Paper 2- Research methods Flashcards
Aims
Outlines the research topic.
E.g. ‘To investigate differences in mathematical ability between genders’.
Always start with ‘To investigate’.
Hypothesis
A statement that predicts the outcome of a study.
Could be directional (one-tailed) or non-directional (two-tailed).
Alternative hypothesis
Directional- States which way they think the results will go.
E.g. ‘Boys will score higher on the maths test than girls.’
Non-directional- States there will be a difference but not what the difference is.
E.g. ‘There will be a difference in maths test scores between boys and girls.’
Null hypothesis
States there will be no difference or any difference is down to chance.
Accepted if the results aren’t significant.
E.g. ‘There will be no difference in maths test scores between boys and girls.’
E.g. ‘…Or any difference will be down to chance.’
Independent variable
The thing that is manipulated/changed by the experimenter.
E.g. The different groups/ different conditions.
Dependent variable
The variable that is measured by the experimenter.
Operationalisation
Explaining how the variables could be changed/measured.
Correlational hypothesis
Not and IV and DV.
Co-variables- two things that are compared for a relationship.
E.g. ‘There will be a positive correlation.’
E.g. ‘There will be a relationship between crime rate and temperature at different times of year.’
What is a target population?
The people that the researcher is interested in.
E.g. People with Sz.
Why do we use a sample?
There are too many people in a population to research them all.
What is a sample?
The people the researcher uses in their study.
The participants.
A small group that is supposed to represent a population.
What is a sampling technique?
The way a researcher selects their participants.
Random sample
Each participant has an equal chance of being selected.
E.g. Name pulled out of a hat.
Opportunity sample
Asking people who are available at that time to take part.
E.g. Researcher might ask parents picking up their children from school.
Volunteer sample
The researcher advertises the study and the people who see the advert can get in contact and take part.
E.g. Local newspaper, poster.
Systematic sampling
Selecting every nth person from a pre-selected list.
Stratified sampling
Selecting people form every proportion of your population- in the same proportions.
Sampling evaluation
…
Define experiment
An experiment involves a change in an independent variable.
The researcher will record or measure the effects of this on the dependant variable.
How the IV is manipulated and under what circumstances varies with the type of experiment.
What are the 4 types of experiment?
- Lab
- Field
- Natural
- Quasi
Lab experiment
- Controlled in an artificial environment.
- Independent variable is manipulated.
- Participants are randomly assigned to conditions.
Field experiment
- Natural environment.
- Independent variable is manipulated.
Natural experiment
- Independent variable is not manipulated
- It is unplanned and has occurred because of a naturally occurring event.
- Could be a natural or controlled setting.
Quasi experiment
- Independent variable is not manipulated- it is based on existing differences between people. E.g. gender, age, personality.
- There is planned manipulation of this naturally occurring IV.
- Could be a natural or controlled setting.
What are self-report methods?
Both questionnaires and interviews are types of self report method. This is because the participant reports their own thoughts and feelings about a particular matter.
What are questionnaires?
- Written questions
- Opened or closed
Open question
The participant can give any answer they wish.
Qualitative
Closed question
There are a set number of responses that a participant selects from.
Quantitative
Types of closed questions
Fixed Choice Option – includes a list of possible options and respondents are required to indicate those that apply to them.
INSERT
Likert Scale – the respondent indicates their agreement (or otherwise) with a statement using a scale (ranging from strongly agree to strongly disagree).
INSERT
Rating Scales – participants select a value that represents their strength of feeling about a particular topic.
INSERT
Evaluation of questionnaires
Some common issues…
Overuse of jargon – using technical terms that are familiar to the person writing the questionnaire that the respondent may not understand.
E.g. ‘Do you agree that maternal deprivation in infanthood inevitably leads to affection-less psychopathy in later life’
Emotive use of language – the authors attitude comes across in the words they select.
E.g. ‘Do you agree that boxing is a barbaric sport and any sane person would want it banned.’
Leading questions – the phrasing of the question indicates a particular response.
E.g. ‘Is it not obvious that student fees should be abolished?’
Double barrelled questions – contains two questions in one (they may agree with one half but not the other).
E.g. ‘Do you agree that premier league footballers are overpaid and should have to give twenty per cent of their wages to charity.’
Double negatives – when two forms of negation are used in the same sentence (can be difficult to decipher).
E.g. ‘I am not unhappy in my job (agree/disagree).’
Interviews
- Spoken questions
- Could be structured, unstructured or semi-structured
Structured interviews
Just like questionnaires, except face-to-face or on the phone.
