Research methods Flashcards
DATA HANDLING
BOOKLET 2
What are the different measures of central tendency?
-mean
-medium
-mode
What’s the mean?
Adding up all values and dividing it by total amount of data available
strength / limitations of mean
+
representative of all data as it includes all values
-
Distorted by extreme values
Medium?
the middle value in data when arranged in order
strength / limitation of medium?
+
extreme values do not affect it
-
not all values are included (highest + lowest not taken into account)
Mode?
The most frequently occurring value
strength / limitation of mode?
+
easy to identify / calculate
-
crude measurement - may not be representative of all data
sometimes more than 1 mode
Measure of dispersions?
how spread out the data are
-Range
-Standard deviation (SD)
Range?
Minusing the lowest values from the highest
Strength / limitations of range?
+
easy to calculate
- only takes into account extreme values
(the highest + lowest = not fully representative of spread of all scores)
Standard deviation (SD)?
Single value that tells us how far scores are deviated (spread out) from the mean.
the higher the SD = greater the dispersion/ spread out the data is
Strengths / limitations of SD?
+
more precise measure pf dispersion as it includes all values
better than range as it less affected by extreme values
-
takes longer to calculate
Display of quantitative data?
Graphs - way of displaying data to see trends/patterns in data
-table
-bar chart
-scatter gram / scatter graph
-histogram
-line graph
Table?
shows descriptive statistic as well as raw scores
Bar chart
Type of graph which the frequency of each variable is represented by the height of the bars
Scatter graphs
represents the strength + direction of the relationship between co-variables in a correlation analysis
Histogram
Displays the distribution of a whole continuous data set
there is no space between the columns like a bar chart
Line graph
Displays continuous data and uses points connected by lines to show how something changes in value over time
Distributions
types of distributions?
normal distribution
positive skewed distribution
negative skewed distribution
normal distribution
- the mode, median, mean are all equal
- most values near middle
- the graph is symmetrical
Positive skew distribution
- The mean is HIGHER than the median/mode
- long tail is on (positive) right side of peak
-the curve is on the left
Negative skew distribution
- the mean is LOWER than the median/mode
- Long tail is on the (negative) left side of peak
-the curve is on the right
DESIGNING PSYCHOLOGICAL INVESTIGATIONS
BOOKLET ONE
before research is done a aim (purpose) and hypothesis (prediction) is needed:
- Experimental hypothesis?
- Alternative hypothesis?
- Null hypothesis?
- An experimental hypothesis predicts what change(s) will take place in the dependent variable when the independent variable is manipulated.
- The alternative hypothesis states that there is a relationship between the two variables being studied (one variable has an effect on the other).
- The null hypothesis states that there is no relationship between the two variables being studied (one variable does not affect the other).
Experimental and alternative hypothesis can be either directional or non-directional.
- Directional hypothesis?
- Non-directional hypothesis?
- A directional (one-tailed) hypothesis predicts the the effect of the IV on the DV, AND in which direction the change will take place. (i.e. greater, smaller, less, more)
E.g. adults will correctly recall more words than children.
- A non-directional (two-tailed) hypothesis predicts that the IV will have an effect on the DV, BUT the direction of the effect is not specified. It just states that there will be a difference.
E.g. there will be a difference in how many numbers are correctly recalled by children and adults.
what’s a pilot study and its purpose?
A small scale trial of the study before it takes place, where potential problems are identified / target behavior categories are identified before
so they can be fixed before so that the study/observation/interview can occur more smoothly
Types of observations + evaluation:
- Participant observation
observer becomes part of the group they’re studying
+ can experience situation as it is and get valuable insight
-lose objectiveness and get bias by identifying with them too much
- Non-participant
researcher remains separate from those they are studying + record behavior in a more objective manner
+ remains objective
-loses valuable insight into the lives of ppl they’re observing
- covert
the ps are unaware they’re being observed (no consent)
+ removes demand characteristics = higher internal validity
- problems with ethics as they have no consent
- Overt
the ps know they’re being observed
+ more ethical as they give consent
- demand characteristics = low internal validity
- naturalistic / unstructured
takes place in setting / context where the target behavior would normally occur
aspects of environment are free to vary
+ high external validity - can apply to everyday situation
- lack of control over situation and extraneous variables = difficult to replicate and low internal validity
- controlled / structured
watching / recording behavior within a structured environment with variables controlled
+ high control over extraneous variables = easy to replicate and higher internal validity
- more demand characteristics and cant generalize to everyday situations - low external validity
Behaviour categories?
target behaviour is broken down into components that are observable and measured.
the 2 sampling methods for observing behaviour categories?
event sampling: target behaviour is first established and then researcher records this each time it occurs
time sampling: records the behaviour in a fixed time frame
issues with observational designs:
- observer bias
- when the observer is actively looking for certain behaviour so more likely to ‘see’ the, + record them
- only notice events that confirm their opinions / hypothesis
how can it be corrected?
