RESEARCH DESIGN Flashcards
define internal validity
the soundness of a design
how strongly we can assert the changes in out DV to the IV
define external validity
how generalisable the findings are
how representative of the world are the findings
problem associated with internal vs. external validity
ensuring internal validity = means the potentially more artificial the study becomes = less externally valid
4 steps to internal validity
sound operationalisation of DV
strong experimental design logic
sound operationalisation of IV
consideration and used of appropriate remedies to control for extraneous variables
manipulation of the IV:
define experimental manipulation
participants allocated to an IV condition by the experimenter
manipulation of the IV:
define individual differences manipulation
a characteristic of the participant determines the level of the IV which they are tested
not strictly experimental
what are the study designs
Repeated measures (within groups) - each participants tested at each level of the IV
Between groups: each participant tested at only one level of IV
Factorial: more than one IV
strengths of a factorial design
more than one IV allows for more precise hypotheses
control of extraneous variables by including as an IV
ability to determine the interactive effect of two or more independent variables
weaknesses of a factorial design
using more than two independent variables may be logistically cumbersome
higher-order interaction are difficult to interpret
what is separation and how is it achieved?
maximising the variation between groups
achieved by operationalisation of the IV
- need min. 2 IVs
- levels need to be distinctly separate
- additional intermediate levels can then be added to identify curvilinear impacts
what is compression and how is it achieved
minimising the variation within groups
achieved by controlling extraneous variables
what are the two forms of extraneous variables
noise creating
Confounding
what are noise creating variables
extraneous variable that have:
no relation to the IV
but can randomly impact the DV to create extra variation in DV
what are confounding variables?
extraneous variables that
- systematically impact the DV
- related to the IV
- potentially explain changes in DV that would be expected from the IV
how can confounds be controlled for
eliminate impact by keeping the variable constant
or building it into the study to measure its impact
what is the trouble with between group designs
two separate groups»_space; participant groups that can differ in any way both relevant and irrelevant to the study
need to make the groups as similar as possible
the issue of between groups design is addressed by…
selection - the process of assigning individuals to an experimental group
matching - equating participants in groups on specific variables
what are the different methods of selection
self-assignment - participant select group
experimenter-assignment - experimenter selects group
Arbitrary assignment - based on seemingly irrelevant criteria
random assignment
what are the pros of random assignment
ensures every member has an equal change of being assigned to either group
max insurance that groups are equal
eliminated systematic group differences
does not eliminate extraneous variables but randomly distributes them
what is matching
using different techniques to equate participants in the groups on specific variables
should be done with variables thought to be related to/ confound the IV
e.g. intelegence, age, gender
what are the two styles of matching
individual matching
distribution matching
pros and cons of individual matching
PRO
- groups equate on potential extraneous variables
CON
- hard to identify what variables to match on
- more variables to match on = harder to match participants
- decreases generalisability of results
what is blocking
building an EV into the design
- makes the EV another IV in the study
- should only be used when you are interested in the effect of the EV
whats the PROS with repeated measures designs
Eliminate problem of group differences - major error source removed
require fewer participants to have good statistical power
