Send the link below via email or IMCopy
Present to your audienceStart remote presentation
- Invited audience members will follow you as you navigate and present
- People invited to a presentation do not need a Prezi account
- This link expires 10 minutes after you close the presentation
- A maximum of 30 users can follow your presentation
- Learn more about this feature in our knowledge base article
Research Methods - AS Psychology
Transcript of Research Methods - AS Psychology
How psychologists conduct their research...
Ethical issues are conflicts between the interests of the researcher, the participants rights and what is acceptable. Ethical issues include:
Informed Consent -
Permission from the person you are studying in order to study them.
False information, holding the truth.
Right to Withdraw -
To be able to leave an experiment at one's own freewill.
Protection from harm -
Protection from abuse, bullying, harm, humiliation etc.
Anonymity, info that people do not want the public to know.
Invading privacy, experiments in the home without the homeowner knowing.
Why do Psychologists conduct research?
QUANTITATIVE DATA - number based (Quantities such as time, amounts, ratings.
QUALITATIVE DATA - Not number based (verbal amounts)
The main method psychologist's use to generate data is the EXPERIMENTAL method, in particular the LABORATORY experiment.
Laboratory experiments are SIMULATIONS where VARIABLES can be controlled, changed (manipulated) and measured by the researcher.
Anything that can be measured and that can be changed (or vary) such as age, weight, time, test scores.
Variables are compared to see whether one will CAUSE or EFFECT another variable.
This is known as the 'Cause and Effect' relationship.
Features of the 'True Experiment'
There are three key features to an experiment (also known as the 'True Experiment');
The researcher manipulates (controls) an INDEPENDENT variable in order to determine if it changes the DEPENDENT variable.
Any other variables that may also change the DV are controlled, eliminated or held constant. These 'other' variables are known as the CONFOUNDING variables.
Participants are selected randomly.
Quasi - Experiment...
When an experiment is lacking in one of the three key features of the true experiment criteria.
The researcher will formulate a HYPOTHESIS - a statement that will predict that one variable (the IV) will affect another variable (the DV).
A HYPOTHESIS IS NOT A QUESTION.
There are two types of Hypothesis -
hypothesis predicts a
'Words that conflict (i.e. colour of word & actual word) will take longer to recall'
. On the other hand, a
hypothesis predicts a
relationship between the two variables
for example; '
Words that conflict (i.e. colour of word & actual word) will have an effect on the time taken to recall'.
Laboratory experiments provide the setting to achieve the highest level of control over variables.
Replicability - Lab experiments are carefully designed to be easy to replicate and achieve similar results.
Access to sophisticated equipment.
Lacks VALIDITY & ECOLOGICAL VALIDITY - by controlling the variables, the experiment becomes artificial and unnatural & the extent to which the results can be generalised.
Demand Characteristics - participants will alter their behaviour because they know they are being studied.
Right to Withdraw
Field experiments are investigations that are carried out in a natural setting (such as a school or hospital) in order to improve the REALISM of the research. As with a lab setting, the IV is still deliberately controlled to show its effect on the DV.
Improved ecological validity.
Reduction in demand characteristics.
Difficult to control independent and confounding variables.
Dependent variables may be difficult to measure.
Confidentiality & Anonymity
In contrast, a NATURAL experiment is one where the researcher makes use of the naturally occurring differences in the IV - there is no control of the IV by the researcher (Quasi-experiment). An example of a Natural experiment is the case study of Genie.
`Reduction in demand characteristics
Researcher has no effect on the research situation - retrospective
Confounding variables are more likely to effect the results therefore it is difficult to establish a cause & effect relationship between the IV & the DV
Research opportunity maybe an issue as specific behaviour may only occur frequently.
Protection from harm
Studies using correlational analysis
Correlation refers to the statistical technique that measures the relationship between two variables. The Correlation Co-efficient is the perfect correlation pattern which is either positive or negative.
