same order and surroundings to increase validity and reliability.
counterbalancing
ABBA to deal with situational variables.
investigator variables
body language, tone of voice, confirmation bias, demand characteristics.
single blind
participant does not know what the study is about.
double blind
both the participant and the investigator do not know what the study is about.
correlation
A mathematical technique to establish a relationship between two quantitatively measured variables.
positive correlation
both variables increase together.
negative correlation
variables change in opposite directions.
coefficient
tells you how strong the correlation is.
benefits of correlation
Indicates a connection between two variables in situations where experimental proof is impossible; does not require manipulation of variables therefore safer and more ethical; ecologically valid as the numbers come from real life.
limitations of correlation
Does not prove a causal relationship; does not reflect a curvilinear relationship; subject to issues with the method used to collect the data.
descriptive statistics
summary of data to illustrate patterns and relationship, but cannot infer conclusions (mode, median, mean).
inferential statistics
tatistical tests that allow us to make conclusions in relation to our hypothesis (Mann-Whitney U, Spearman’s Rho).
nominal data
category data (measure of central tendency is mode).
ordinal data
data ranked in order (measure of central tendency is median).
interval data
data measured on a fixed scale (measure of central tendency is mean).
self reports
Questionnaires and interviews are two types of self-report where the participant tells you how they are thinking and feeling.
closed questions
limited choice which provide quantitative data. Easy to compare and analyse, but difficult to provide depth and explanation.
open questions
give qualitative data, writes depth and detail but difficult to analyse and compare.
likert scale
a data collection method on a scale which uses an odd number to have a midpoint. Reflects strength of feeling, qualitative data with no explanation.
semantic differentials
similar to a likert scale but asks people to place themselves on a line between two extremes. Used to measure attitudes.
social desirability bias
when people choose the middle option in a self-report if they don’t know or don’t want to appear extreme.
response bias
when people continue to tick the same box. Can be stopped by reversing half the questions to be framed positively and others negatively or using split-half method.
split-half method
subtly repeats questions to ensure the opinion remains consistent.
structured interview
all planned questions.
unstructured interview
a conversation.
semi-structured interview
uses some set questions but allows participants to expand on answers. The best method as it provides qualitative and quantitative data.
benefits of self report
Allow participant to give views rather than just inferring from conversation; can study large sample easily and quickly; examine a number of different variables; asks people to reveal behaviour and feelings from real life situations (ecologically valid).
limitations of self report
Social desirability bias; validity issues from unclear questions; low response rates; leading questions; quantitative data does not include reasoning; qualitative is hard to analyse; reliability and validity in context of the situation.
improving validity
Qualitative is more valid than quantitative by being able to see with greater ease; compare self-report with another on the same topic to establish concurrent validity; avoid leading questions; add open ended questions; reinforce confidentiality to reduce social desirability bias.
improving reliability
Ensure questions are not ambiguous; interviews must be standardised.
participant observation
the observer acts as part of the group being watched.
non-participant observation
they do not become part of the group.
naturalistic observation
natural environment.
controlled observation
variables are controlled and manipulated by the experimenter.
structured observation
determines the behaviours to be observed and the sampling to be used.
unstructured observation
where the observer records everything that happens.
overt observation
the participant knows they are being studied.
covert observation
do not know they are being studied.
time sampling
observations may be made at regular time intervals and coded.
time point sampling
observations are made at fixed intervals.
time event sampling
observations are made at a fixed period.
event sampling
keeping a tally chart of each time a behaviour occurs.
behavioural categories
used in structured observations to decide what is going to be observed and how it is going to be observed.
coding frames
used to make recording behavioural categories easier by listing different behaviours as different ‘codes.’ Allows information to be recorded quickly.
validity of observations
Demand characteristics and observer bias can reduce validity. Can be improved by using wider categories, a single-blind technique, or self-reports.
reliability of observations
Difficult to replicate observations due to confounding variables; check consistency through inter-rater reliability; using good coding schemes.
internal validity
how properly the experiment was conducted to produce truthful/accurate results.
external validity
can the findings (ecological and population) of the study be generalised?
face validity
does something look like it will measure what it is supposed to measure?
construct validity
does it measure all aspects of what is being assessed?
concurrent validity
when a test correlates well with a pre-validated test.
criterion validity
when a test correlates well with a pre-validated test.
ecological validity
whether the study reflects real life situations.
population validity
whether the sample is representative of the wider target population.
internal reliability
whether the procedure is standardised and consistent.
external reliability
the extent to which the results of a procedure can be replicated.
inter-rater reliability
agrees beforehand what will be observed and pilot studies are conducted to ensure this works. Each observer then observes the same thing independently. The results are compared at the end of the observation, and the greater the level of similarity, the greater the inter-rater reliability. This means that coding schemes (measuring tools) are consistent by producing the same results with different people. Increases internal reliability.
