research methods 2

Helpfulness: 0
Set Details Share
created 2 years ago by morganlynn1224
updated 2 years ago by morganlynn1224
show moreless
Page to share:
Embed this setcancel
code changes based on your size selection

main characteristic of a study



independent variable

factor of interest to the experimenter in which they systematically manipulate

the one being studied to see if it has an effect/ will influence behavior

must have at least 2 levels - ex: dose 1 and dose 2


subject variables

characteristics inherent in the subjects


field experiments

research conducted outside the lab, using experimental and non-experimental methods


3rd variable

any variable that you have not included in your design that you are not trying to find a cause and effect


factor that is varied vs. held constant

independent variable vs. control


dependent variable

result that is observed

measured outcomes of experiments


situational variables

variations in the features of the environment the participant encounters

ex: person in need of help - one bystander vs many bystanders is the different situations

room with all women vs. room with mixed genders



big request then small request

makes people more likely to say yes because the small request seems so small compared to the big request from before


task variables

variation in type of task performed by participants

ex: levels of complexity of a puzzle, maze, etc.


instructional variable

variations in instructions given to groups of participants

ex: participants in a memory study are told to memorize the content in different ways


control group

identical to those in experimental group, but do not get the manipulation


random sample

allows us to generalize to the population of interest (representativeness)


interaction effects

IV's interact if the effect of one of the variables differs depending on the level of the other

multiple IV's allow for hypotheses


random assignment

roughly gets rid of extraneous variables (group equivalence)


extraneous variables

uncontrolled factors, not of interest to the researcher, that might influence the behavior being studied



a variable that co-varies with the independent variable

could be an explanation for results

as the IV changes, the other moves at the same time

ex: age vs. driving ability
confound = years of experience
extraneous variable = eyesight, car they drive


quality of study depends on...

choice and operational definition of dependent variable


subject variables

compare groups based on already existing characteristics of individuals

ex: gender, age, personality characteristic


ceiling effect

the results of a study are so high that no difference can be determined between conditions


floor effect

the results of a study are so low, usually because the task was too difficult, so there is no observed differences between conditions


when can we rule out extraneous variables

only in a manipulated study, not when using subject variables.

when we manipulate the variable, we are able to remove the extraneous variables, with subject variables we cannot


why can't we say subject variables cause the change

we cannot eliminate alternative explanations because we are unable to rule out extraneous variables and hold everything else constant

we can say the IV precedes the dependent variable and covaries with it/ the groups performed differently on the dependent measure


studies using subject variables are called

ex post facto studies, or quasi experiments


statistical conclusion validity

making conclusions about the associations between the independent and dependent variables in a study

were statistics used and interpreted correctly?


how is statistical conclusion validity reduced

(this is bad)

researcher leaves out some of the results, uses the wrong scale of measurement or wrong analysis, violate some assumptions required for some particular analysis


construct validity

did you describe the IV and DV properly?

are the operational definitions accurate?


external validity

how well do the findings from a study generalize beyond the specific context of the experiment

do the results generalize to other populations

can you say this is relatable to other parts of the country/world?


convenience sampling

using people that are available to you

ex: PSY 110 studies

must be skeptical about that


internal validity

to what degree is a study methodologically sound and confound-free

are the results most likely due to the independent variable and not some other factor - if so, high internal validity


threats to internal validity

pre-post design - history, maturation, regression, testing, instrumentation

participant problems - subject selection, attrition

alternate explanations for the results


pre-post studies

will people change as a result of some experience?

pretest -> experience -> posttest



an event that takes place between pre and post testing that produces change unrelated to the treatment program



natural changes in the organism due to the passing of time

the participants mature in some area which changes the results

ex: college students questioned as freshman may have adjusted and matured to college life by the second time they are questioned


regression (towards the mean)

if your pre-test measure was crazy high, your next testing measure will have a score closer to the mean

based on chance alone

having a control group allows a researcher to spot possible regression effects



having taken the test before affects your results on the next test

practice effect may arise because they have taken the test before and know what to expect


ecological validity

will results from a lab be transferred to the real world well

research with relevance for the everyday activities of people trying to adapt to their environment

another form of external validity
do results generalize to other times?


