Download presentation
Presentation is loading. Please wait.
Published byOswald Fleming Modified over 9 years ago
1
Research Methods revision
2
Peer review What do peer reviews achieve? Allocation of funding: public bodies can peer review studies to gather whether or not they will be worthwhile Publication of research in scientific journals and books: this gives scientists the opportunity to share the results of their research. Peer review process can be used to eliminate incorrect or faulty date entering the public domain. Assessing the research rating of university departments: all university departments are expected to conduct research and this is assessed in terms of quality future funding. Good research = better funding.
3
What problems occur with peer reviews? Unachievable ideal: is not always possible to find an expert to review a research proposal. Poor research could therefore be passed on. Anonymity: gives people the chance to ‘settle old scores’ or bury rival research. Publication bias: researchers may publish something if it will prove their own research. Also, positive results tend to be published more often. Preserving the status quo: peer review has a preference for data that already follows what other current theories have found rather than dissenting. Peer review therefore can make the pace of development slower.
4
Specific research methods used: 1.Experimental 2.Observations 3.Correlational analysis 4.Surveys 5.Interviews 6.Case studies 7.Pilot studies
5
Lab experimentADVANTAGESDISADVANTAGES within an artificial environment where the researcher has full control over the variables Specialised equipment Cause and effect Easy to replicate Low ecological validity Demand characteristics Field experimentADVANTAGESDISADVANTAGES is within a more natural environment but researcher has some control. Higher ecological validity Lower demand characteristics Control not always maintained Equipment not practical Natural experimentADVANTAGESDISADVANTAGES real life environment where experimenter has no control. No/few demand characteristics Low control Cannot establish cause and effect Time consuming 1. EXPERIMENTAL
6
CONTROLLEDADVANTAGESDISADVANTAGES Is where the conditions are contrived by the researcher. This type of observation may be carried out in a laboratory type situation. Easy to replicate Easy to analyse Low ecological validity Researcher bias Demand charac. NATURALISTICADVANTAGESDISADVANTAGES Is when observation is within the participants natural environment. Higher ecological validity Qualitative data Low validity Too specific PARTICIPANTADVANTAGESDISADVANTAGES Observer is involved with the participants he/she observes. Qualitative date Very in-depth First hand experience. Costly Hard to arrange – time consuming Low validity 2. OBSERVATIONAL NON-PARTICIPANTADVANTAGESDISADVANTAGES Observer is NOT involved with the participants he/she observes. Not relying on memory Ethical issues
7
This is the use of statistical correlation to evaluate the strength of the relations between variables. 3. CORRELATIONAL ANALYSIS ADVANTAGESDISADVANTAGES Shows there is a connection between two variables. This evidence can strengthen research and theories Shows a relationship Causality cannot be established as there is always going to be other variables (extraneous variables) involved. Linear graphs do not accurately represent variables.
8
This is a set of questions that can be written out and delivered to participants in order for them to fill out. 4. SURVEYS ADVANTAGESDISADVANTAGES Can approach sensitive topics Sent them out to many people cheaply and quickly i.e. Post or email. Lots of data – both qualitative and quantitative. If badly worded, can effect participants responses. (i.e. Ambiguous or leading) Very low response rate (33%) – especially if via post. Demand characteristics
9
UNSTRUCTUREDADVANTAGESDISADVANTAGES The interviewer has an idea of what they will ask they interviewee but has no questions pre-prepared. Better validity Speak openly Detailed info Flexible Difficult to analyse Training for interviewer needed Harder STRUCTUREDADVANTAGESDISADVANTAGES The interviewer has a pre-set list of questions that they read from a sheet as they ask the interviewee. Less likely to deviate Data analysis easier Easy to generalise Less training needed Lack of validity – formality of sit. No further questions Demand charac. And social desirability. 5. INTERVIEWS SEMI-STRUCTUREDADVANTAGESDISADVANTAGES The interviewer has some questions ready but may ask more questions depending on interviewee’s answers already given Qualitative data Keep focus but still gather more data. Training needed Time, money and effort Social desirability
10
Focusing on one individual or a group of people and using different research methods to attain data. 6. CASE STUDIES ADVANTAGESDISADVANTAGES Lots of data attained – both qualitative and quantitative which will be very in-depth Can use case studies to challenge existing theories Researcher bias Retrospective data that is not always reliable Very narrow study if focus is on one individual Cannot generalise results to everyone
11
