psy 203 exam 3

Why would we want to run a factorial design instead of two single-factor designs when we have two independent variables?

How many factors are there in a 6X8 factorial design?

A researcher is running a 5x5 between-subjects factorial design. How many conditions will each participant experience in this study?

one condition

Calvin wants to know how people with depressive disorders will respond to monoamine oxidase inhibitors (MAOIs), but Calvin thinks that the subject variable relationship status might affect the results of the experiment. He matches everyone on relationship status and randomly assigns one person from each pair to the Treatment group, and the other to the Control group. What type of experimental design is Calvin using?

matched group design

A researcher wants to determine how socioeconomic status (SES) influences delay of gratification. Participants are matched on all variables but their SES and sorted into three groups: low SES, middle SES, and high SES. Then they're exposed to the delay of gratification procedure. What type of research design is this?

ex post facto

People with more referrals are more likely to be hired. What type of correlation is this?

positive

Two-IV (Two-Factor) Experiments

two IVs are manipulated

What do factorial designs assess?

overall (main) effect of each IV and interaction between IVs, if there is one

Y x Z Factorial

how many numbers=how many IVs

Within-subjects 2x2 Factorial Design

each subject is exposed to every level of each IV

Factorial Designs

a design with 2 or more IVs in which each level of each IV is combined with every level of the other IV

Types of factorial designs

between-subjects, within-subjects, and mixed

Between-Subjects 2x2 Factorial Design

each subject is exposed to only one condition of each level

Mixed Factorial Design

at least one between-subjects factor and at least one within-subjects factor

Within-Subjects Factorial Design

every subject is exposed to all levels of the IV

determining number of conditions in a factorial design

multiply the number of levels for each IV together

use a factorial design when...

you want to determine if there is an interaction between IVs

between-subjects factorial design

each subject is exposed to only one condition of each level

limitation of a single factor design

can't determine if there is an interaction between two or more IVs

main effect

differences between levels of an IV averaged over all levels of every other IV

interaction

the effect of one IV changes as a function fo the level of another IV

calculating main effects

does Factor A (or B) have an overall effect on DV?

determining if there is an interaction

if the answers are different to the two interaction questions, then there is an interaction; if not/if the answers are the same then there is no interaction

Interaction

EFFECT of one IV depends upon the level of another IV

Questions for Interaction

For Factor A (or B): What is the effect of Factor A on Factor B1?

What is the effect of Factor A on Factor B2?

Rule of Thumb for Interactions

only for 2x2 factorial designs: if the lines on the graph are not parallel, then there is an interaction

cross-over interaction

the two (or more) functions (lines) cross over at least one other line; indicates an interaction

repeated measures design (mixed factorial design)

separate groups of subjects tested (measured repeatedly); repeated measure=within-subjects factor; counterbalancing not possible

relational types of research (non-experimental)

contingency analysis and correlational research

correlational research

what is the nature and degree of relatedness between two variables? do variables co-vary (correlate)? if so to what degree?

when to interpret interactions relative to main effects

interpret interactions first then main effects

purpose of relational research

to determine in a non-experimental way if 2 or more variables are related to one another; does one variable change as a function of another variable?

contingency analysis

does binary value/outcome of one variable depend upon the binary value of another variable?

types of correlations

positive and negative

negative correlation

values of two variables change together in opposite direction

zero correlation

there is no linear relationship between the two variables; the variables are uncorrelated

degree of relatedness for correlation coefficient (r)

indicated by the value (0, .3, 1.0)

positive correlation

values of two variables change together in the same direction

correlation coefficient (r)

Pearson's r; indicates "direction" and degree of linear relatedness between two variables; range -1.0 to +1.0

direction of relatedness for correlation coefficient (r)

indicated by the sign (+/-)

strong correlation

r=.70 to 1.0

(positive or negative)

moderate correlation

r=.4 to .6

correlational vs. causal

causation cannot be determined from correlational data because no variables were manipulated

directionality problem

probably a causal effect between two variables, but the effect is in the opposite direction than what yo think; could be A-->B or B-->A

ordinate

y-axis

weak correlation

r=.1 to .3

fallacy of affirming the consequent

logical fallacy when you

abscissa

x-axis

interpretation problems with correlations/relational research

for positive or negative correlations: directionality or third variable

third variable

there are two correlated variables but no direct causal relation between them; they are both related to another (third) variable

nonlinearity

the variables are related to one another, but DV not described by a straight line (y=mx +b) but by another shape (ex: hyperbola)

quasi-experiments

quasi=almost; at least one non-manipulated variable

nonequivalent control group design

non-random assignment 2 groups (experimental and control) with a pretest for both groups, treatment for experimental group, an posttest for both groups

spurious correlation

false correlation; third variable; either two variables are correlated with one another because related to a 3rd variable or two variables only seem related

truncated range

narrowed range; if one variable's values don't vary then you won't get a correlation between the two variables; must have a large enough range with variability

common types of quasi-experiments

nonequivalent control group design, interrupted time series design, factorial design with non-manipulated variable, and developmental designs

factorial design with non-manipulated variable (PxE Designs)

p=person/participant

stable baseline performance

requirement to use small-n design; behavior does nor vary outside the range of some predetermined criterion

interrupted time series design

take repeated measures before and after treatment; treatment "interrupts" measures; O=outcome, T=treatment, nonrandom assignment

problems with interpreting interactions with factorial design with nonmanipulated variable

the thing you think might be deriving the effect might just be correlated with something else causing the effect

permissible to use a non-stable baseline when

treatment is expected to change behavior in the opposite direction

scatterplot

a graph depicting the relationship shown by a correlation

split-half reliability

a form of reliability in which half of the items (e.g., the even-numbered items) on a test are correlated with the remaining items