Contain standardised pre-set questions.
Often use a computer with the pre-set questions on – CAPI (computer assisted personal interviewing)
Sometimes includes a list of pre-determined answers (quantitative data).
Unstructured interviews
A conversation.
The interviewer has a general idea of the topics they want to discuss but the actual questions and sequence of questions develop during the course of the interview.
The interviewer is able to probe for deeper answers and follow interesting avenues that may come up.
Semi-structured interviews
Still has a list of issues/questions.
However, questions can be asked in any order.
If something interesting comes up the interviewer can veer away from the standardised questions.
Questions generally open-ended but data can also be collected.
Evaluation of interviews
Structured - the data that is produced is numerical- easier to analyse
Unstructured - Quantitative data- more difficult to analyse and record. However, more depth and detailed data collected.
Naturalistic/controlled observation
Naturalistic Observation:
A research method carried out in a naturalistic setting, in which the investigator does not interfere in any way but merely observes the behaviour in question.
Controlled Observation:
Observing behaviour under controlled conditions.
Overt/Covert observation
Overt Observation
The participants are aware that they are being observed.
Covert Observation
The participants are not aware that they are being observed.
Structured/Unstructured observation
Structured Observation:
The researcher determines precisely what behaviours are to be observed and uses a standardised checklist to record the frequency with which they are observed within a specific time frame.
Unstructured Observation:
The observer recalls all relevant behaviour but has no system.
Participant/Non participant
Participant Observation
The researcher gets involved with participant activity so they can experience it for themselves, joins in.
Non participant Observation
The observer remains separate from the participants to maintain objectivity.
Conducting an observation
To conduct an effective observation a psychologist may use behavioural categories.
These are used in structured observations as a check list
The target behaviour is broken down into behavioural categories and then operationalised.
E.g. when observing infant behaviour, behavioural categories could be smiling, crying, sleeping etc.
Continuous recording
All instances of target behaviour are recorded.
Event sampling
Counting the number of times a particular behaviour occurs in a target individual or group, doesn’t take account of time – just a tally.
Time sampling
Recording behaviour within a pre-established time frame e.g. take note what a target individual is doing every 30 seconds of some other time interval.
Correlation
When two things are measured in order to identify if there is a relationship between them.
A single numerical value is produced that is used to describe the relationship.
There are four possible outcomes – positive correlation, negative correlation, no correlation or curvilinear correlation.
Positive correlation- Both variables increase together.
Negative correlation- As one variable increases, the other decreases.
No correlation- No relationship between the variables.
Curvilinear correlation- The relationship is predictive although it is not linear but curved.
Correlation co-efficient
Number between 0 and 1.
Tells us how strong the correlation is – the nearer to 1 the stronger.
It has a plus or minus sign in front of the number which tells us whether the correlation is positive or negative
E.g. -0.85.
INSERT PICTURE
Correlation or experiments
In an experiment the researcher controls or manipulates the independent variable.
In contrast, in a correlation there is no manipulation of the two variables.
It is therefore not possible to determine cause and effect.
Experimental design
How psychologists organise their groups of participants.
- Independent groups
- Mathced pairs
- Repeated measures
Independent groups
A separate group of participants for each condition of the IV.
E.g. IV music/no music = one group has music playing, a different group has no music.
Matched participants
A separate group of participants for each condition of the IV but they are fitted for certain characteristics.
E.g. IV gender = separate group of males and females but I will make sure that the groups are matched for age and income.
Repeated measures
Every participant completes all conditions.
E.g. one group has music playing then the same group does another test without music.
Cannot be used if the IV is gender, age etc.
Evaluation of experimental design
Independent:
+ No demand characteristics
+ No order effects
- Individual differences
Matched:
+ No individual differences
- Demand characteristics
- Order effects
Repeated: \+ No demand characteristics \+ No order effects \+ Individual differences minimised - There will always be some individual differences
What is a pilot study?
This is sometimes conducted to test the design.
It is also conducted to test the measures used.
It can be used to test for reliability…to test and re-test.
It is also used to identify extraneous variables so controls can be put in place for the actual study.
It is also used to ensure all the ethical issues have been dealt with.
What are ethics?
Ethics can be defined as ‘the consideration of what is acceptable or right behaviour in the pursuit of a personal or scientific goal’ (Cardwell, 2000).
What are the ethical guidelines?
- Informed consent
- Right to withdraw
- Deception
- Confidentiality
- Protection of participants
How do we gain informed consent?
Participants should be issued with a consent form detailing all relevant details that might affect the participant to participate in the research.