Inter observer reliability:
- carried out by 2 researchers at least
- record separately
- compare observations they got after
- find correlation
=reduces bias and pick up on more detail
issues with behaviour categories
- must be observable + measurable + self-evident
- categories must be exclusive and not overlap
- ensure all possible forms of that behaviour included
observational design AO3
- event sampling useful for behaviour thats infrequent and easy to miss
- time sampling reduces amount of observations needed
- might be unrepresentative of the entire time as they only measuring behaviour in a specific time period
Experimental designs / research design
(the way the ps are used in experiment)
-what are the 3 types?
- independent group design
- repeated measure design
- matched pair design
independent group design
different ps in each condition
+order effect not problem
-participant variables may be problem
-2x many ps as repeated measure
-more time and money needed
repeated measure design
same ps are used in both condition.
can do counter-balancing or randomisation
+ps variables controlled
+less ps needed than independent group designs
-order effect present ( guess aim/ get better by practice/ tiredness from doing 2 conditions)
matched pair design
ps matched in pairs on a shared variable (IQ/age) amd 1 of each pair go to each condition
+no order effect
+Ps variables reduced
-time consuming
-can never be fully matched on every variable so some ps variable still present
sampling
(methods used to select ps from target population)
what are the 5 types of sampling techniques?
- random sampling
- systematic sample
- stratified sample
- opportunity sample
- volunteer sample
random sampling and AO3
randomly select ps (generator / picking names in hat)
+no bias
-time consuming
systematic sample and AO3
selecting names at regular intervals from sampling frames (EG every 4 names)
+objective and no bias
-time consuming
stratified sample and AO3
stratify target population into groups (AGE/CULTURE/GENDER) and randomly select from each strata.
+representative sample = reflects composition of population
-identified stratas wont reflect all ways ppl are different
opportunity sample and AO3
using anyone that’s available at the time to take part
+convenient
+less time and money needed than random
-unrepresentative of target population
-researcher has control over selection (bias)
Volunteer sample and AO3
ps self select / volunteer themselves
+easier + requires little input from researcher
+less time consuming
+ps more engaged as they volunteered
-volunteer bias ( asking for ‘specific’ characteristics in volunteers)
what’s correlation analysis ?
- relationship between 2 co-variables
- measures strength and direction of an association between 2 co-variables
- plotted on scattergram
positive correlation v negative correlation
no correlation?
Positive correlation
: as one co-variable increases so does the other
+ 1 = perfect positive correlation
+ 0.8 = positive correlation
Negative correlation
: as one co-variable increases the other decreases
-1 = perfectly negative correlation
Case studies?
a detailed study of a single individual / situation etc
AO3 of case studies
+ provides rich source of data
+ specific to that specific individual / scenario they’re studying
-difficult to generalise the results - have low population validity
-difficult to replicate case studies = hard to examine the reliability of findings
analysis + interpreting qualitative data
- the 2 ways?
content analysis
thematic analysis
content analysis?
- research techniques where ppl are studied indirectly through their communication and words
- coding system is established
- count number of times a word/phrase appears in the qualitative data = turns to quantitative data instead
AO3 of content analysis
+data already exists so don’t need to get permission to analyse it
-may suffer from lack of objectivity
(esp when more descriptive forms of thematic analysis used)
Thematic analysis
- Identify implicit /explicit ideas within the data that’s recurring
- Themes emerge once data has been coded
- theme = a recurring idea
- developed into broader catogeries
AO3 of thematic analysis
+high external validity
-may suffer from lack of objectivity - it’s subjective
(esp when more descriptive forms of thematic analysis used)
Reliability ?
internal reliability ?
external reliability ?
reliability - consistency
internal - the consistency within the test
external - producing the same results each time the test is carries out
ways of assessing reliability
(cheching)
- split half method
- inter rater reliability / inter observer reliability
- test-retest
split half method?
this compares one half of the test with the other to check whether the scores of a variable are consistent
inter rater / observer reliability?
more than 1 observer observing same thing
- record data separately
- test of correlation between 2 sets of scores
test - retest?
