PSY282 chapter 15
Terms in this set (122)
a desugn that compares two conditions of one independent variable
So all participants are consistently exposed to the conditions to the same degree.
Why seek reliable manipulation?
So that differences between scores between conditions are really due to the differences in our independent variable.
Why seek internal validity?
So that the way the samples operate is really the way that the general population operates
Why seek external validity?
Used to analyze a two sample experiment that consists of independent samples
Independent samples t-test
We randomly selected participants and assign them to a sample without regard to other participants selected for either sample
inherent, personal characteristics that distinguish one individual from another, may include physical and mental ability, attitudes and emotions,personal history and experiences, and social or economic level.
Participant variables that fluctuate within conditions threaten this because differences between participants cause them to respond differently and inconsistently, resulting in error variance.
What threatens reliability in a between subjects design?
Participant variables that fluctuate BETWEEN conditions threaten this, because w e do cannot know whether differences in scores between the conditions are due to our manipulation or to the participant variable.
What threatens internal validity in a between-subjects design?
A participant variable that is correlated witht hte influence of the independent variable or with performance on the dependent variable.
Describe potential variables to control
By mixing them, so that differences in a variable are balanced out in each condition.
Who does random assignment control participant variables?
We control it by systematically balancing or counterbalancing its influence within each condition. We could do this by making the variable a part of the seleciton criteria or pretesting participants.
How can we balance a participant variable?
Limit the population based ont hat variable, so we keep the variable ocnstant and it can't influence the results.
Alternative to counterbalancing a participant variable
Increases internal vilidity by eliminating a potential confounding that might occur w/random assignment AND Increases power by reducing the error variance, the more similar the participants, the less variable the scores are likely to be within each condition.
PROS of limiting population
If we become to selective, we may create a restriction of range problem AND because we are more selective in choosing participants, they represent a more limited populaiton, and so external validity is reduced.
CONS of limiting the populaiton
Increased power and internal validity, therefoe the advantages of limiting the population usually outweight the disadvantages.
What do MOST researchers opt for, even at the expense of external validity?
Depends on how much it threatens internal validity
What influences final decision to counterbalance an extraneous variable?
random selection of 2 indep. samples from pop., DV normally distrib. or roughly(if n>30) and interval or ratio scale, homogeniety of variance, sample sizes in each group don't have to be equal.
Assumptions of Independent samples t-test.
True variance of the populations represented by samples are equal
Homogeniety of variance
the probability of obtaining our difference between means when the null hypothesis is true.
What does a independent samples t-test determine?
distribution of all possble differences between two means when they are drawn from the raw score population described by the null hypothesis.
Sampling distribution of differences between the means
The weighted average of the sample variance
the estimated standard deviation of the samploing distribution of differences between the means.
standard error if the difference
Describes a range of differences between two population means, any one of which is likely to be represneted by the differences between out tow sample means.
What does a confidence interval for the difference between the two population means do?
means to combine scores from the different amounts or categories of that variable
collapsing across avariable
By computing the point-biserial correlaiton coefficient.(r pb)
How is the strength of a relaitonship between the IV and the DV in a two sample experiment described?
measures the proportion of variance in the DEP. scores that is accounted for by changing the conditions ot the IV.
The SQUARED point-biserial coefficeint measures what?
the proportion of variance accounted tor
the more consistently the Depndent scores change as the conditions of the Indpendent variable change.
The larger the effect size the WHAT?
Larger differences in scores between the conditions AND smaller variability of scores within conditions AND larger "n"
How can the power of the 2-sample t-test be increased?
Participants are selcted for a condition without regard for hwo is selected in another condition, and each parti icpant serves in only one condition(has two independent samples).
prior to conducting a study we measure potential participants on the variable to be controlled
Not guaranteed to balance part.variables within eahc condition,works less well with small samples, when it does balance out a vairable effectively the variable can still fluctuate within the condition so it can produce larger error variance among scoresw within the condition resulting in a weaker relationship
Three potential problems with random assignment
Partiicpant scores measured under TWO conditions of theinde pendent variable. Group 1 exp condition 1, group 2 does not exp.condition 1. OR Group 1 caffein and group 2 no caffeine
created by matching each subject in one sample with a subject in the other samnple OR by repeatedly measuring the same subject under all conditions (AKA: related samples)
Design in which each participant in one condition is matched with a subject in every other condition along an extraneous subject variable
Order effects that occur from one order of conditions or trials that cannot be balanced out by another order
nosymettrical carryover effects
Design in which partiicpants are measured before and after a treatment
Design in which each participant is measured repeatedly under all conditions of an independent variable
the standard deviation of the sampling distribution of mean differences betweeen dependent samples in a two-sample experiment
standard error of the mean difference
bias that arises due to participants exxpereinces that influence repeated measures
bias that arises due to the changes that occur as an individual grows older and more amture that influence repeated measures
bias that rises when participants fail to show up for a study or discontinue the participation before the study is completed
the sampling distribution of all possible alues of t that occur when samples of a particular size represent the raw score population described by the null hypothesis
statistical procedure that is appropriate for significance testing when the scores meet the requirements of a parametric test, the design involves matched groups OR repeated measures, AND there are only two conditions of the independent variable.