how is internal validity reduced

any uncontrolled, extraneous factor

big problem when there is no control group


pretest sensitization

participants taking the pretest get a hint on what the test is about so they pay closer attention to that topic before taking the posttest and score higher than normal



when measurement instrument changes from pretest to posttest

- is posttest "easier" than pretest?
- are observations more accurate at posttest?
- is the same measure appropriate at both tests?


subject selection effects

participants in different conditions must be equivalent to each other, except for the independent variable

ex: effect of teaching method on performance
students take lecture or lecture/discussion and who does better?

could this be because of the method or because of the difference in the students?


participant problems

characteristics of the participants can affect the internal validity of your study



people who did not finish the study for any reason (moving, death, lost interest)

longer studies lose more people over time


what should you do about attrition?

compare the people who dropped out for similarities/differences and make note of this - was there a connection?

are the people who dropped out different in characteristics than the ones who finished?


how is attrition a threat to external validity?

maybe the people who finished were the ones who needed the money, while the people who dropped out didn't.

this means the final sample relates to desperate people, and not everyone


why can't you have pre-post threats when you only test a person once?

because you have nothing to compare them to

how can you know there was an effect?


what increases the likelihood of pre-post threats?

the longer the time between the pre and post test

longer amount of time = longer time for one of these threats to occur


two types of experimental design

between subjects (independent groups) and within subjects (repeated measures)


between-subjects design (independent groups design)

different groups of participants contribute data for different levels of the independent variable

each participant only experiences one level of the IV - either control or IV group

ex: each person gets either A OR B


within-subjects design (repeated measures designs)

same participants contribute to all the levels of the independent variable

each participant experiences all levels of IV

often used for sensation and perception studies

ex: each person gets A AND B


pros of between-subjects design

- each person comes in equally naive to the independent variable (come in, do experiment, leave)

- they don't have to stay long because they only do one level of IV (come in, do experiment, leave)


cons of between-subjects design

- needs a large number of subjects to fill all experimental conditions

- differences between conditions may be due to IV (you hope so), but could be due to pre-existing differences between the groups

- need to create equivalent groups that are the same in all important ways except the IV


when would a between-subjects design be the only design option?

- when the independent variable is a subject variable
ex: males vs females, introvert vs extrovert

- if experience of participating at one level makes it impossible to participate in other levels
ex: eyewitness memory from watching a video - you can't "unsee" it without bias/influence


random assignment

method of placing participants into different groups, where each subject has an equal chance of being placed in any group

spreads potentially confounding factors evenly throughout the groups

sometimes you end up with unequal N


ways to create group equivalence

random assignment, block randomization, matching


block randomization

used to ensure an equal number of participants per group

each condition of the study has a participant randomly assigned to it before any condition is repeated a second time



participants are grouped together on some trait and then randomly distributed to different groups

good because you can control for factors that would threaten the internal validity of the study

useful when there are not enough participants for random assignment to work well

ex: find two people with same IQ and match them


conditions needed for matching

- you need a good reason to believe that the matching variable is correlated with the dependent variable

- you need a good way of measuring the matching variable you're controlling for


pros of within-subjects design

- fewer people needed to fill all experimental conditions because everyone experiences everything

- don't need to worry about equivalent groups, you are your own control


cons of within-subjects design

- sequence or order effects - once a person has completed the first part of a study, the experience could influence performance in later parts


progressive effects

after trial 1, performance is better on trial 2 - practice effect

after trial 1, performance is worse on trial 2 - fatigue/boredom


carryover effects

order effect - the sequence in which the conditions are presented has an influence

