research methods Flashcards
1
Q
aim
A
reasonably precise statement of why a study is taking place, purpose and what’s being studied
2
Q
hypothesis
A
More precise than aim and predicts what is expected to happen
3
Q
experimental/alternative hypothesis
A
States the direction of your results, difference in conditions not due to chance
4
Q
null hypothesis
A
there will be no difference between conditions.
Results are due to chance.
5
Q
directional hypothesis
A
More precise, nature of the effect, more/less
6
Q
non-directional hypothesis
A
Nature/direction not specified
States there will be a difference
non/very little past research
past research is contradictory
7
Q
independent variable
A
Change it to see effect on DV
8
Q
dependent variable
A
Measured result
9
Q
extraneous variables
A
Another factor that has an effect on the DV that is not controlled
10
Q
participant variables
A
Any individual differences between PTS that may affect the DV
11
Q
investigator effect
A
Interaction with PTS and their behaviour towards PTS
12
Q
reliability
A
A measure of consistency
is it repeatable
13
Q
pilot study
A
Try out in a small scale trial run
Check procedures and design before wasting time and money
14
Q
random sample
A
equal chance of being selected.
+ unbiased
- not generalisable/ representative
15
Q
opportunity sample
A
easily available at time
+ easy, no money
- bias as small part of target population
16
Q
volunteer sampling
A
relies solely on volunteers
+ variety of PTs = representative
- bias as motivated
17
Q
systematic sampling
A
every nth member of target is selected from list
+ no researcher bias
- difficult n time consuming
18
Q
stratified sampling
A
strata (subgroups)
+ no researcher bias as randomly selected
- strata can’t recognise all ways PTs are different so not representative
19
Q
stratified sampling
A
strata (subgroups)
+ no researcher bias as randomly selected
- strata can’t recognise all ways PTs are different so not representative
20
Q
informed consent
A
making PTs aware of the aims, procedures, rights and what the data will be used for.
consent letter
21
Q
Deception
A
Deliberately misleading PTs/withholding information
22
Q
right to withdraw
A
Can leave at any time and withdraw data
23
Q
protection from harm
A
Not at risk
protected from all types
24
Q
confidentiality
A
Data protection act - right to have personal data protection
25
invasion of privacy
Control information about yourself fake name, blur face fake voice, censor number, please
26
debrief
Thanks.
Research purposes only
Explain aim
Info about both conditions
Check welfare
Questions
27
lab study
highest level of control over variables, technical equipment used
+ any effect on DV due to IV, replication is possible
- Like generalisability, demand characteristics
28
Field experiment
IV deliberately manipulated in a more natural setting
+ higher mundane realism than lab as natural environment, valid. PT is unaware they are being studied.
- Loss of control of extraneous variables precise replication not possible unaware of being studied = no consent
29
natural experiment
Researcher takes advantage of pre-existing IV.
+ provide opportunities that may not be undertaken for ethical reasons. High external validity - real world issues as they happen
- Little control - can’t demonstrate relationship relationships as IV isn’t manipulated
30
quasi experiment
Almost experiment, lacks. Know who to focus on but not what to change.
+ controlled conditions so any effects from DV
- Can’t randomly allocate PTS to conditions which could lead to confounding results
31
experimental design
Method of control imposed by the experimenter to control for participant variables
32
experimental group
Compared to control group to determine if the IV had an effect on the DV
33
independent groups design
PT is only take part in one condition.
+ no practice/order effects less likely to guess aim
- No control over PT variables
solution: random allocation
34
repeated measures design
Each PT takes part in all conditions of experiment
+ less PTS so more control over PT variables
- May guess aim in the 2nd condition, order effects
solution: counterbalancing 1/2 A then 1/2 B.
35
Matched pairs design
two different groups, each PT from group one paired with PT from group 2 on key variables, placed in separate groups.
+ no order effects, decreases risk of demand characteristics
- Time consuming, may not control all PT variables
solution: pilot study.