t-test for dependent samples
statistical procedure used for significance testing that is appropriate when the scores meet the requirements of a parametric test, the design involves independent samples AND there are only two conditions of the independent variable.
t-test for independent samples
A design in which all factors are within subjects factors, for each partiicpant in one condition there is a comparable participant in the other condition.
type of factor created when an independent variable is studied using dependent samples in all conditions, either because participants are matched or repeatedly measured
Individual differences in between-subject experiments are produced by WHAT?
The probability of rejecting Ho when it is false.
Power is defined as WHAT?
Two sample between-subjects approach AND the Matched-subjects design
What are the two possible approaches to collecting independent samples for a between subjects experiment?
two seperate samples/gorups woth different people in each group or condition
two sample between subjects approach
Different people in each group or condition, but individuals are equated or matched on certain factors (gender, age, SES, etc.)
Selection of dependent samples for mop.,DV normally distrib.(if N>30) and interval or ratio scores, homogeniety of variance, sample size in each group have to be EQUAL.
What are the assumptions of Dependent sampels t-test?
Two sets of data from ONE sample of people
two-sample within-subjects approach in a dependent samples t-test
participants perform better over time
participants perform worse over time
one trial impacts another
Habitual pattern (bias toward responding in a particular way because of previous responses made
subject history variables
older, wiser, more expereinced
participant quits study
sample mean difference/estimated standard error
A repeated measures Design in which each participant is measured under all conditions of an independent variable.
pretest-post-test design is also know as WHAT?
subject history, maturation, mortality, and order effects
What are possible confoundings of repeated measures design.
that the population mean difference falls in the interval
When we construct a 95% confidence interval we are sure of WHAT?
the strength of the relationship
A significant result on an independent-samples t-test can tell us that a relationship exists between two variables but it CANNOT tell us the WHAT?
t(34)=2.3, p<.05, two-tailed.
Significant written in APA
t(34)=2.3, p>.06, two tailed
describes reults that are considered too unlikely to result from chance sampling error if the predicted relationship does not exist, indicates rejection of the null hypothesis
The difference betweent he sample means in our study AND the difference in the population means divided by the standard error of the difference.
What does t-obt indicate in an independent samples t-test?
seldom occur and thus poorly represent two population means with no difference.
If t-obt is in the region of rejection, it shows that means like those in the two samples on which t-obt is based DO WHAT?
The variance in one population is equal to the variance in the other population
What does the homogeniety of variance state?
rejecting Ho when it is false
Power is defined as the probability of WHAT?
seldom occur and thus poorly represent two population means with no difference.
If t-obt is in the region of rejection, it shows that means like those in the two samples on which t-obt its based do WHAT?
Individual differences in between-subjects experiments are produced by WHAT?
amount of expected error when you use your sample mean difference to represent population mean difference
Waht does the standard error of the difference tell us?
If a researcher predicts that the experimental treatment will (mu1-mu2<0) how will the null hypothesis be stated?
A standard erro variance using the weighted average of the sample variance
What is a pooled variance?
What does the null hypothesis state in a twotailed independent samples experiment?
collapsing accross a variable
Combining scores from different levels of a variable is WHAT?
minimize the variability of the raw scores
One way to increase power is to WHAT?
Larger difference inswcores between conditions, smaller variability within conditions, a larger N
How do you go about minimizing variability across scores?
Which of these research terms is different from the others...order effects, mortality effects, history effects, research effects?
Michael is using a repeated-measures design in a 2 year study of depression. In the study participants come in monthly to provide depression rating. However, he is worried that participants continuing expereinces between measurements could change their responses. In other words Michael is concerned about WHAT?
individuals change moment to moment so may differ from one condition to the next, could be demand characterisits(part.recog.purpose or hyp of exper.). OR subject history, ,subject maturation..all of thes may reduce internal validity.Subject mortality may reduce external validyt
What are common criticisms of repeated-measures or within-subjects designs?
