Chapter 9 - Within-subjects design

define the within-subject design

single group of participants in all conditions

what are the two principle accomplishments of the within-subject design

1. equating groups by using same subjects

2. reducing within-group variance by controlling for individual differences

describe the calculation of the f-ratio in the within-subjects design

between condition variance (condition effects + error) / within condition variance (error)

what are the main sources of confounding in within-subjects designs (threats to internal validity)

1. confounding from environmental variables

2. confounding from time-related variables

a (short/long) timespan between two measurements will show a greater risk of impacting results

long

define carryover effects

exposure to one manipulation may produce persistent consequences influencing future measurements

define progressive error

changes to behavior that are generally related to experience in research but not due to treatment

define practice effects

type of progressive error, progressive improvement through treatment condition

define fatigue effects

progressive decline in performance through treatment conditions

explain how counterbalancing can control for or reduce order effects in within subjects design

disrupt the systematic relationship between treatment order and any order effects and prevents these order effects from, accumulating in a particular condition by spreading them out

for counterbalancing to work, there must be (odd, fair, equal) numbers of participants in each counterbalanced order

equal

define Latin square counterbalancing

delivery of treatment such that each condition appears in every position once ex:

a b c

b c a

c a b

using a latin square, each condition is equally preceded and succeeded by (each, highest and lowest, extreme, important) conditions

each

explain the danger of asymmetries in counterbalancing orders

if carryover effects are asymmetrical, counterbalancing orders will not yield the same results (i.e., performance on b better after a but not before)

why are within-subject designs not used with experimental conditions producing lasting effects which cannot be reversed

carryover will affect subsequent measurement