- same test to same person given on different occasion
- sufficient time left between the 2 tests
- a test of correlation between the 2 sets of scores
= +0.8 or more
Ways of improving reliability?
- interviews
2.questionaires
- Interviews:
- use same interviewers
- ensure all interviewers are properly trained
-use structured interviews - Questionnaires:
- test-retest method
-to get 0.8+ or more
- observations
- experiments
- observations:
- inter - observer method
-use specific behaviour categories
/ operationalize behaviour categories - Experiments:
- use standardised procedure
- replicate procedure
validity?
internal validity?
external validity?
validity - accuracy
Internal - the extent to which you measure what you set out to measure - the effect of IV on DV
external - how well the results can be generalised beyond the study
the types of external validity?
ecological validity - extent to which findings can be generalised to other situations
population validity - how far the sample of ps is representative of rest of population
mundane realism - the degree to which the setting or procedure reflects that in real life
experimental realism - the degree to which the results reflect realistic behaviour
Ways of assessing / testing validity?
Face validity - does it appear valid?
- just by looking at it should appear valid + pass to experiment to check
concurrent validity - do the results from a test match up with another well recognised test
- test of correlation - +0.8 or more
ways of improving validity
- Questionnaires
- Interviews
Questionnaires:
- use lie scale in questions to assess consistency of responses they give
-make it anonymous - less socially desirable answers + bias
Interviews:
- interpretative validity - extent of which the finding of the interviewers match the ps
- use different sources as evident ( triangulation)
- Experiments
- Observations
Experiments:
-standardised procedure
-blind trials - ps unaware of procedure / hypothesis
-double blind - neither ps or experimenter know
Observations:
- covert observations (uncover)
-Behaviour categories should not be too broad / overlapping
STATISTICAL TESTS
BOOKLET
levels of significance
what % should psychologists be at least certain for it to be valid? drug trials/medicine?
= what’s the confidence level used in psychology?
what does it mean
psychologists need to have at least a 95% confidence level / significance
99% for medical science / drug trial
level of significance used - 0.05 / 5%
written as p ≤ 0.005
= there’s only a 5% or less likelihood that results are due to chance
psychologists use a 0.05 level of significance as its half way between type 1 + 2 error
if null hypothesis rejected?
if null hypothesis accepted?
null hypothesis rejected = results are significant
(accept alternative hypothesis)
null hypothesis accepted = results are insignificant
(reject alternative hypothesis)
Errors - wrong hypothesis is accepted
( 5% possibility )
Type 1 error?
- Null hypothesis is rejected when it should’ve been accepted
- accepted the alternative hypothesis instead
(should’ve rejected it)
= too lenient
- claim to have found significant difference when they acc haven’t
Type 2 error?
- Null hypothesis is accepted but it should’ve been rejected
- rejected the alternative hypothesis instead (should’ve accepted it)
= too strict
What’s a false positive error?
False positive error - TYPE 1 ERROR
- Rejected the null hypothesis when should’ve accepted it
- too lenient - optimistic
( 0.1 / 10% )
What’s a false negative error?
False negative error - TYPE 2 ERROR
- Accepted the null hypothesis when should’ve rejected it
- too strict / pessimistic
( 0.01 / 1% )
What has to be considered when deciding which statistical test to use?
- Looking for difference or relationship?
- What’s the experimental design?
(independent group design [unrelated designs]
/repeated measure / matched pair design [related designs]) - what’s the level of measurement?
( nominal / ordinal / interval )
Levels of measurement
- Nominal data
- Ordinal data
- Interval data
- Nominal data
- data represented in categories (eg females + males)
- (cant be put in rank order as there’s no scale measurement) - Ordinal data
- data put in rank order of size
- but there’s no scale of measurement with equal units - Interval data
- data put on a scale that has equal units (cm / seconds)
- precisely defined size
- most precise form of data
When is a sign test used?