If the participant agrees, this is signed.
For investigations involving under 16 year olds, a signature of parental consent is required.
Name the three alternative ways of getting consent if it is impractical to get informed consent
1) Presumptive consent
2) Prior general consent
3) Retrospective consent
How should a researcher deal with deception and protection from harm?
- Full debrief
- Right to withdraw
- Counselling
How to deal with confidentiality
- Maintain anonymity
- Protection of data (reassurance of this)
Validity
- Accuracy
- External validity- can we accurately generalise.
Internal validity
Are we measuring what we set out to measure?
If there are extraneous variables this is lowered - because we are no longer testing the effect of the IV and DV.
Extraneous variables
‘Nuisance variables’
- These affect the DV but don’t vary systematically with the IV (it is a random error- might not affect everyone in the same way)
- e.g. temperature of room, participants mood etc.
Confounding variables
Affect the DV and do vary systematically with the IV (affects everyone in the same way).
e.g. if all the participants in condition 1 were more intelligent than those in condition 2 (these variables have a direct impact on the DV - they confound the findings of the study).
External validity
Population validity - Is our sample representative?
Ecological validity - Is the environment accurate to real life?
Validity over time - Is the experiment still accurate to todays society?
Demand Characteristics
- Type of extraneous variable
- Difficult to control
- Participants may guess the aims of the research and then may act in a way that they think is expected.
Investigator effects
- Unwanted influence of the researcher on the experiment.
- This may be unconscious behaviour, such as smiling more with one condition compared to another.
Experimenter bias – when the experimenter effects the results e.g. through their interpretation, through body language, facial expressions, the way they speak etc.
Interviewer bias – when the interviewer affects the responses of the interviewee
e.g. Greenspoon effect – when the interviewer makes affirmative noises e.g. mmmmhhhmmmm after certain answers, this affects the way the participant responds i.e. they think ‘that must have been a good answer, I will try to give more like that’
Participant reactivity
Hawthorne effect – when the added attention of being in a study affects participant behaviour e.g. they may be shy or show off
Demand characteristics – when participants think they have figured out the aims of the experiment and change their behaviour
Social desirability bias – when participants try to look good by answering/behaving in a socially acceptable way.
Controlling variables
Randomisation:
The use of chance wherever possible to reduce the researchers influence. E.g. randomly assigning participants to conditions, randomly generating the order of a list of words for a memory experiment etc.
Standardisation:
All participants should be subject to the exact same process – the only thing that should be different is manipulation of the IV. E.g. written instructions, time limits, doing everything in the same order etc.
Counterbalancing:
Control for order effects, half do condition one first while the other half do condition two, then they swap.
Single blind design:
The use of deception, misleading the participants.
Double blind design:
When both the participant and the researcher are unaware of the aims of the study.
Reliability
- Consistency
- If they did the test on another day, would they get the same results?
- Is there standardised procedures and instructions i.e. is there consistency in the way the experiment is conducted?
Inter-rater reliability
Are the observers scoring in the same way?
How can you check reliability?
1) Conduct the test again and see if you get the same results
2) Conduct a spearmans rho test comparing the scores:
Testing for a correlation (if there is no correlation between the observers etc then it is not consistent so not reliable).
How can you improve reliability?
- Observers familiarise themselves with behavioural categories.
- Conduct a small scale pilot study.
- Compare the data observers have gotten by calculating a correlation coefficient.
- Operationalise variables if needed.
- Repeat.
- Randomisation - The use of chance to reduce the researchers influence eg randomly assigning participants to conditions.
- Standardisation - All participants should be subject to the exact same processes eg written instructions, time units.
- Counterbalancing - Control the impact of order effects, half do condition 1 while half do condition 2, then swap.
- Single blind - the use of deception, misleading the participants.
- Double blind - When both the participant and the researcher are unaware of the aims of the study, e.g. doctors doing drug testing using a placebo should be unaware who took the drug as it could lead to experimenter bias. (Usually more than one researcher)
What is peer review?
- Also called ‘refereeing’.
- Peers assess the scientific work of others who are in the same field before it can be published.
3 main purposes:
- Allocation of research funding.
- Validate the quality and relevance of research.
- To suggest amendments or improvements.
Why do we need peer review?
Sir Cyril Burt:
- Falsified his findings.
- Claimed that intelligence was inherited.
- This led to 11+ test which determined which school everyone went to.
- Kamin challenged him.
- He was exposed as a fraud.
- Move away from grammar schools and 11+ test.