Quantitative measure of relationships between variables (number based)
Useful in understanding complex relationships
Only measures the degree of a relationship and does not assume the cause of it
Not all relationships are linear (one directional)
Informed consent (Natural experiment)
Confidentiality - use of findings
Behaviour is observed and recorded with no manipulation of variables. Observations are usually conducted in a natural setting and study naturally occurring behaviour. However there have been observational studies carried out in laboratory settings.
High in validity
High in ecological validity
Control of confounding variables
Replication is difficult
Right to Withdraw
Participants provide information knowingly about specific things relating to themselves. There are two types of questionnaires; Closed questions & Open Ended questions.
Simplicity, speed & cost
Less researcher impact
There are three types of interview; Structured Interview, Unstructured Interview & Semi-Structured Interview...
- These produce quantitative data, planned questions, structure allows focus to be maintained throughout interview, training is required, however because of focus, interviewer cannot prompt interviewee.
Unstructured Interview -
Do not follow a rigid structure, focus isn't narrow, few questions are used to initiate the interview but then the interview expands based on the answers provided, qualitative data is more difficult to anyalyse than quantitative data, interviewee will feel that they can express themselves more in this interview due to the lack of structure.
- This is typically a mix between the two previous types of interview, some planning beforehand concerning the questions to be used, the interview still holds a lot of freedom over the direction it goes, interviewer encourages interviewee to expand on their answers.
Tackles sensitive issues
Data can be misinterpreted
Interpersonal variables - gender, age, ethnicity etc.
Protection from harm
Right to Withdraw
Case Studies are in depth studies that are conducted over a long period of time, of an individual or group and undertaken in a real-life context (Longitudinal). Alternatively a case study can be an exceptional or unique circumstance of something that cannot be ethically tested such as the case study of Genie.
Rich & interesting data
Challenges existing theory
Findings may be subjective
Repeated Measures Design:
Same participants in each condition
Independent Groups Design:
Two or more groups for each condition
Matched Pairs Design:
Involves independent groups, but participants are matched with others that share similar characteristics. In this case it could be age & gender.
Specifying a set of operations or behaviours that can be measured or manipulated.
For example: An hypothesis is an operationalisation of the experiment aim because the wording has been changed to a testable statement.
Key aspects of investigation design
The term 'operationalising' means to simplify a concept or idea into a simpler form so that it can be studied and measured. It narrows the research focus on a broad term, for example progressive behaviour can be operationalised into 2 categories; Verbal Behavior and Physical Behaviour.
Anything other than the IV that can affect the DV.
An uncontrolled variable that produces an unwanted effect on the DV, which affects the results.
EV's can occur through RANDOM ERRORS or CONSTANT ERRORS.
Random errors cannot be predicted such as state of mind if the participant, levels of motivation, noise, temperature, previous experiences preceding the experiment.
Constant errors effect the DV in a constant way and can effect the experiment in more ways than one. Participant differences and errors of measurement are some constant errors.
One way of reducing Random Errors is by allocating participants randomly to experiment conditions, with the assumption that random errors will balance out over the course of the experiment.
One way of reducing Constant Errors is by COUNTERBALANCING. This is a method used in repeated measures design to combat *order effects, practice, boredom and fatigue in an experiment. It ensures that each condition is likely to occur in a particular order. If there are only 2 variables then these will have an equal chance of being 1st & 2nd.
In a repeated measures design, an extraneous variable arising from the order in which conditions are presented, e.g. a practice effect or a fatigue effect.
Correlational analysis is a way of assessing reliability of results. Reliability of the observer can be improved by operationalising key variables.
The ways in which the test itself can be assessed for it's reliability can be done using the SPLIT-HALF METHOD or the TEST-RETEST METHOD.
The split in half method is used in question based research, where the test is split either by odd or even numbers, in half or random parts. Tests are conducted separately, and results are correlated. Results are deemed reliable if there is a positive correlation.
The test-retest method tests the stability of a test or questionnaire over time. This is done by presenting the same test to the same participants on two different occasions. It has to be done within a specific time period.
Assessing & Improving Validity
There are three types of validity; Internal, External & Test.
The extent to which it can be certain that the research findings are the results of the variables stated in the hypothesis, the variations are the result of manipulation of IV, and not EV's or CV's.
