experimental methods, between/within-group designs, internal validity

method of agreement

shows a certain factor is necesarry for bringing about a certain affect (looks for factors that are present when an effect occurs)

method of difference

looks for factors that are present when an effect occurs, but absent when it does not

method of concomitant variation

quantitative change in the effect is associated with quantitative changes in a given factor

indepedent variable

manipulated by the experimenter

experimenter (type e) variable

also known as the active variable; similar to the independent variable; can be manipulated by the experimenter

example of type e variable

the type of music played in Ruth's study

dependent variable

measured by the researcher; any changes that occur are dependent on the independent variable

subject (type s) variable

also known as the attribute variable, cannot be manipulated by the experimenter

example of type s variable

age, gender, race

true or false: the independent variable can only ever be a type e variable

false, the independent variable can be a type e or a type s, depending on the study

moderating variable

affects the relationship between the independent and dependent variable

mediating variable

also known as the intervening variable, affects a dependent variable when there is no direct relationship to the independent

extraneous variable

also known as the confounding variable, affects the dependent variable, but is not considered by the experimenter

discrete variable

a quantitative variable that has either a finite number of possible values or a countable number of possible values

example of a discrete variable

grades, number of students in a class

continuous variable

a quantitative variable that occurs within a range and only takes certain values

example of continuous variable

reaction time, test score

method of loci

use of familiar locations as cues to recall items that have been associated with them

what is the 'maxi' portion of the maximincon principle?

the part of the principle that allows systematic variance to be maximized (true score)

what is the 'min' portion of the maximincon principle?

the part of the principle that allows random error variance to minimized

what is the 'con' portion of the maximincon principle?

the part of the princople that allows extraneous error variance to be controlled.

type 1 error

rejecting the null when it should be retained

type 2 error

retaining the null when it should be rejected

between-group designs

participants are exposed to only one level of the independent variable, but the dependent variables may be multiple (participants may be tested several times)

random groups design

a type of between-groups design that utilizes random selection, random assignment, block randomization

matched groups design

a type of between-groups design that utilizes frequency distribution control, precision control, and yoked control

concerns with between groups designs

demand characteristics, social desirability, and external validity

within-groups design

participants are exposed to all the levels of the independent variable and may be tested once at each level or several times at each level

reasons for using within-subjects designs

small sampling frame, matching required on several variables, answers research questions

problems with within-subjects designs

order effects, carryover effects

order effects

the order in which the effects are introduced will impact the dependent variable

carryover effects

knowledge gained from the first effect impacts the second impact

assymetrical carryover effects

whatever is learned at the first effect is either detrimental or enhancing, depending on what is learned first

symmetrical carryover effects

what is learned on the first level increases performance on second level

linear practice effects

has a ceiling effect; you become successful because you have learned how to solve the problem

nonlinear practice effects

insight; no learning involved - you look at a problem and automatically know the solution

range effects

the performance of the dependent variable is effected by the rage and most ranges have the most responses at the middle

how to control order effects

counterbalancing (including all levels at each ordinal position equally often)

when is a within-subjects design required?

when participants past experiences may influence results

block randomization

a process of randomization that first creates treatment blocks containing one random order of the conditions in the experiment; subjects are then assigned to fill each successive treatment block

limitations to within-subjects designs

differential transfer, cannot study subject-variance effect

threats to internal validity

history, maturation, testing, instrumentation, selection, statistical regression, experimental mortality or attrition

history

past experiences that pose a threat to internal validity (can affect all designs)

maturation

changes over time that pose a threat to internal validity (can affect the study if participants are tested over time)

testing

a threat to internal validity when the knowledge of the test in a test-retest experiment impacts the dependent variable (effects extreme scores evaluated at only two points)

instrumentation

a threat to internal validity when there are inconsistencies in data collection - measuring tool is changed in the middle of the study (can affect all designs)

selection

a threat to internal validity that occurs when there is a non-representative sample (occurs when random selection and assignment are not used - affects all quasi-experimental designs)

statistical regression

occurs when scores begin to regress in the direction of the mean (can affect studies that focus on extreme scores that are evaluated at only two points in time)

experimental mortality/attrition

a threat to validity that occurs when participants drop out or lose interest in a study (can affect all designs)

selection interaction

a threat to internal validity that occurs when any threats impact just one group of the study (effects designs with a non-equivalent control group)