WHY EXPERIMENTS?
* TO KNOW THE CONSEQUENCES OF A TREATMENT:
* 2 EQUIVALENT GROUPS OF PEOPLE
* EXPOSE ONLY ONE GROUP TO THE TX
* EXAMINE THE GROUPS TO SEE IF THEY DIFFER
* EXPERIMENTAL SIMUPLATION STUDIES, QUASI-EXPERIMENTS, AND TRUE
   EXPERIEMENTS.

EXPERIMENTAL SIMULATION STUDIES
* DONE WHEN IMPRACTICAL TO DO A TRUE EXPERIMENT - YOU APPROXIMATE ONE -
* "ACTING OUT" THE IMPORTANT ASPECTS OF THE TREATMENT CONDITION.
    EX. ZIMBARDO EXP.
* PARTICIPANTS KNOW FROM THE OUTSET THAT TYE ARE BEING ASKED TO PLAY A
   ROLE OR ACT OUT A PART
* PROBLEM: SOME PARTICIPANTS MAY BE SELF-CONSCIOUS - MAY ACT
   DIFFERENTLY - NOT NATURALLY
* PROBLEMS: INTERNAL VALIDITY?

QUASI-EXPERIMENTS
* MORE LIKELY TO USE THIS THAN SIMULATION STUDIES - INVOLVE NATURALLY
   OCCURRING MANIPULATIONS
* NATURAL EXPERIMENTS:
* RESEARCH DESIGN IN WHICH THE RESEARCHER HAS ONLY PARTIAL CONTROL
   OVER HIS OR HER INDEPENDENT VARIABLES.
* EXAMPLE: INSTRUCTOR TEACHING STYLES AND STUDENT PERFORMANCE - DEAL
   WITH THIS BY: ASSESSING THE WAYS IN WHICH THE GROUPS MAY HAVE BEEN
   DIFFERENT PRIOR TO THEIR EXPOSURE TO THE TREATMENTS OF INTEREST.
 
QUASI-EXPERIMENTS (CONT')
* PERSON BY SITUATION QUASI-EXPERIMENT:
* ALMOST ALWAYS TAKE PLACE IN THE LABORATORY
* RESEARCHERS USE RANDOM ASSIGNMENT TO DETERMINE THE TREATMENT
   CONDITIONS TO WHICH DIFFERENT PARTICIPANTS ARE EXPOSED.
* HOWEVER, RESEARCHERS MANIPULATE AT LEAST ONE INDEPENDENT VARIABLE
   AND MEASURE AT LEAST ONE ADDITIONAL INDEPENDENT VARIABLE.
* THE MEASURED VARIABLE IS USUALLY A STABLE INDIVIDUAL-DIFFERENCE
   VARIABLE- SUCH AS PARTICIPANTS’ ATTITUDES OR THEIR LEVEL OF SELF-ESTEEM.

TRUE EXPERIMENTS
* CRUCIAL INGREDIENT: RANDOM ASSIGNMENT OF PARTICIPANTS TO DIFFERENT
   TREATMENT CONDITIONS.
* SPECIFIC KINDS OF TRUE EXPERIMENTS DEPEND UPON:
* DOES THE EXPERIMENT CONTAIN ONE OR MORE INDEPENDENT VARIALBES
* WHETHER THE PARTICIPANTS IN AN EXPERIMENT SERVE IN ONLY ONE OR MORE
   THAN ONE EXPERIMENTAL CONDITION.

TRUE EXPERIMENTS (CONT')
* ONE-WAY DESIGNS:
* DESIGN IN WHICH THERE IS ONE AND ONLY ONE INDEPENDENT VARIABLE
* THE SIMPLEST KIND OF ONE-WAY DESIGN IS THE TWO-GROUPS DESIGN
* TWO GROUPS DIFFER IN A MEANINGFUL WAY IN TERMS OF A SINGLE VARIABLE.
* THIS CAN ALSO BE ONE GROUP RECEIVING A TREATMENT BUT DIFFERENT
   LEVELS, I.E., 50 MGS OF DRUG VS. 100 MGS. - NEEDS TO BE ONLY 2 LEVELS.
* EXAMPLES: MEMORY: CHRISTOPHER COLUMBUS DISCOVERING AMERICA;
   PERCEPTION AND STEREOTYPING - GLASSES MAKE YOU LOOK MORE INTELLIGENT.

