- Created by: Peter Jones
- Created on: 28-06-10 11:45
The Issue of Reliability
Reliability: Its the extent to which the way in which you measure a vairable or concept is consistent so repeating measures would produce similar results.
In the context of an Experiment, reliability refers to the ability to repeat a study and replicate it and produce similar results with conditions all being the same.
In an Observation, two observers should produce the same results and the extent to which the observers agree is called Inter-Observer reliability. To improve inter-observer reliablity, observers should be trained carefully in the coding system so that they are both looking for the exact same thing.
Internal Reliability: Its the level at which something is consistent within itself for example questions in an IQ test should all be measuring the same thing.
External Reliability: Its the measure of consistency over several occasions so say an experiment was repeated several times over a few months, if they all produced similar results then it has high external reliability.
The Issue of Validity
Internal Validity: How much the study managed to test and measure what they set out to measure.
External Validity: The extent to which the results of a study can be generalised to the rest of the world. Its also known as Ecological Validity.
Things that effect Experimental Research: Extraneous Variables, Demand Characterists, Research Design and Mundane realism.
Things that effect Observational Research: The coding system, Observer Bias and the fact some behaviours are not observable.
Things that effect Self-Report Techniques: Social Desirability, Biased Interviewer, Demand Characteristics, Are the questions measuring things correctly and does the individual fully understand the questions.
When psychologists get a sample they want to minimise costs and maximise generalisability as lots of participants cost a lot of money.
Opportunity Sample: Participants are selected by those who are most easily available but can tend to be biased. E.g if you try and get people on a monday morning in a town centre you wont get professional workers as they are all working.
Volunteer Sample: Participants are selected by asking for volunteers, however its also biased as the sort of people who volunteer are a certain type and may be there for different reasons, money, spare time etc..
Random Sample: This is unbiased as participants are selected by all members of a population are calculated then random ones selected giving an equal chance
Snowball Sampling: In some studies where it is difficult to find participants with a certain characteristics you get one or two participants to direct you towards others. E.g someone with an eating disorder.