120 likes | 216 Views
Research Design (chpt 3). A design is the practical plan of how to investigate a study Contains the nitty-gritty of how to conduct the research What subjects are going to do what Which measurements/scales are going to be used. Design Terminology. Variable: a thing which is measured
E N D
Research Design (chpt 3) • A design is the practical plan of how to investigate a study • Contains the nitty-gritty of how to conduct the research • What subjects are going to do what • Which measurements/scales are going to be used
Design Terminology • Variable: a thing which is measured • Independent variable (IV): a variable which is manipulated by the researcher • Dependent variable (DV): a variable which changes as a result of the research process • Eg: in a study which loks at how nutrition affects intelligence • Independent variable: Nutrition • Dependent variable: Intelligence
Design Terminology 2 • Eg: In a study which looks at how intelligence affects matric scores • Independent variable: intelligence • Dependent variable: matric scores • Intervention: the “thing” done to the subjects to check for changes in the DV • Eg. In a study to see how nutrition affects IQ, the feeding program is the intervention
The design overall Think of a design as a machine: IVs go in, and DVs come out. We are interested in how the IVs affect and change the DVs
Importance of the design • The design is the scaffold on which the entire study hangs • A faulty design will lead to false conclusions • The design must be carefully evaluated before performing the study • Once the data has been collected, the design cannot be changed (too late!)
Design notation • Designs are written with a special notation: • An O represents an observation (measurement) • An X represents an intervention • An arrow represents the passage of time Observe, intervene, Oserve again Intervene, then observe Each Line represents one group of subjects
Design notation (2) • Designs can be complex, using several groups: The Solomon 4 Group design (1 single study with 4 groups) (very safe design, but requires many subjects)
Design Validity • A study has validity when it has the capacity to study what it aims to study • A study that claims to study intelligence must have some measure of intelligence in it • A study with poor validity is powerless • Campbell (1979) came up with a list of threats to validity, eg • Subject dropping out • Bad measures of variables
Design Validity (2) • Validity of designs is improved by eliminating rival hypotheses • Show that the only likely explanation is yours • Identify as many possible rival hypotheses as you can, and change the design to reduce their impact • Eg. If you suspect age might be a rival hypothesis, record subject’s ages so you can later analyze age and show it was not a factor
Some design descisions • Unit of analysis: what are you talking about? Groups? Individuals? Idiologies? • Time: is the study longitudinal (follow people over a long time) or cross-sectional (a snapshot in time)? • These descisions affect the conclusions that can be drawn! Must be carefully chosen
Type of study • Another descision: descriptive, relational or experimental? • Descriptive: paints a picture of how things are right now. No hypotheses are tested • Relational: investigates the relationship between two or more variables as they occur naturally • Experimental: investigates how one or more variables cause another variable
How to stuff up your own work • Research is a specific social situation – you can “cause” subjects to behave in a particular way • Experimenter effect: Your race, gender etc. suggests to people how they “should” behave • Demand Characteristics: The research setting or measures can “suggest” to the subjects how to behave • You can eliminate this by careful control or observation