Lecture 2 Flashcards
Explain
Finding out what caused behavior
Control
Changing causes
Causation
Experiment - one factor directly affects another factor
To Show Causation We Must Demonstrate That
- Changing the first thing produces a change in the second
- There is no other possible cause for the change in the second thing
Important – rule out alternative explanations or hypotheses
Components Of An Experiment
Population And Sample Dependent Variable (DV) Operational Definition Reliability And Validity Bias
Population
Members of a specific group
Sample
Relatively small subset of a population that is selected to represent the population
Representative Sample
Characteristics and behavior of the sample reflect those of the population (ensures generalizability)
Representative Sample Achieved By
Random sampling
- Selected members in an unbiased manner - all members have an equal chance of selection
Descriptive Statistics
Summarize the data collected from the sample
Inferential Statistics
Generalize from the sample to the population
Dependent Variable (DV)
The measure taken
What you record (depends on what the participant does)
Operational Definition
Specification of how property of interest will be measured
Validity
A dependent variable is valid if it measures what it is supposed to
Threat to validity arises from any unintended component that is reflected in a score
Therefore a poor operational definition can result in an invalid dependent variable
Reliability
A dependent variable is reliable if, under the same conditions, it gives the same measure, and contains a minimum of measurement error
Unreliable data reflect error and provide a biased perspective
If a measure lacks reliability it also lacks validity
Bias
A biased dependent variable is consistently inaccurate in one direction (i.e. always high, or always low)
A Good Dependent Variable Is
Valid
Reliable
Unbiased
Ceiling Effect
When a task is so easy that all scores are very high
Floor Effect
When a task is so difficult that all scores are very low
Floor And Ceiling Effects
Both mask differences between groups
Data Types And Scales Of Measurement
Every dependent variable has a data type
The data type determines what sorts of analyses you can perform on your data, and the conclusions you can draw
Data types are associated with different scales of measurement
The type of scale of measurement dictates the type of conclusion that you can draw
Nominal Scale
- Categorizes without ordering
- Numbers that substitute
- E.g. gender 1 – Female, 2 – Male
Ordinary Scale
- Categorizes and orders the categories
- Bigger means more
- Distance between points on the scale is not considered equal
- E.g. rugby team standings
Interval Scale
- Categorizes, orders, and establishes an equal unit of measurement in the scale
- Knowing how much more
- Distance between points on the scale considered equal
- E.g. Celsius temperature
Ratio Scale
- Categorizes, orders, establishes an equal unit in the scale, and contains a true zero point
- Allows ration statements e.g. twice as big
- E.g. number of items recalled in a memory task
Examples
• John is twice as lazy as Peter - Ratio • John is lazy but Peter is not - Nominal • John is lazier than Peter - Ordinal