ATI Design (Aptitude by Treatment Interaction)

form of PxE factorial design; educational research; examines possible interactions between an aptitude (person) variable and a treatment (environment) variable

Archival Research

a method in which existing records are examined to test a hypothesis

Partial Correlations

a multivariate statistical procedure for evaluating the effects of third variables; if the correlation between X and Y remains high even after removing Z, then Z eliminated

test-retest reliability

a form of reliability in which a test is administered on two occasions and the correlation between them is calculated

interrupted time series design with switching replication

program is replicated at a different location at a different time

archival data

data initially collected for a purpose not related to a current research study and used later for a specific purpose in the current research

small N designs

within-subjects designs using small n of participants

baseline (A)

pretreatment phase in which treatment is not applied and performance measured

Reversal ABA Design

small-n design in which subjects experience baseline, treatment, and return to baseline conditions; treatment withdrawal design

Multiple Baseline Across Subjects (AB) Design

all subjects get baseline condition, then treatment condition started with first subject while the rest continue with baseline until all exposed to treatment

When/Why Small N Designs are Used

unusual situation or phenomenon, devising/evaluating individual treatment, little subject-to-subject variability, or good control over extraneous variables

One-Shot Case Study

post-test or pre-post design; one participant is exposed to a baseline condition and treatment continues

Repeating-Treatment ABAB Design

small-n design in which subjects experience baseline, treatment, return to baseline, and return to treatment conditions

Interaction (A-B-A-B-BC-B) Design

small-n design in which multiple treatments are compared across adjacent phases/conditions

Multiple-Baseline Across Behaviors (AB) Design

small-n design; baseline, treatment started on one behavior while reset still in baseline, then treatment staggered across the other behaviors

Need Special Precautions with Small-n Designs if

evaluating irreversible treatment effects

General types of small-n designs

reversal/ABA/withdrawal, repeating treatments (ABAB), multiple-baseline, changing criterion, and interaction designs

concerns with using a reversal/withdrawal (ABA) Design

phase B's effects might be irreversible/permanent and /or it is undesirable (unethical) to return to baseline

alternating-treatments design

small-n design; compares within the same study two or more treatments on the same behavior (DV); rapidly presents the treatments multiple times

B

designations for treatment phase in small-n design notation

rational for using reversal/withdrawal (ABA) designs

if subjects behave differently in B than in the second A phase, then the change in behavior was most likely due to the treatment

design to use when it is undesirable to end on a baseline phase

repeating treatment or ABAB design

changing criterion ("shaping") design

small-n design; successively implement stricter criteria for reinforcement; criterion is new baseline for the progressives step up in criterion

Purpose of Baseline

serves as reference condition with which to judge treatment effects

A researcher is running a mixed factorial design to assess the effects of notification frequency and notification method on positive feelings toward a company. Every participant experiences both conditions for notification frequency and only one condition for notification method. What is the within-subjects factor?

Notification frequency

A researcher is running a factorial design to assess the interaction between political ideology and informant source on acceptance of community programs. What is the person variable in the factorial design?

Political ideology

What is one difference between a typical within-subjects design and a repeated measures design?

repeated measures design= no counterbalancing

As part of the suggestions to be a competent and ethical researcher, what should we do when we have written a new experiment that we are ready to run?

Test out the program before running our full set of participants

In which of the following situations could we use a contingency analysis to determine the relationship between variables?

What type of observational research is a person doing if they go to a park and record the number of families with and without children who use the park?

What type of subject reactivity countermeasure are we using if we count the number of sticky fingerprints on windows next to tables in a restaurant to determine the preferred seating areas?

What time sampling technique would we use to record the number of times that an employee slept on the clock if we expected a high frequency of this type of behavior?

Researchers repeatedly counted the number of people without handicap stickers, license plates, or placards who parked in handicapped spaces at two different stores for several weeks. Then, as a treatment, they posted signs on the doors of one store (but not the other) that cars illegally parked in handicap spaces without a proper permit will be towed. They continued to assess illegal parking in handicap spaces. What type of quasi-experimental design did they use?

interrupted time series design

What is a defining characteristic of quasi-experimental design

no random assignment of participants to groups

Researchers wanted to know how comprehensive or abstinence-only sex education affected the number of unplanned pregnancies in teens and young adults. They used a factorial design with a non-manipulated variable to determine this relationship. Which variable was the environmental or manipulated variable?

type of sex education