Peer review evaluation
+ Essential so that high quality research is produced.
+ Keeps a check on dishonest psychologists.
- Expensive
- Time-consuming
- Bias
- Subjective
- Publication bias - Editors of journals want to publish significant ‘headline grabbing’ findings to increase credibility. This could mean that research which does not meet these criteria is ignored. This creates a false impression of the current state of psychology if journal editors are being selective in what they publish.
- Anonymity - ‘peer’ is usually anonymous. Reviewers may use their anonymity to criticise rival researchers. This is also made more likely by the fact that many researchers are in competition for funding. This can be avoided by making the reviewer public.
Types of data
Quantitative - Anything numerical
Qualitative - Data that is not numerical, expressed in words e.g. observation notes, interview responses etc.
Primary - Data that is collected by the researcher themselves
Secondary - Data that has been collected by someone other than the person doing the research e.g. statistics, meta-analysis etc.
Meta-analysis – Collates findings from several studies that have already been done.
Types of data evaluation
INSERT PICTURE
Mathematical symbols
= < << > >> ~ directly proportional to?
Measures of central tendency
…find the average
Mean - sum of values divided by the number of values
Mode - most frequently occurring number
Median - the middle value
Measures of dispersion
…find the spread of the data
Range - highest value minus the lowest value
Standard deviation - calculates how far scores deviate from the median
Strengths and weaknesses of different measures
Mean:
+ uses all of the data in the calculation so is more representative of the data as a whole.
- includes extreme values.
Mode:
+ not affected by extremes
- isn’t representative of the data as a whole
Median:
+ not affected by extreme values
+ can be used when data isn’t interval (units of variable sizes)
- median may not be in the data set
- isn’t representative of the data as a whole
Range:
+ includes all data pieces
- only takes the most extreme values into account
- is unrepresentative of the data as a whole
Standard deviation:
+ all of the data is included in the calculation so is more representative of the data as a whole.
- is affected by extreme values
Bar graph
- Used for data in discrete categories. The bars are separated by a gap to show they are not continuous.
- A bar chart should not be used to plot individual participant scores but the total or mean or percentage scores for each group.
- The DV goes on the y axis and the IV goes on the x axis.
Histogram
- Used for continuous frequency data.
- The bars are touching to show that the data is continuous.
- The x axis is made up of equal sized intervals of a single category. The y axis represents the frequency.
- Sometimes a frequency polygon is drawn by joining the midpoints at the tops of the bars.
Line graph
- Used for continuous frequency data.
- The x axis is made up of equal sized intervals of a single category. The y axis represents the frequency.
- Useful for comparing two sets of frequency data on one graph would not be easy to see on a histogram.
Scatter graph
- Used for correlational data.
- The co-variables go on the axes.
- The dots are not joined but sometimes a line of best fit is drawn.
How to identify and draw appropriate graphs for certain data sets
INSERT PICTURE
Normal distribution
- Bell shaped curve.
- The mean, mode and median all lie at the midpoint.
- Most scores occur around the middle with fewer being clustered as they occur above and below the mean.
- The tails of the curve, which extend outwards, never touch the horizontal x axis as more extreme scores are always theoretically possible.
Positively skewed distribution
- A positive skew is when the long tail is on the positive side of the peak, and some people say it is “skewed to the right”.
- Most scores fall below the mean.
- The mean, mode and median are not in the same place - the mean gets pulled to the right because it is affected by extreme values.
Negatively skewed distribution
- A positive skew is when the long tail is on the positive side of the peak, and some people say it is “skewed to the right”.
- Most scores fall below the mean.
- The mean, mode and median are not in the same place - the mean gets pulled to the right because it is affected by extreme values.
How could you improve validity for questionnaire, case-study, experiment, observation.
Questionnaire - a lie scale, assure participants its anonymous.
Case-study - check with ps you have understood how they feel
Experiment - use standardised procedures
Observation -
What is a case study?
A case study is an in-depth investigation, description and analysis of a single individual, group, institution or event.
Often involve analysis of unusual individuals or event. E.g. rare disorders, also typical cases.
Characteristics: longitudinal, qualitative.
Case history of individual - interviews, observations, gather data from friedns and family.
Evaluation for case studies
+ The case study method offers rich, in-depth data.
Allows us to find out so much more about an individual.
Includes subjective experiences that quantitative data doesn’t show.
+ Case studies can be used to investigate rare behaviour and experiences.
Avoids ethical issues as case studies are naturally occurring and the researcher doesn’t inflict damage.