A -> B has different effect than B -> A



present conditions in more than one sequence

used to minimize progressive effects

less effective at minimizing carryover effects because the order in which the sequence is presented will have an effect
ex: unpredictable vs. predictable noise study - people who heard UPN first get annoyed when later doing PN and people who start with PN later think they can do UPN well cause they already did it once


counterbalancing when testing once per condition

complete counterbalancing

partial counterbalancing


complete counterbalancing

every possible sequence of conditions is used at least once

but this may require many sequences if there are many conditions because if the number of levels of the IV increases, this adds new possible sequences


partial counterbalancing

a subset of total possible sequences are used

use randomization to select for a few of all of the possible sequences


latin square

a way of making sure that each sequence is used only once and in only one order
ex: is A is first in any row, it will never be first again

ex: AB will never happen again but BA might


counterbalancing when testing more than once per condition

reverse counterbalancing and block randomization


reverse counterbalancing

for each participant, present the conditions in one order and then again in reverse order

ex: ABCD then DCBA


block randomization for controlling sequence effects

every condition occurs once before any condition is repeated a second time

use some kind of randomization to give a random order

ex: BCDA then CADB


types of design when independent variable is age

cross sectional or longitudinal


cross-sectional design

a between-subjects design

ex: compare 3 y/olds to 4 y/olds to 5 y/olds all at the same time

problem of cohort effects - history effect, rapid changes, people born at different times vary in many ways


longitudinal design

a within-subjects design

ex: compare language performance in the same group of kids at 3, 4 and 5 yrs/old

problems? attrition can occur - people move, drop out, are sick at follow up

feasibility of collecting data for 3 years - school board changes and doesn't allow data collection anymore



group of people born at about the same time


terman study

studied gifted children from 1877-1956 to see if students who scored high on the tests early in life, really went on to be gifted

longest running repeated-measures design in the history of psychology

found that the students (who started in the top 1% of the population with avg. IQ of 150) went on to produce tons of great things like books, scientific papers, etc

had no problems with attrition - the percentage of people who remained in the study even over that many years was over 90%



preconceived expectation about what should happen in an experiment


experimenter bias

going into the study, the experimenter has some ideas of what they want the study to produce

the way the experimenter innocently interacts with the subject can influence results that confirm your hypothesis

ex: smiling at correct answers


rosenthal effect

independent variable was experimenter expectancy!!

study was conducted of participants seeing pictures and rating how successful the person looked

experimenters were told that either 1. to expect negative results or 2. to expect positive results

even though the pictures were the same for both groups, the negative experimenters group saw negative results and the positive experimenter's group saw positive results


double blind procedure

neither the experimenter nor the participant know what to expect of the study, neither know what condition group each participant is placed in

usually a researcher sets up the experiment and a grad student collects the data


single blind procedure

the subjects do not know which experimental group they are in, but the experimenter does


subject bias

knowing you are in an experiement can make you change your normal behavior

good subject - gives you the data they think you want

bad subject - wants to mess with data


ways of experimenter bias

ways in which instructions are given

facial expressions when answers are given
ex: participant changes response to get "positive" reaction from experimenter

personality characteristics
ex: preschoolers do better when experimenter is caring vs indifferent


ways to control for experimenter bias

minimize contact between experimenter and participant

ex: computerize assessment or double-blind procedure

have a research assistant administer the study rather than the researcher


hawthorne effect

you change your behavior simply because you know you're being studied

belief of participant that they are part of a special group and focus of attention, always get positive outcomes from this but it is NOT due to IV


evaluation apprehension

stress because you know you're being observed

make an effort to present yourself as a good subject


demand characteristics

characteristics of the study that demand the participant's attention

as a result, tend to show the participant what the study is about

(happens more in within-subjects because they see all varying levels)


controlling participant bias

minimize demand characteristics
ex: use deception so participant behaves more naturally

use of placebo or control group
ex: everything the same except content of treatment

manipulation check
ex: ask participant what they think the hypothesis is


anonymous studies

people still lie!


western electric

6 women workers were taken from an electric plant and put into a special room making special parts. they were observed for how efficient and quick they could make the parts. even when working conditions changed - work week extended, less rest given - they still produced more and more parts.

this is the origin of the hawthorne effect.

really they should have looked at this more deeply because it turns out that if you looked at the parts made per hour, it was much less than per week. they were afraid to complain about extended weeks because they didn't want to lose potential bonus money for being in the study


when are demand characteristics a bigger problem?