test-retest
repeating the same test on the same sample at a different point in time to measure internal reliability. Increases external reliability.
deductive reasoning
hypothesis -> experiment. Logical and tests theories to drill down.
inductive reasoning
observation -> theory. Based on experiences and only takes one exception to falsify.
what are the categories for ethical guidelines?
respect, competence, responsibility, integrity
respect
Informed consent, right to withdraw, confidentiality.
competence
Work within ability, consult with colleagues.
responsibility
Protection of participants, debrief.
integrity
Deception
one-tailed hypothesis
predicts the direction of the relationship.
two-tailed hypothesis
does not know the direction of the relationship.
sampling
Used because it takes too long to study everyone. It uses a sample to be representative and generalisable to the wider population.
population
the group from which the sample is drawn/who you are interested in studying.
opportunity sampling
taking people who are available at the time.
benefits of opportunity sampling
Quick, cheap, and easy.
limitations of opportunity sampling
Not representative because it misses out people.
random sampling
every member of the population has an equal chance of being chosen.
benefits of random sampling
Should be representative.
limitations of random sampling
Expensive; time consuming; if people decline it is no longer representative.
self-selected sampling
volunteer when asked or in response to an advert.
benefits of self-selected sampling
Quick, cheap, and easy. Have consented.
limitations of self-selected sampling
Not representative; higher demand characteristics.
stratified sampling
classify the main population into categories and choose a sample in the same proportion as they are in the population.
benefits of stratified sampling
Representative.
limitations of stratified sampling
Time consuming; can only be stratified on one category; group must get bigger to have the same proportions; if people drop out the group is no longer representative.
snowball sampling
using a small group of people to ask others to take part.
benefits of snowball sampling
Useful for hard to reach groups.
limitations of snowball sampling
Time consuming; participants may decline.
bar charts
used for discrete categories (nominal data).
pie charts
used when data represents a whole.
histogram/line graph
to show a pattern in a whole data set (ordinal or interval data).
scatter graph
display the findings of correlational studies.
primary data
gathered by yourself for the study.
secondary data
gathered from another source.
measures of dispersion
Give an indication of how spread out results within a data set are.
range
(biggest number - lower number)
benefits of range
The simplest measure of spread. Quick and easy.
limitations of range
Can be distorted by outliers and does not account for all values.
variance
Variance- gives an idea of how dispersed scores are around the mean.
The average of the squared differences from the mean.
benefits of variance
Takes every score into account.
limitations of variance
Cannot be directly compared to original data, as it is squared.
standard deviation
tells us the average distance of each score from the mean.
The square root of the variances and 68% of scores fall in one standard deviation of the mean.
benefits of standard deviation
More precise as accounts for all values.
limitations of standard deviation
Much harder to calculate.
significant difference
means something has been found to be significant after being tested with inferential statistics. The observed value is compared with the critical value to see if it is significant and the common level for significance in psychology is p<0.05.
critical value
p<0.05
chi square
independent, nominal
binomial
repeated, nominal
mann-whitney u
independent, ordinal
wilcoxon signed rank
repeated, ordinal
spearman’s rho
correlation
in which tests does the observed value have to be greater than or equal to the critical value?
chi-square and spearman’s rho
negative skew
left
positive skew
right
report writing
abstract, introduction, method, results, discussion, references, appendices
abstract
summary of investigation.
introduction
sets the scene and background research.
method
design, apparatus, participants, procedure.
results
descriptive and inferential statistics.
discussion
explaining and evaluating results.
references
Boring, E.G. (1929) ‘A History of Experimental Psychology’, New York: Century. Page 20.
Author. initials, data, title, publisher, pages.
Sperry, R.W. (1968) ‘Hemisphere deconnection and unity in conscious awareness. ‘American Psychologist,’ 23, 723-733.
Author, initials, date, title, book, volume, pages.
appendices
consent form, instructions, raw data, ethics, calculations.
non-parametric tests: chi square
- Calculate the totals for each row and column.
- Then Expected Frequencies = (Row total ×Column total)/Overall total
- Do this for each cell then calculate the Chi-square value for each cell = sigma (observed - expected)^2 / expected
- Add all the numbers from each cell together to give your final Chi-Square observed value.
- Work out the degrees of freedom = (Number of Rows – 1) x (Number of Columns – 1)
- Look it up on the table to get the critical values.
- The observed value must be greater than or equal to the critical value for the test to be significant.
what do parametric tests assume?
the data collected is normally distributed
that the data from both levels of the iv has equal variance
that the level of measurement is interval data
variance formula
s² = sigma ( x - x̄ )² / n - 1
standard deviation formula
s = √sigma ( x - x̄ )² / n - 1
type one error
accept the operationalised hypothesis and reject the null.
type two error
accept the null hypothesis and reject the operationalised.