-
-
Content analysis
A type of observational research where people are studied indirectly via the communications they use e.g. emails, letters, media.
The aim is to summarise the communication in a systematic way so conclusions can be drawn.
The researcher uses behavioural categories to count the number of times something occurs.
Thematic analysis
- A technique when analysing qualitative data.
- Themes or categories are identified and then data is organised according to these themes.
- Thematic = qualitative data produced
- Content = either but generally quantitative
What are inferential tests?
- Draw conclusions about our data.
- Tell us whether our results are significant enough that we can generalise with any certainty.
- Based around probability – assess the probability that the results could just be down to chance – if there is a low probability of this then we can generalise.
What does significance mean?
- If a test shows our results are significant we accept our alternative hypothesis.
- If they are not significant we accept our null hypothesis.
- A test is significant if it meets the level of probability we have chosen.
Levels of significance
p≤0.05 = 5% level of chance (only 5% probability results are down to chance).
p≤0.01 = 1% level of chance (only 1% probability results are down to chance).
Type 1 and type 2 error
- Type 1 error – The alternative hypothesis is accepted when the null hypothesis should have been accepted.
- Type 2 error – the null hypothesis is accepted when the alternative hypothesis should have been accepted.
Chi square
When would you use a chi square test?
- ) Tests for a difference or association
- ) Level of measurement is nominal
- ) Independent groups design
- It begins with a contingency table
- You will be told the observed value of chi square (²)
- You will then be asked to find the critical value of chi square from a table.
- Degrees of freedom – calculated by (no. of rows – 1) x (no. of columns – 1)
- If the final observed value of chi-square is greater than the critical value the test is significant and we can accept the alternative hypothesis.o
When would you use a spearman’s rho test?
1) tests for a correlation
2) level of measurement is ordinal
3) correlational design
The observed value of rho is the correlation coefficient
•The test then uses a table (table of critical values) to calculate the critical value of rho.
•We also need to know how many participants there are in total.
•If the final observed value of rho is greater than the critical value the test is significant and we can accept the alternative hypothesis – but!…check the direction is the same as in the hypothesis (i.e. positive/negative correlation)
When would you use a Pearson’s R test?
1) test for acorrelation
2) level of measurement is interval (parametric)
3) correlational design
What do you need to know?
•You will be told the observed value ‘R’.
•The test then uses a table (table of critical values) to calculate the critical value.
•We also need to know how many participants there are in total and calculate degrees of freedom when looking at the critical values table. (N- 2= df)
•If the final observed value of R is greater than the critical value the test is significant and we can accept the alternative hypothesis – but!…check the direction is the same as in the hypothesis (i.e. positive/negative correlation)
Unrelated T test
1) test for a difference
2) interval data
3) independent groups design
What do you need to know?
•You will be told the observed value ‘R’.
•The test then uses a table (table of critical values) to calculate the critical value.
•We also need to know how many participants there are in total and calculate degrees of freedom when looking at the critical values table. (N- 2= df)
•If the final observed value of R is greater than the critical value the test is significant and we can accept the alternative hypothesis – but!…check the direction is the same as in the hypothesis (i.e. positive/negative correlation)
Related t test
- ) Tests for a difference
- ) Interval data
- ) Repeated measures design
What do you need to know?
•You will be told the observed value of T.
•The test then uses a table (table of critical values) to calculate the critical value.
•We also need to know how many participants there are in total and calculate degrees of freedom when looking at the critical values table. (N- 1= df)
•If the final observed value of T is greater than the critical value the test is significant and we can accept the alternative hypothesis.
Mann-Whitney U
1) test for a difference
2) ordinal data
3) independent groups design
What do you need to know?
•You will be told the observed value of U
•The test then uses a table (table of critical values) to calculate the critical value.
•We also need to know how many participants there are in each group
•If the final observed value of U is less than the critical value the test is significant and we can accept the alternative hypothesis
Wilcoxon t test
- ) Tests for a difference
- ) Ordinal data
- ) Repeated measures design
What do you need to know?
•You will be told the observed value of T.
•The test then uses a table (table of critical values) to calculate the critical value
•We also need to know how many participants there are in total
•If the final observed value of T is less than the critical value the test is significant and we can accept the alternative hypothesis.
Psychological report
Abstract: summaries the aims, hypothesis, method, findings and conclusion
Introduction: large section of writing that outlines the background research and explains why this study is being conducted and how it relates to the background research.
Aims and hypothesis
Method: explains how the experiment was carried out: design, participants, materials, procedure
Results: descriptive statistics, inferential statistics, discussion
References