in within-subjects designs because participants are exposed to all levels so they may start catching on to the differences in exposures


placebo effect

being given a "fake" pill and thinking it actually worked



something that dies not have active ingredients

used to control for expectancies


manipulation check

take a small group of participants from the study, stop the study for them, and ask them what they think the hypothesis of the study is

this is to see if demand characteristics are playing a role


"other" manipulation check

to check if your IV is actually working

ex: you have a high/low anxiety group - give them an anxiety test to see if they actually are high/low


what is the defining feature of an independent groups design

random assignment to groups


why does an independent groups design have to be between subjects?

because the participants cannot be exposed to both levels of the IV


when is a matched group design useful?

when there are a small number of participants

there could be some attribute of the subjects that could affect the outcome

there is a good way to measure that attribute


matched groups design

use a matching procedure to match subjects on similar characteristics that are thought to affect the outcome

then use randomization to match each of those participants into either IV group


matching then random assignment vs. matching in ex post facto

matching then random assignment creates equivalent groups

matching in ex post facto makes the groups more similar to each other, but does NOT make group equivalence because they are groups based on subject variables


nonequivalent groups (ex post facto)

control groups are non-equivalent

groups are made on subject variables so you do not randomize

used to test differences in already existing conditions (IQ, head injury, etc)


repeated measures

always within-subjects, all participants experience both levels


how do you know if the two levels differ (one IV does better)?

t-test will tell you if this difference is larger than you expect due to chance alone

t-test for independent groups - between-subjects

t-test for dependent groups - within-subjects


gabriel and richtel article

a paper discussing the mixed results on how well a new, computerized teaching system works. many places said it does, many said it doesn't, many said you cannot base learning off standardized test scores, many found flaws in experimental design


how can you test for non-linear effects

test multiple levels of an IV

allows you to test for other explanations


why can't you use a line to show means

this assumes that it is a continuous variable


what does multilevel mean

more than 3!
there is always 2 levels

control vs. experiment is not multi level


bransford and johnson balloon study

measured the amount of ideas that could be recalled by people who were given a weird paragraph to read, the weird paragraph and an image directly corresponding to it, or the paragraph and an image somewhat corresponding with it.

perfect example of how adding levels to an experiment can significantly enhance understanding of some phenomenon


why is the yerkes-dodson law an example of a non-linear study?

it is shaped like an inverted -U

if you looked at low arousal and medium arousal, it would look like as arousal increases, so does performance

if you looked at only low and high arousal, it would look like there is no difference

in reality, low and high arousal means low performance, medium arousal leads to the best performance


multiple level design

allows you to test for other explanations and non-linear effects


placebo control group

participants are led to believe that they are receiving some kind of treatment but they really aren't

just thinking you are getting treated actually does help sometimes


waiting list control group

used in treatment outcome studies

people are "told" they are on the waitlist but they always end up getting the treatment at the end, but this way they can compare people with the treatment to people without it


why is a wait list control group not sufficient

the people in the actual control group are told not to do anything else (ex: seek other treatment) but people who think they're on the waitlist may go try to get treated otherswise


yoked control group

each member of the control group is matched to a member of the experimental group so that the time spent participating or the types of events encountered are kept constant

used when the participants in the experimental group are exposed to a varying number of events for a variable amount of time

ex: learned helplessness studies


advantages of multi-level designs

allows you to test for non-linear effects

can test more specific hypotheses and rule out alternative factors that could affect outcome

ex: caffeine and reaction time


mozart effect

people told to remember as many numbers as possible while listening to mozart, gentle rainstorm or nothing.

everyone did better on the 3rd trial comapred to 1st no matter what order the IV came in

problem is practice effect! the more you recall numbers, the easier it becomes for you


problem with using multiple t-tests to compare possible pairs of conditions?

the probability of type-1 error increases with every t-test


instead of t-test you can use

ANOVA - then use t-tests if you find a difference

problem with ANOVA - only tests if there is a difference, not which group it is in


presenting discrete vs. continuous data

write numbers out in text - hard if more than 3 levels

make a table with results

make a graph
discrete - bar graph
continuous - line graph


factorial design

study with more than 1 IV or factor, not just multiple independent variables

you must test for interaction effects!!


variable does not = ...

value or level!