TRUE EXPERIMENTS (CONT')
* ONE-WAY MULTIPLE-GROUPS DESIGN
* ONLY A SINGLE INDEPENDENT VARIABLE - HOWEVER IT TAKES MORE THAN TWO
   LEVELS
* EXAMPLE: ONE GROUP GETS A PLACEBO; FOUR OTHER GROUPS GET DIFFERENT
   DOSAGES OF THE DRUG.
* HOWEVER, IT TAKES MORE TIME, MORE RESOURCES, AND MORE RESEARCH
   PARTICIPANTS
* MAJOR LIMITATION OF ONE-WAY DESIGNS: THEY ONLY ALLOW RESEARCHERS TO
   LOOK AT ONE INDEPENDENT VARIABLE AT A TIME.

TRUE EXPERIMENTS (CONT')
* FACTORIAL DESIGNS:
* NEED TO LOOK AT THE EFFECTS OF MULTIPLE VARIABLES ACTING TOGETHER.
* FACTORIAL DESIGNS: YOU CREATE EVERY POSSIBLE COMBINATION OF ALL OF THE
   LEVELS OF YOUR INDEPENDENT VARIABLES.
* SIMPLEST DESIGN: 2 X2 DESIGN. EACH NUMBER IN THE DESIGN REFERS TO THE
   NUMBER OF LEVELS OF ONE OF THE INDEPENDENT VARIABLES.
* YOU CAN FIGURE OUT THE TYPE OF STATISTICAL ANALYSIS THAT SHOULD BE
   USED TO ANALYZE THE DATA FROM THE FACTORIAL EXPERIMENT. ONE-WAY -->
   ONE WAY ANOVA
* 2 X2 AND 2 X 3 ARE PRETTY COMMON DESIGNS.
* EXAMPLE: TAKING SOCIAL CONTEXT INTO CONSIDERATION WHEN JUDGING OTHER
   PEOPLE’S PERSONALITIES.

TRUE EXPERIMENTS (CONT')
* FACTORIAL DESIGNS - CONTINUED:
* EXPERIMENTS WITH FACTORIAL DESIGNS ALWAYS HAVE AT LEAST TWO
   INDEPENDENT VARIABLES
* FACTORIAL DESIGNS: ALLOWING RESEARCHERS TO ANSWER QUESTIONS ABOUT
   MORE THAN ONE INDEPENDENT VARIABLE AT THE SAME TIME.
* MAIN EFFECTS: STRAIGHTFORWARD (SIMPLE) EFFECTS OF AN INDEPENDENT
   VARIABLE IN A FACTORIAL STUDY. MAIN EFFECTS EXAMPLE: SUBLIMINAL
   SELF-HELP TAPES -

TRUE EXPERIMENTS (CONT')
* FACTORIAL DESIGNS - CONTINUED:
* THESE DESIGNS CAN TELL RESEARCHERS ABOUT TWO OR MORE MAIN EFFECTS
   HOWEVER, THE MOST IMPORTANT OF THESE ADVANTAGS IN THE FACT THAT
   FACTORIAL DESIGNS ALLOW RESEARCHERS TO DETECT INTERACTIONS.
* INTERACTION: CAN ONLY BE APPLIED TO DESIGN WITH MORE THAN ONE
   INDEPENDENT VARIABLE BECAUSE IT REFERS TO A SITUATION IN WHICH TWO OR
   MORE INDEPENDENT VARIABLES WORK TOGETHER TO INFLUENCE A DEPENDENT
   VARIABLE.
* MEANING THAT THE EFFECT OF ONE OF THE INDEPENDENT VARIABLES DEPENDS
   ON THE LEVEL (PRESENCE OR ABSENCE) OF THE OTHER INDEPENDENT VARIABLE.