36
observational design
Structured or not, refers to the way data is collected
37
structured
Research users various systems to organise observations behavioural categories often used: event sampling, time sampling, behaviour checklist
38
unstructured
more likely to produce qualitative data small scale uses interobserver reliabiltiy
39
evaluation of observational design
Structured: easier and more systematic. quantitative so easier to compare.
unstructured: so difficult and time-consuming
Behavioural categories: need to be exclusive, observable, measurable
event sampling: when it’s quite frequent but not representative if complex.
Time sampling: reduce number of observation saving time and money
40
observational techniques
Naturalistic - behaviour observed in natural setting
Controlled - behaviour observed under controlled conditions
covert - not aware of being observed
Overt - aware of being observed
PT - researcher joins and takes part
Non-PT - research observes from outside
41
evaluation of observational techniques
Natural- high external validity. lack of control, no replication.
Controlled - low external validity but can replicate
Covert - increase validity as no demand characteristics but ethical
overt - demand characteristics ethically acceptable
Participant - increase insight increase validity however lose objectivity
Non-participant - no valuable insight but maintain objective distance
42
correlational analysis
Measuring strength and relationship between two and more variables to see a trend/pattern.
Plotted on a scattergram
Correlation coefficient - ranges from 0-1. and -1-0.
43
advantages and disadvantages of correlations
+ quick and economical
- Impossible to establish cause and effect
44
Self report techniques
non-experimental methods and include questionnaire and interviews
state and explain feelings related to the topic
45
questionnaire
Written method
closed questions - research determines range of possible answers
Open questions - doesn’t restrict range of possible answers
46
primary data
Original data collected specifically for investigation by researcher
47
secondary data
Collected by somebody else, already exist before
48
Matches of central tendency
mean - best with interval data, representative, distorted if extreme values
median - original data (ordered), not affected by extreme values, not as sensitive
Mode - nominal data, good for categories, not when more than 1 mode.
Range - direct info, only 2 most extreme so not representative, dispersion
Standard deviation - scores deviate from mean, precise, can be distorted
49
positively skewed distribution
Mode at Central point, median moves to the left
Starts positive and declines
50
negatively skewed distribution
Median moves to the right starts negative and then increases
51
Peer review
Assessment of scientific work by experts to ensure before publication.
52
evaluation of peer review
Anonymity - best to remain anonymous but some don’t to criticise
Publication bias - want a fancy title and positive results
burying research - criticise research that contradicts their own
53
evaluation of interviews
simple, minimum training, easy to analyse, fast
Interpret question wrong , bias, low response rate
54
sign test
Count up + and -
Ignore 0
Smaller of these 2 numbers = S
Compare to critical value in table at 0.05
N value is number of participants not including 0s.
If S value is equal to or less than the critical, the results are significant
55
content analysis
A way of analysing data, need a coding unit.
flexible process
Avoid ethical issues
High ecological validity
Observed bias reduces objectivity and validity
Time-consuming
56
Evaluation of case studies
used to investigate human behaviour and experiences that are rare which would be unethical to study otherwise
Offer a rich detailed insights
Difficult to generalise findings from individual cases as each one has unique characteristics
Ethical issues confidentiality many easily identified because of unique characteristics
57
Coding and quantitive data
if large data set to be analysed, info must be categorised into meaningful units
58
thematic analysis and qualitative data
Involves identifying implicit/explicit ideas within the data
A theme refers to any idea that is recurring
Themes may then be developed into broader groups
Research and then collects new set of data to test the validity of the themes
If these themes explain the new data to adequately the research will write up the final report
59
5 easy steps for content analysis
identify relevant categories
Read/watch/listen to the written/visual/verbal artefacts
Count number of times this category occurs
Identify wider themes and find examples that fit from qualitative info
Irrelevant compare behaviours before and after
60
internal reliability
A measure of the extent to which something is consistent within itself
61
external reliability
Measure of consistency over several different occasions, is it consistent over overtime?