2 level design means...

IV1 has 2 levels


interaction effect

the effect of 1 IV depends on or changes with the level of the other independent variable


main effect

ignoring the other variables


continuous variable

variable in which a number of intermediate values exist

ex: dosage of drug


interaction vs. main effects on a table

main effects are on the margins - these tell you if A OR B have an effect, not the effect of them together

interaction effects are in the box - this tells you if the interaction of one changes with the other


main advantage of factorial design

you can show interactions of groups

line graph with X means there is an interaction

parallel lines means there is no interaction


if the means in the margins are different...

there is a main effect for that variable


if the patterns within the boxes are different...

there is an interaction effect


factorial design can be:

all between-subjects

all within-subjects

at least one of each (mixed factorial)


text case study 19 - spider study

tested people with high and low self-efficacy for their fear when watching a video with a spider coming at/away from them when being told they are tied up or free to roam the room

this was a good example of why you need to do enough trials to find an effect


text case study 18 - students and studying

college students studied for an exam with headphones either playing cafeteria sounds or no sounds and then took a test on that material in either a noisy or quiet room. noisy and quiet studying was tested with noisy and quiet testing for interaction effects


P x E factorial design

P = person or subject variable (always between)

E = environment created by experimental manipulation (could be between or within)

PxE can never be all within-subjects, only mixed or all between


key word for between-subjects design


means they are only receiving one treatment, which means it is between


block randomization to control for sequence effects

tests multiple orders to test for order effects

advantage - no preset orders so you can test for many orders

disadvantage - less people per order, when you have more orders


# people needed for PxE mixed design

# people needed for between-subjects x # conditions


# people needed for PxE all between-subjects

# people for each condition x # conditions

Related pages

infraspinous musclenames of dna basesabdominal reflexsentence for vicissitudeadrenal gland and pancreassclerenchyma under microscopeamino acids phosphorylatedgrantia characteristicsonly digestive structure with three muscle layersfunction of the mesencephalonplant cell telophasehomework passes to printxanthine derivatives mechanism actionnet atp production in glycolysisfluid mosaic of cell membraneneurotransmitters of the sympathetic nervous systemthe emotion of anger usually appears at the age ofexplain the processes of transcription and translationthe function of the goblet cells is topsittacines and passerinesap biology test questions campbellquizlet maternity nursingmarketing kotler kellerwhat does atcc stand for in microbiologybinary fission explanationjames rosenquist house of firephotosynthesis light reaction and calvin cyclebrain functions quizlethuman oogenesisduring close vision the ciliary muscle ishuman anatomy and physiology marieb 9th edition pearsoncampbell biology pearsonlayers integumentary systemthe starting molecule for glycolysis isgeometry flash cards printablethe numbers in french 1-100the enzyme pepsin digestswhat activates pepsinosmoregulation quizpneumatic bonescomparison of nixon's d├ętente policies to previous cold war strategiesa virus is a tiny infectiousabnormal areolaeasy drawings of puppiesdiboron hexahydride formulawhere are lysosomes located in an animal cellnames of countries and capitalsone celled microorganisms are known asthe average temperature in the tropical rainforestmasteringaandp comspecifies the amino acid sequence for a proteinhow many pairs of spinal nerves are thereascending tracts in the spinal cord carryaxon synapse100 non rebreather maskthe digestive system quizanterior view of heart labeledmicrobe classificationsite of spongy bone in the adultendergonic chemical reactionjava introduction bookprintable ekg flashcardshow many daughter cells are in mitosismembranous systemstimulates spermatogenesisvocab level g unit 12venipuncture side effectsphysioex exercise 11 blood analysis answerswhich of the following statements about joints is trueaxial skeleton quiztranslucent cells containing keratinclusters of lymph nodesap bio chapter 22 testis b6 water soluble50 states and capitals study listheterozygous for a traitthe thyroid gland is embedded in the parathyroid tissueoperant conditioning worksheetsbeauty pageant dietcellular respiration easy