Psychology Revision
Research Methods
Is Psychology a science?
Yes because…. No because…
Ithas influences in medicine eg Theoriesto do with people
drugs. can not be tested enough.
Goal of science is to create a Psychologists often use
reasonable explanation to
retrospective studies and
describe reality and psychology
does this.
there are too many factors
prone to bias. This process is
It too unscientific.
overlaps with biology and
medical conditions.
We do not know why some
Scientifictheories are always treatments work and continue
theories, they never become to use treatments that do not
final and only explain given work / aren’t very successful.
phenomenon.
Designing investigations
Sampling
There are many ways to take samples from the target
population, such as:
Random sample – everyone has an equal chance of being
selected.
Systematic sample – taking every nth name from a register.
Opportunity sample – selecting whoever is available at the time.
Volunteer sample – participants volunteer.
Stratifiedsample – important subgroups in the population are
selected in proportion to the target sample.
Designing investigations
Types of Experiments
There is more than one type of experiment, these
are:
Laboratory Experiments – These are controlled,
scientific and conducted in an artificial setting.
Field Experiments – These are conducted in a natural
setting but variables are still controlled.
Natural Experiments – These are conducted in a natural
setting where the independent variable is not
manipulated by the experimenter, but occurs naturally.
Designing investigations
Experimental Design
There are also different types of experimental
design:
Matched pairs – Participants are matched with another
participant with similar attributes. They do one condition
of the experiment each, their results are then
compared.
Independent groups – Participants are placed in groups.
Repeated measures – Each participant does both
conditions of the experiment.
Designing investigations
Variables
You must be able to operationalise (name)
your variables.
Independent variable – the variable that is
manipulated by the experimenter.
Dependent variable – the variable that you
measure.
Extraneous variables – Other variables that might
have an effect on the dependent variable.
Measures of Central Tendency
Averages
There are three ways to measure averages.
Mean – This takes all measurements into account but
is easily skewed by anomalous results.
Median – This is not affected by outlying results, but
doesn’t take all measurements into account.
Mode – Often gives us a sensible answer and can be
used with nominal data; however sometimes the
most common result does not represent the entire
sample.
Measures of Central Tendency
Ranges
There are two ways to measure variation
amongst the results.
Range – This is the measurement between the lowest
and highest result. This is useful for knowing the
distance between the highest and lowest result but
doesn’t give us any information about the distance
between each result. It is also massively effected by
outlyers.
Standard deviation – This is the average spread
around the mean. It takes into account all pieces of
information and is an indicator of outlying results.
Reliability
Reliability refers to the ability to repeat the test and achieve
consistent results. There are methods to measure reliability:
Scorer reliability – This is when everyone recording data records it in the
same way, much like marking a test. All scorers must mark the same, so
that if two different scorers were to mark the same test, they would have
the same result.
Test-retestreliability – This is when a test is repeated and the same results
are achieved or the same overall correlation or result is concluded.
Split-Half reliability – This measures internal consistency and is useful if it
isn’t possible to repeat the test. After the test has been taken, one selection
of questions is correlated with the other to make sure that no side was
particularly harder than the other. e.g. the IQ test increases in difficulty as
you progress through it. It could be best to take all odd numbered questions
as one half of the test, and even numbers as the other half to avoid the
possibility of having a difficult/easier part.
Validity
Validity refers to the extent to which you are measuring what you’re
supposed to be measuring.
Content/face validity – This is the easiest way of checking if something is valid.
It is done by examining the data and deciding whether it looks right.
Predictive validity – This is when the results are used to predict future
behaviour, for example, the interview used in job selection is supposed to
predict how good a person will be at the job.
Concurrent validity – This method compares the test score with another
independent method of testing the same thing. For example, scores found from
a new method of measuring height might be compared to the results of
measuring the same thing with a current measuring device, e.g. a ruler.
Construct validity – This is putting the device into a real life setting and seeing
if it works. For example, if a test was supposed to measure anxiety, in a
situation where people are often anxious (such as a sky dive) the test would
show high levels of anxiety compared to in more relaxed situations.
Ethics
There are often many ethical issues in
psychological studies. The most common ethical
issues are:
Informed consent
Right to withdraw
Deception
Protection from harm, both physical and psychological
Debriefing
Confidentiality
Research that raises sensitive social issues, such as
race, genetics, sexuality.
Analysing Data
Types of Data
Qualitative / Quantitative
Observations and interviews often give us
qualitative data. It is difficult to objectively
analyse qualitative data but it gives us much
more detail than quantitative data.
Quantitative data is very easy to analyse
objectively as it can be shown on a graph or
chart and can be easily categorised.
Analysing Data
Presenting Data
There are lots of ways to present
quantitative data.
Tables.
Bar charts – non continuous data.
Histograms – continuous data
Line graphs – more than one set of data
Scatter graphs – correlation data
Probability and Significance
Types of Error
There are two types of error:
Type 1 error – This is when you accept the
experimental/alternative hypothesis when the
results were down to chance, therefore you
should have accepted the null hypothesis.
Type 2 error – This is when you accept the
null hypothesis when you should have
accepted the alternative hypothesis.
Probability and Significance
Inferential Statistics
To be accepted as an acceptable investigation, they
must have a significance level of 0.05% or lower. This
is usually measured by statistical tests such as:
Sign test
Chi squared
Wilcoxon
Mann-Whitney
Spearman rho
Probability and Significance
Inferential Statistics
Table for remembering statistical tests
Design Type of Data
Nominal Ordinal
Repeated Measures Sign Test Wilcoxon
Matched Pairs Sign Test Wilcoxon
Independent Measures Chi squared Mann-Whitney
Correlation ------- Spearman rho
Submitting Research
Peer review
When a psychologist conducts a study, it must go
through a peer review before it is published. This
consists of another psychologist reading the report of
the experiment and then concluding whether or not it
is valid.
This is good because it makes sure only valid research
is published.
However, there is a chance that peers may be biased
and refuse to publish valid research.
There is also the risk of others reproducing the same
failed experiment because things that are not
published are not recorded.