TRUE EXPERIMENTS (CONT')
* FACTORIAL DESIGNS - CONTINUED:
* EXAMPLE: GROWTH RATE OF BEAN PLANTS: INDEPENDENT (1) AMOUNT OF
   SUNLIGHT, (2) AMOUNT OF WATER. 2 X 2
* TO FIND INTERACTIONS: DO A SIMPLE EFFECTS TEST: SEE WHICH SPECIFIC MEAN
   COMPARISONS ARE SIGNIFICANT IN THEIR FACTORIAL STUDY. - CALLED t-TESTS.
* FACTORIAL DESIGNS ALLOW US TO LOOK FOR MORE THAN ONE EFFECT AT A TIME
   IN A SINGLE STUDY. - THEY TELL US THE WHOLE STORY TOO. (COMPREHENSIVE
   AND EFFICIENT). ALLOW US TO SEE HOW DIFFERENT VARIABLES WORK TOGETHER
   TO INFLUENCE WHAT WE ARE LOOKING FOR/OUR RESULTS.
 
TRUE EXPERIMENTS: ARE THEY ALWAYS MORE INTERNALLY VALID THAN QUASI-EXPERIMENTS?
* IN STUDIES LOOKING AT SELF CONCEPT - QUASI-EXPERIMENTAL DESIGNS MIGHT
   BE BEST - HARD TO MANIPULATE PEOPLE’S SELF CONCEPT.
WITHIN-SUBJECTS DESIGNS:
* WITHIN-SUBJECTS/REPEATED MEASURES DESIGNS: THOSE IN WHICH EACH
   PARTICIPANT SERVES IN MORE THAN ONE (PERHAPS ALL) OF THE CONDITIONS OF
   AN EXPERIMENT.
* MAIN DIFFERENCE: THE FACT THAT THE WITHIN-SUBJECTS ANOVA TAKES INTO
   ACCOUNT THAT THE OBSERVATIONS IN THE DIFFERENT CELLS OF A
   WITHIN-SUBJECTS DESIGN ARE NOT STATISTICALLY INDEPENDENT (BECAUSE
   THEY ARE NO LONGER COMING FROM SEPARATE PEOPLE).
* ADVANTAGES OF WITHIN-SUBJECTS DESIGNS: REQUIRE FEWER PARTICIPANTS;
   ESPECIALLY IN COMPLEX FACTORIAL DESIGNS THAT HAVE MANY DIFFERENT
   CONDITIONS.

WITHIN-SUBJECTS STUDIES (CONT')
* DISADVANTAGES OF WITHIN-SUBJECTS DESIGNS:
* (1) PEOPLE’S PSYCHOLOGICAL STATES CHANGE AS THEY WORK THEIR WAY
   THROUGH A TASK (TIRED, EXCITED ETC.): ORDER EFFECTS OR SEQUENCE EFFECTS
    - CO-OCCUR WHEN FATIGUE OR THE PASSAGE OF TIME BEGINS TO TAKE ITS TOLL
   ON PERSON'S RESPONSES.
* (2) CONTAMINATION EFFECTS: OCCUR WHEN SOME ASPECT OF PEOPLE’S
   EXPERIENCE IN EARLIER CONDITIONS OF A WITHIN-SUBJECTS STUDY INFLUENCES
   THEIR RESPONSES IN A LATER CONDITION OF THE STUDY (ALSO REFERRED TO AS
   CARRYOVER EFFECTS AND INCLUDE PRACTICE EFFECTS AND INTERFERENCE
   EFFECTS).
 
WITHIN-SUBJECTS EXPERIMENTS (CONT')
* SOLUTIONS:
* USE CONTROL TECHNIQUE: COUNTERBALANCING - THE RESEARCHER VARIES THE
   ORDER IN WHICH PARTICIPANTS EXPERIENCE THE DIFFERENT CONDITIONS OF THE
   STUDY
* KINDS OF COUNTERBALANCING:
* COMPLETE COUNTERBALANCING: PRESENT EVERY POSSIBLE ORDER OF ALL OF
   THEIR EXPERIMENTAL CONDITIONS. DIFFERENT PARTICIPANTS RECEIVE
   DIFFERENT ORDERS
* IMPRACTICAL TO USE COMPLETE COUNTERBALANCING? USE REVERSE
   COUNTERBALANCING: GENERATE A SINGLE ORDER AND THEN REVERSE IT -.
   CONDITION C IS ALWAYS SMACK IN THE MIDDLE OF THE EXPERIMENT. -
* DO PARTIAL COUNTERBALANCING: CHOOSING A LIMITED NUMBER OF ORDERS SAY
  10 OR 12 AT RANDOM FROM THE POOL OF ALL POSSIBLE ORDERS.