A practice attempt before the real study to find if there is any problems with the upcoming experiment. 7. PILOT STUDY ADVANTAGESDISADVANTAGES Gives the researcher useful knowledge of how the task with be carried out. Eliminates any faults which will save money as they will not occur during the real thing Participants know it is a pilot study and may act differently to make it easier for the environment Participants may sabotage the research by giving the experiment false faults
12
Sampling... Opportunity sampling Whoever is around at the time is chosen to take part in a study – based on availability. ADVANTAGESDISADVANTAGES Quick and therefore the most common type used Convenient and easy Not representative – therefore any results can not be generalised to the public Volunteer sample Participants see an advert for a study and sign up. ADVANTAGESDISADVANTAGES Consent given instantly Motivated particpants social desirability Not representative – therefore any results can not be generalised to the public
13
Sampling... Random sample When everyone is given an equal chance (i.e. Picked out from a hat. ADVANTAGESDISADVANTAGES Convenient and easy More representative then a volunteer or opportunity sample Does not guarantee to be representative Time consuming. Stratified sample Ensuring the ratio of people is equal. 50:50 ADVANTAGESDISADVANTAGES Most representative Can generalise from this data to the rest of the population Sorting out participants for the study can be time consuming
14
Experimental designs: Design descriptionADVANTAGESDISADVANTAGES Independent groups: different people in different conditions No order effects Reduction in demand characteristics Different people – hard to make a comparison Lots of different people – expensive and individual dif. Repeated measures: same people doing all the conditions Fewer participants Comparing them with themselves. No individual differences Order effects – individuals becoming tired or bored INTRODUCE COUNTERBALANCING Matched pairs design: similar people doing same condition Increase validity. Individual differences decreased Lower order effects People never similar Time consuming matching people
15
RELIABILITYRELIABILITY INTERNAL RELIABILITY Concerns the extent to which there is consistency. Different parts should give consistent results throughout. i.e. IQ tests should all have questions of similar difficultly Split-half method: split the test in half by odd and even numbers – this can assess questionnaires. EXTERNAL RELIABILITY Concerns the extent to which there is a measure of something that is consistent over time. Test should produce consistent results regardless of when used. This can be assessed with the test-retest method: repeating the test at a later date.
16
How do you improve reliability?
17
VALIDITYVALIDITY INTERNAL VALIDITY Concerns the extent to which the changing IV is entirely reliable for the DV and not an extraneous variable. Milgram’s study was valid as the participants believed the study was real. Face validity: examines/assesses the test EXTERNAL VALIDITY The extent to which things can be generalised to other people, times, situations etc. Milgram’s study did not have it as it was set in a lab, not in a natural environment. Predictive validity – two sets of scores are obtained at different times. Allows accurate prediction of future behaviour.
18
How do you improve validity?
19
GRAPHS ! BAR CHART FREQUENCY POLYGON HISTOGRAM LINE GRAPH SCATTER GRAPH
20
Significance levels P < 0.01 Stricter level of significance. It means we are 99% sure the results are accurate and there is a real effect. Used more for drug studies as the results need to be accurate in order to be safe. This type of significance will lower the chances of a type 1 error (FAR) where you falsely accept the research hypothesis. P < 0.05 A conventional level of significance. This means you are 95% sure that it is a real effect. This level would be used more for studies that have smaller samples and are not too important. This type of significance will also lower the chances of a type 2 error (FAN) falsely accepting the null hypothesis.
21
TYPES OF DATA Nominal Where simple categories are used – i.e. Categorising people as smokers or non-smokers. A numerical value often cannot be assigned. You can normally count up frequencies within these categories (i.e. Number of women or number of men.) Ordinal Where numerical value is used, but based on ranks or ratings. i.e. Doctors may be ranked in preference by patients. Ordinal data is very subjective as based on personal opinions of people. Interval data This is when data is in the form of equal units e.g. Score on an objective test. In such a case, if one participant has scored 20 on one test, and another has scored 10, then it is correct in saying participant one performed twice as well as participant two. Ratio Interval data with an absolute value of 0 i.e. Distance or speed
22
Choice of statistical test What level of measurement is the data? NOMINAL Chi-square Is the study about difference between two sets of data or a relationship between two variables? RELATIONSHIPS (CORRELATIONAL) Spearman’s Rho What type of design was it? INDEPENDENT GROUP Mann Whitney REPEATED MEASURE Wilcoxen
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.