62
split half method (internal reliability)
IQ test items are divided in half
PTs compare both (repeated measures design)
Scores on both halves are compared
If test is reliable scores on both half should be the same
Two sets of scores are correlated and the correlation coefficient calculated should be +.8 for it to be reliable
63
test retest method (external reliability)
give PT’s same test on different occasions
If test reliable results should be the same each time
Must be time between tests so can’t recall answer answers but not too long as views can change
Two sets of scores correlated and correlation coefficient calculated
64
inter-observer reliability
One researchers interpretation of events may differ to others so better to conduct in pairs
Observers need to watch the same event but record their data separate separately
The two sets of data are correlated and CC is C to see if similar
If positive and significant observers are reliable
65
how can we improve reliability?
Questionnaires: replace open questions with closed
Interview : use same interviewer. Structured use audio recordings.
Experiments : lab, test in same conditions
Observations : make sure behavioural categories have been operationalised. Audio recordings.
66
Validity
Extent to something being true
67
Internal validity
Does it measure what it’s intended to?
68
external validity
Can it be generalised beyond the research setting?
69
types of external validity
Ecological validity -Generalise from one setting to another
Population validity - generalise to other groups
Temporal validity - generalise to different time periods
70
how to assess validity
face validity - does it look like it’s measuring what it’s intended to?
Concurrent validity - compared test results of yours with a well established test on the same topic.
71
Improving validity
experimental research: use a control group. Standardised procedures.
Questionnaires : contain a lie scale. State it will stay anonymous.
Qualitative methods: use these as more in-depth
72
why do we use stats test?
To determine whether the results are significant or due to chance
To reject/accept null/one/two tailed hypothesis
73
how do we decide which starts test to use?
Is the researcher looking for a difference or correlation?
If a difference which experimental design was used
Which level of measurement was used?
74
A difference or a correlation
If a difference, often an experiment so experimental hypothesis.
If correlation , alternative hypothesis
75
experimental design used?
independent groups: take part in one condition (unrelated design
repeated measures : each PT takes part in all conditions (related design)
Matched pairs: involves use of two groups of different PTs related design
76
level of measurement used?
nominal data- categories, discreet
Ordinal data - ordered doesn’t have equal increments, lacks precision use ranks
Interval data : numerical scale units of equal size
77
Parametric test
most powerful.
must be interval data
Data should be drawn from a population expected to show normal distribution for the variable being measured.
Should be homogeneity of variance - set of scores in each condition should have similar dispersion
78
critical value
The rule of R
if R, calc >_ critical = significant
if no R, calc <_ critical = significant
79
writing the results of a stats test
The result_ significant.
The calculated value of _ is_ the critical value of_.
When using a _ hypothesis and a_ 0.05 significance level and N/df is_.
Therefore the _ hypothesis will be accepted and the _ hypothesis is rejected.
80
type one error
False positive.
Research accepts the one/2 tailed hypothesis and rejects null but should’ve been another way.
Think they’ve found significance but haven’t.
81
type two error
False negative.
Researcher accepts null and reject one/2 detailed hypothesis but should’ve been another way.
They think they haven’t found significance, but have
82
1. abstract
short summary at the start.
Includes aim, hypothesis, method, results and conclusion.
Psychologist read this to determine whether to examine further
83
2. introduction.
Review of relevant theories and studies linked to your research.
84
3. method
in detail so can be replicated.
Design, sample, ethics, materials, procedure and method
85
4. results
descriptive statistics, measures of tendency and dispersion.
Data table , stats tests, content analysis.
86
5. discussion (conclusion)
link results to research in introduction, identify improvements and limitations of the study.
Consider wider impacts on society
87
6. referencing
List of all researchers mentioned in above to give credit
includes full details of their work to avoid plagiarism and people can easily find the original source