The extent to which the results can be generalised to the population as a whole. This can take two forms:
Population Validity - The extent that the results can be generalised to other groups of people.
Ecological Validity - The extent to which the results can be generalised outside of the research setting.
Face Validity - Experts assess whether a test is appropriate
Content Validity - Experts assess whether the test instruments are appropriate.
Concurrent Validity - Comparing the new test with one that validity has already been established.
Predictive Validity - The same as Concurrent but 2 sets of scores are obtained at different points in time.
Scales of Measurement
Bar charts are for representing NOMINAL or ORDINAL data from average scores from different samples. EACH BAR is SEPARATED as there is no order or relationship between categories. Usually to prevent bias, bars are ordered either alphabetically or in ascending/descending order.
Histograms are for representing INTERVAL or RATIO scales of measurement. The bars are touching to indicate a relationship between the values on the x-axis. (Horizontal)
Used as an alternative to the histogram, but is particularly useful when presenting data from 2 or more conditions. 2 or more lines can be drawn on the same graph to show comparison.
Scatter grams are used to show a correlation; a relationship between 2 sets of variables to determine how much they do or do not CO-VARY. Correlations can be positive or negative.
A way of presenting raw data to identify patterns or trends.
Measures of Central Tendency - Average values such as the mean, median & mode.
Measures of Dispersion - How much the scores vary.
Measures of Disperse - RANGE:
The difference between the highest and the lowest scores, +1 if the scores are whole numbers e.g.
7, 9, 11, 12, 15, 21, 27, 32, 42, 58, 76
If the data is represented in decimals the +0.1 needs to be added, 2 decimal places 0.01 and so on.
If the data is in half units the 0.5 needs to be added.
Inter quartile Range
2 3 7 8 10 11 16 18 21 22 26 28
First you calculate the MEDIAN.
You then identify 50% of the numbers that lay either side of the median value. (In this case 50% is 6 digits, so 3 above and 3 below the median score)
The upper & lower boundaries of the inter quartile range are identified and a mean taken.
Nominal Scale of Measurement
Involves distinguishing between different categories of a variable. Data is organised into categories that are mutually exclusive, the labels are merely names, no order of the groups.
Ordinal Scale of Measurement
Involves organising the measured characteristics into categories, these categories can be placed in a logical order, based on their meaning. Ordinal categories indicate the rank position in a group e.g.
Very confident, confident, not all confident
Strongly agree, agree, neither agree nor disagree, disagree, strongly disagree.
Interval Scale of Measurement
Involves measurements that can be ordered and the intervals on the scale are equal because they are based on some standard unit of measurement. The point used to indicate zero is arbitrary. An example of interval data is IQ scores; it is not possible to claim that someone with an IQ of 120 is twice as intelligent as someone with an IQ of 60.
Ratio Scale of Measurement
Ratio data are measurements on a scale that has equal intervals but also a genuine zero point e.g. height in cm or weight in kg. With a fixed zero point can make it possible to make ratio statements i.e. to claim that someone whose height is 1.9m is twice as tall as someone who is 95cm.
Repeated Measures Design - Advantages & Disadvantages...
There is no issue with order effects which occur when participants' performance is positively or negatively affected by taking part in two or more experimental conditions.
There is the potential for error resulting from individual differences between the groups of participants taking part in the different conditions.
Independent Groups Design Advantages & Disadvantages
Individual differences between participants are removed as a potential confounding variable. Also fewer participants are required, since data for all conditions are collected from the same group of participants.
The range of potential uses is smaller than for the independent groups design.
Order effects may result when participants take part in more than one experimental condition.
Matched Pairs Design Advantages & Disadvantages...
A matched pairs design combines the advantages of both independent groups and repeated measures design.
Achieving matched pairs of participants can be difficult and time consuming. It depends on the use of reliable valid procedures for pre-testing to identify the matched pairs. Complete matching of participants on al variables that might affect performance. This design is relatively uncommon, with their use restricted to specific situations where a matching process is highly desirable in order that experimental success can be achieved.