- Test for difference
- Nominal data
- Repeated measures
How to do the sign test?
- Calculate the difference between the 2 sets of scores of the ps
- calculate how many ps increased [ + ]
calculate how many ps decreased [ - ] - how many ps didn’t increase /decrease - minus this number from the total number of ps later when looking at the table
- Find the OV (observer value) - the number of the less frequent sign [ - / + ]
- compare with the critical value ( cv)
= OV needs to be equal or less than the CV for findings to be significant
Statistical tests:
- Spearman’s Rho
- predicts correlation
- data is related ( comes from same person )
- level of measurement: Ordinal / Interval / ratio
- Needs number of Ps
- Significant = OV must be greater / equal than CV
= Null hypothesis rejected
- Pearson’s r
- Predicts correlation
- Data is related
- level of measurement: Interval data
- Significant = OV must be greater / equal the CV
= Null hypothesis rejected
- Chi squared
- Predicts a difference between 2 conditions
- data is unrelated / independent
- Independent group design used
- level of measurement: Nominal data
- Degree of freedom needed
- Significant = OV must be greater / equal the CV
= Null hypothesis rejected
- Sign test
- predicts a difference between 2 sets of data
- data is related (comes from same person)
- Repeated measure / Matched pair design used
- level of measurement: Nominal data
- Significant = OV must be less / equal the CV
= Null hypothesis rejected
- Wilcoxon T
- Predicts a difference between 2 sets of data
- Data is related
- Repeated measure design used
- level of measurement: Ordinal / Interval / Ratio
- Significant = OV needs to be less / equal the CV
= Null hypothesis rejected
- Man Whitney U
- predicts difference
- data is unrelated / independent
- independent group design used
- level of measurement: Ordinal / Interval / Ratio
- number of ps in each group needed
- Significant = OV needs to be less / equal the CV
- Unrelated T test
- predicts a difference
- data is unrelated / independent
- Independent group design used
- level of measurement: Interval
- Significant = OV must be greater / equal the CV
- Related T test
- predicts a difference
- data is related
- repeated measure used
- level of measurement: Interval
- Significant = OV must be greater / equal the CV
Statement of significance - explaining why findings are significant / insignificant
what needs to be included?
- the number of ps
- type of hypothesis (1 or 2 tailed )
- level of significance
- OV
- CV
- If the OV is less/greater/equal the CV
- if the null hypothesis can be rejected / accepted
Referencing (last page)
what does it include?
- Author. Surname followed comma + the first initials.
- Year of publication of the article (in bracket). End with full stop.
- Book title (in italics ). Capitalize first letter of the first word. End with full stop.
- Edition (in brackets), if other than first.
After the title but before the period. - Name of the publisher, full stop.
Include a DOI
Author, A. (Year) Title of article. location of publication. publisher.
What are the different sections of reporting a psychological investigation?
1.Abstract
2.Introduction
3.Method
4.Results
5.Discussion
6.References
- Abstract?
first section in a psychological report / journal (always written last).
Summary (150 words) of the:
* Aims
* Hypothesis
* Method
* Results
* Conclusions
Provides an:
I. overview of the entire report
II. efficient way of gaining information without having to read an entire study
- Introduction
- Explains why the study is being carried out
- Researcher reviews previous research (theories + studies) to provide background information + a rationale for the current research.
- info = moving from the general to the specific
- Method
Contains a detailed description of the methodology (allows for replication)
Includes details of the:
* design
* Sampling method/ participants
* apparatus/materials
* procedure
* ethical considerations
- Results
details about what they found in their investigation.
It includes:
* descriptive statistics ( tables, graphs, measures of central tendency and dispersion)
- inferential statistics (e.g. results of statistical tests, including calculated values and significance levels).
- If qualitative research has been conducted, the results section would involve description of the categories and themes, along with examples.
= finishes with the rejection / acceptance of the null hypothesis.
- Discussion
- interprets the results of the study
- makes criticism of the methodology used
- considers the implications of the results for future research
- suggests real-world applications
- Referencing
Includes full details of any sources, such as:
* journal articles
* books
that are used when writing a report.
What’s the appendix?
- Contains the material too bulky for the body of the report ( consent + debriefing forms, instruction sheets or stimulus materials and raw data and calculations.
- These are numbered so they can be referred to in the text and easily found by someone reading the report