Chapter 9 Flashcards
3 properties of a well designed experiment
1: vary at least one independent variable
2: ensure initial equivalence of participants
3: control al extraneous variables
3 types of independent variables
environmental manipulations - manipulating aspects of the research setting
instructional manipulations - variation in instructions to participants
invasive manipulations - manipulating conditions in a participant’s body, ex stimulation, surgery, or drugs
pilot test
initial test with some participants to ensure that the different levels of independent variable are different enough to be detected by participants
.are these two levels of lights different enough to be perceptible or significant?
manipulation check
question during an experiment designed to determine whether the independent variable was manipulated successfully. like, did the participant notice the light was brighter in this condition compared to the participants in other conditions where the light was dimmer?
subject/participant variables
things we cannot manipulate but vary, like the gender of a participant
simple random assignment
every participant has an equal probability of being placed in any experimental condition
matched random assignment
- researcher obtains participant scores on a measure known to be relevant to outcome of experiment (typically a pretest measure of the dependent variable)
- participants are ranked from highest to lowest
- creates clusters based on the number of conditions in the experiment. for ex, 2 highest scoring would be split into condition a and condition b. next 2 highest scoring would be split as well. all the way to bottom scoring
randomized groups design
simple random assignment or matched random assignment
between-groups design
interested in behavior BETWEEN groups of participants
as in participants in condition a versus those in condition b
within-subjects design
.each participant experiences all conditions or all levels of independent variable
.don’t have to worry about randomizing, because no differences between conditions b/c all conditions experienced by each participant
.has more power!
.subject to order and carry over effects tho
order effects
behavior affected by the order in which they participate in the various conditions of the experiment
.three types: practice, fatigue, sensitization.
practice effects
performance improves because complete the dependent variable several times
.like if you do a simple memory test, they’re going to improve at it each subsequent time!
fatigue effects
participants become bored, tired, or less motivated as experiment progresses
.may perform least well on experimental condition they receive last
sensitization
after completing dependent variable several times, participants may begin to realize what the hypothesis of experiment is, then start to respond differently!
.aka they become sensitized to the purpose of the experiment
counterbalancing
guards against order effects
.present levels of independent variable in different orders to different participants
latin square design
.method of counterbalancing
.each condition appears once at each ordinal position (1st, 2nd, 3rd, etc.) and each condition precedes and follows every other condition once (SEE PAGE 193!!)
carryover effects
effect of a particular treatment condition persists even after the condition ends
.one level of IV is still present when the next level is introduced
treatment variance
.aka PRIMARY VARIANCE
.portion of variance in participants’ scores that is due to the independent variable
confound variance
.aka SECONDARY VARIANCE
.portion of variance in participants’ scores that is due to extraneous variables that differ systematically between the experimental groups.
.MUST BE ELIMINATED AT ALL COSTS
systematic variance
.aka BETWEEN-GROUPS VARIANCE
.part of total variance that reflects differences among the experimental groups
.composed of both treatment variance and confound variance
error variance
.aka WITHIN-GROUPS VARIANCE
.result of unsystematic differences among participants
.participants differ when enter experiment in terms of ability, personality, etc.
.experimenter will treat participants in slightly diff ways…
.measurement error…
.want to REDUCE but can’t really eliminate
internal validity
degree to which a researcher draws accurate conclusions about the effects of the independent variable
.experiment internally valid when it eliminates all potential sources of confound variance
counfounding
something other than the independent variable differs in some systematic way
.a fatal flaw!
biased assignment
.aka selection threat
.effects are due to nonequivalent groups rather than independent variable
differential attrition
loss of participants during a study that is different under each condition because of the nature of each condition
.e.g. one condition might be more stressful than the other, leading the ones that finish that condition to be different (more hardy) than the other condition which made less people drop out
pretest sensitization
taking a pretest may lead participants to react differently than they would have if they had not been pretested.
.give a pretest, then same test after experiment… are participants just getting better scores because they’re doing the same test a second time, or have they actually changed because of the IV?
history effects
one’s past experiences or history causes them to react differently to IV than other participants… like if everyone had to watch a rape video in one condition… and a couple of the people had experienced rape, it would dramatically skew the results
demand characteristics
.participants act in a way that they think the experimenters want them to act
.think if they had to go to the bathroom while being watched… no doubt EVERYONE would ALWAYS wash their hands
threats to internal validity, GO
.Biased assignment .Differential Attrition .Pretest Sensitization .History effects .Misc confounds
Sources of error variance
.individual differences .transient states .environmental factors .differential treatment .measurement error (make sure understand, on page 204 if need)
Experimenter’s dilemma
more tightly experimenter controls the setting, the more internally valid the results but lower the external validity
.but it’s ok because we don’t always prioritize external validity, not always wanting to generalize to whole population or something
.however if results aren’t internally valid, they can’t ever be generalized or externally valid
.so internal validity is more important
pros cons of web-based research
PROS
.large sample sizes
.diverse sample sizes
.easy to get specific samples with specific characteristics
.less susceptible to social desirability effects and experimenter expectancies… etc.
CONS
.no way of confirming one’s identity or what they say, could take it many times
.can’t control setting where people take test
.people often get bored halfway through and don’t complete