Class Notes (1,100,000)
CA (650,000)
UTSC (30,000)
Psychology (9,000)
PSYB01H3 (200)
Anna Nagy (100)

Chapter notes

Course Code
Anna Nagy

This preview shows page 1. to view the full 5 pages of the document.
Chapter 8- Experimental Design 1
In the experimental method:
-all extraneous variable controlled
-advantage: allows a relatively unambiguous interpretation of results
-manipulates independent variable (IV) to see if groups differ in levels on the dependent
variable (DV)
oall other variables are kept constant- through either direct experimental control or
-confounding variable: varies along with the IV
-confounding occurs when the effect of the IV and an uncontrolled var are intertwined (dont know
which one is responsible for observed effect)
-a good experimental design- involves eliminating possible confounding that results in alternative
explanations (need to eliminate them)
-when results can are confidently attributed to the independ var (IV)= internal validity
*Chapter Focus* on true experimental designs that provide the highest degree of internal validity
*Chap 11- Quasi* lacks crucial element of random assignment; but still allows inference that an IV has
an effect on the DV
-simplest possible experimental design has 2 vars: the IV and the DV
-IV has 2 levels
oExperimental group
oControl group
-Either vars are kept constant in both groups, or use randomization so any extraneous vars affect
groups equally
-The basic simple experimental design can take 1 or 2 forms:
-1. Posttest-Only Design
oResearcher must
(1) obtain 2 equivalent groups of participants
(2) introduce the IV
(3) measure the effect of the IV on the DV
Participants (R= randomly assign)Experimental Group Measure
Control Group Measure
Steps to design:
(1)choose participants and assign to the two groups
You're Reading a Preview

Unlock to view full version

Only page 1 are available for preview. Some parts have been intentionally blurred.

Chapter 8- Experimental Design 2
a.procedure must eliminate any potential selection differences
i.people cannot differ in any systematic way- i.e. High income ppl- grp 1; low
income- grp 2
ii.can use random assignment into groups
iii.can have same participants, participate in both conditions
(2)Choose two levels of the IV
a.i.e. experimental grp- receives treatment; control group- does not
b.i.e.2- test btw 1 hr of training with 1 grp and 10 hrs with the other
(3)Lastly, Effect of IV is measured
a.Same measurement procedure is used for both groups
b.B/c groups are equiv- any differences can be attributed to the IV
i.Thus having internal validity
c.Normally, a statistical significance test would be used to assess the difference btw grps
-2. Pretest-Posttest Design
oOnly difference btw this design and the one above is that in THIS design a pretest is given
before the experimental manipulation is introduced
makes it possible to ascertain that the groups were in fact equivalent at the
beginning of the experiment
not necessary though, if participants are randomly assigned
oespecially in lrg grp sizes- randomization very effective
othe larger the sample, the less likelihood that the groups will differ systematically prior to
statistically method available to determine, but RULE OF THUMB= 20,30 ppl min,
per condition
Advantages and Disadvantages of the Two Designs
-pros and cons of whether to include a pretest
oAllows to assess equivalence of (even with randomization)- maybe with small groups *imp
oCan use as a selection process of participants
oMight need it to find highest and lowest scorers on a test (for example extent of change on
oNecessary when its possible participants will drop out of experiment
Dropout factor =mortality
oPretest enables you to assess effects of mortality (looking at pretest scores of dropouts) and
how it affected the final results
Can find: is mortality a plausible alternative explanation
oTime-consuming & awkward to administer (for particular experiment)
oCan also sensitize participants to what you are studyingand figure out your hypothesis
Thus, act differently than they would if they didnt know the manipulation
TO AVOID problem- Pretest can be disguised- either with diff experimenter OR
embed pretest into a set of irrelevant measures (of no interest )
You're Reading a Preview

Unlock to view full version