- Created by: nagina
- Created on: 18-05-14 21:05
Empirical Methods - Information you have collected yourself through direct observation/experiment. Have to collect facts. (Opposite of meta analysis.)
Objectivity - Data has to be objective. can't have experimenter bias. has to be unbias throughout research. Lab studies unbias, controlled setting.
Replicability - way to demonstarte validity. have to be able to repeat it. if outcome is the same study will have high replicability. scientists should record procedure so someone else can repeat study and verify original results. (Case studies, natural observation can not be repeated)
Theory construction - cant just have facts. explanations/theories must be constructed to make sense of facts. Theory is important so we can make predictions with people behaviour. (could be limited) Parasychology has no theory, all facts.
Hypothesis testing - good theory must have good hypothesis. must have testable expectations. if scientist fail to find support for their hypothesis must modify theory. If no support is found then must take null hypothesis; Falsification.
Falsification - This is when the researcher has proved their theory wrong, therefore have to falsify work. reject null hypothesis and accept alternative hypothesis.
Nomethetic - When a theory is applied to everyone. (science)
Idiographic - When you take a theory and apply it to an individual. (Psychology)
Induction - When you think of your hypothesis, gather research and propose a theory.
Deduction - when you propose a theory, find research to support you theory.
Psychologists which come from the same family (e.g behavioural, biological) review someones work. Check for errors, make sure its of a high quality. peer reviewers are unpaid. (links with inter-rater reliability.) increases validity, ensures no bias.
peer review serves 3 main puposes:
1.) Allocation of research funding - research paid by various govenment/charities. in 2008 had £605 million to spend, do reviews to enable them to decide which research is worthwile.
2.) Publication of research in scientific journals/books - scientific journals provide scientists with oppurtunity to share results of their reserach, therefore mjst follow scientific principles.
3.) Assessing the research rating of university departments - All university science department expected to conduct research, assessed in terms of quality. future funding for department depends on recieving good rating from RAE peer review.
Now people using the internet to peer reveiw. can be a bad thing as anyone is able to peer review it without having any psychological background. can be expensive, subjective, picking up fraud. (smith, 1999)
Finding an expert - not always possible to find an expert to review research. poor research may be passed as reviewer may not understand it. (decreases validity)
Anonymity - Peer reviewer is anonymus, so reviewer can be honest and objective. now have open reviewing.
Publication bias - journals seen as bad as newspapers for seeking eye- catching stories, want sensationalist stories, so its interesting.
Preserving status quo - want all research to go with existing theories, resistant to large shifts in opinion
Cannot deal with already published research - once research study has been published results remain in public view even it has been shown to be fradulent or have poor research practices. (Cyril Burt)
Independent variable - thing you are changing/manipulating (e.g. sleep, food) aim to control IV, have limited extrenuous variables.
Dependent variable - Thing that has been affected by IV (e.g. performance on math test)
Lab experiment - (Follow scientific methods)
+ Conducted in control environment, good control on DV
+ Limited extrenuous variables - internal valdity will be high
- not very realistic - lacks mundane realism
- cant be generalised in real life - lacks external validity
- Demand characteristics
Conducted in a more natural surronding
+ High in external validity, high in mundane realism
- demand characteristics
- Lacks internal validity, not everything is controlled, has some extrenuous variables.
Can be called observation, not manipulated or controlling variables.
+ No demand characteristics, get rich data, high in mundane realsm
- not ethical,no informed consent, privacy issue, right to withdraw - participants dont know they are being studied.
- Externuous variables - not controlling anything
Independent measures - Split group in 2: 1st group is treatment group, 2nd group controlled group. Both groups do both both levels of experiment.
+ No order effects, quicker
- individual differences
Repeated measures - One group take part in both conditions
+ fewer people used,no individual differences
- Takes longer
Matched pairs - Matched with someone with same characteristics e.g. age, school etc.
- individual differences
Level of measurement
Nominal data - Data is in sepeate catergories e.g. height - tall,medium, short
Ordinal data - data that is ordered in some way e.g. first, second, third
Interval - measured by units e.g. measuring heights, 5ft 1", 5ft 2" etc
Mann - whitney = U
use mann whitney when:
- independent measures is used
- The data is ordinal or interval
Will coxen = T
use will coxen when:
- Repeated measure is used
- when data is ordinal or interval
Spearmens rho = RHO
Use spearmens Rho when:
- When the study is a correlation
- When the data is ordinal or interval
Chi - square = X2
use a chi - square when:
- when study is a correlation
- when their is nominal data
Type 1 error (5%) - when you think you have found an effect but havent. Reject null hypothesis and accept alternative. use 10% but then correct it by using 5%.
Type 2 error (1%) - when you think you havent found an effect but have. Accept null hypothesis and reject alternative.
Psychologists use a 5% signifincance level because there is a 5% probability that the results are due to chance. If significance level is below 5%, null hypothesis is rejected and alternative is accepted. if significance level is above 5%, null hypothesis accepted and alternative is rejected. however there is a 95% chance that the results are due to an effect of the experiment. Use 1% sig level when you want it to be more stringent, mainly used in doctors research.
Anything negative is not significant.
One tailed - directional
Two - tailed - Non - directional
Title - tell the reader what the report is about
Abstract - provide the reader with brief summary of study
introduction - introduce the background of study, should lead to aim, hypothesis of study.
Method - describe how study was done. include design, apparatus, procedure
Results - summarise findings. include tables/graph/statistics
Discussion - dicuss findings and implications, should hypothesis be accepted? disscuss alternative explanations and methodological problems
references - inform reader about sources of information
Appendices - can be used for detailed information.