Same participant repeatedly observed in ONE condition,
Same participant repeatedly observed in ALL conditions
Try to obtain measures in short time span,make mechanics easy, interesting,(demand characteristics) and brief to reduce mortality.
What are some ways to counter influences of history, maturation,mortality, etc?
subject history, subject maturation, subject mortality, and demand characteristics and order effects.
A repeated measures design may be confounded by WHAT?
Influence of performing a series of trials they include....Practice effects, fatigue effects,carry-over effects, response sets,
What are order effects?
geting better at the task over trials
getting worse at the task over trials
the expereince of any one thrial that influences scores on subsequent trials
from previous trials, developing a habitual response for subsequent trials
John is doing a within-subjects study in which he presents sets of stimuli that vary along several perceptual dimensions( e.g. color and shape). He is concerned that the stimuli in one condition will influence performance in subsequent conditions. To reduce this possibility, he should make use of WHAT?
equal N's in each group such that each score in one condition be paried with a particular score in the other condition.
The assumptions of the t-test for dependent samples are the same sas those for the t-test for independent samples EXCEPT for the requirment THAT WHAT?
Matched groups design
Jack is doing a project in educational psychology. Before the experiment, he rank orders all the aprticipants according to achievement test scores. He takes the top two scores and randomly assigns one to the experimental group and one to the control group. THen he takes the next two scores and randomly assignes one to the experiemental group and one to the control group. He continues in this manner until all participants have been assigned. What procedure is Jack using?
Use a between subjects design
Suppose that you discovered that the task in your study does not alow repeated testing. What should you do?
t-obt will increase.
If all other factors are held constant, a decrease in the variability of the D scores will have what kind of effect on the t-obt?
Both repeated measures and within subjects design would be appropriate
two random samples were taken in which GRE scores of 15 male engineering students (male group) were compared with those of 15 female engineering students (female group). The appropriate design for testing the significance of the difference between the means is a WHAT?
If the researcher thinks thatthe experimental treatment will produce an increase in a related samples t-test, how will the alternative hypothesis be stated?
by converting the raw scores to difference scores and then dividing by the standard error of the mean difference and then comparing t-obt to a t-critical value
How is the t-test for dependent samples performed?
Ho: muD< or equal to 0
If a researcher predicts that the experiemental treatment will produce an increase in a related-samples t-test, how will the null hypothesis be stated?
a within subjects or repeated-measures design
Aysha has her participants expereince every condition of her study. That way, each participant serves as her own control. Such a research technique is called WHAT?
Independent samples t-test and Dependent samples t-test
A scientist has conducted a two-sample experiemnt what two versions of a parametric procedure are available to him?
Whether the design involves independent samples or not
What is the deciding factor between the ind.samples t-test and dep. samples t-test?
Increase internal and external validity
What positive impact does counterbalancing have on internal and external validity?
Increases error variance and reduces power
What negative impact does counterbalancing have?
May alert participants to variables being studied or communicate demand characteristics
What problems arise from pretesting partiicpants?
a within-subjects design, because the dependent samples t-test is more powerful
what is more powerful a dependent samples or independent samples design? Why?
They reduce internal validity because differences in responses between conditions may result from these factors instead of the conditions the independent variable. They reduce external validity because the behaviors and participants do not represent those found in other situations where these factors are different or not present.
How do the influences subject mortality, subject history, and subject maturation bias your results?
by testing the same participants under all conditions of the IV.
How is a repeated-measures design created?
It provides participants who have virtually identical participant variables in all conditions
How does a repeated-measures design control subject variables?
Conditions are prone to confounding by part. history and maturation,by demand characterisitics resulting from expereincing all conditions, by subject mortality, and by order effects between conditions.
What problems are associated with a repeated-measures design?
Graph the results, compute the confidence interval, and compute thte effect size(rpbsquared)
A researcher has obtained a statistically significant two-sample t-obt. What additional procedures are needed to complete the analysis?
a repeated measures (within-subjects design and it is analyzed using a dependent samples t-test.
What type of design is present in a pretest-posttest design and how is it analyzed.
a. It is impossible to test people when thhey ar emale and then again when they are female b.match each male with a corresponding female on relevent participant variables
A rather dim student proposes testing the conditions of "male" and "female" as a repeated measures study..What is wrong with this idea and what control techniques can be applied instead?
When carryovover effects are ono-symmetrical, when the task does not allow repeated testing, or when extensive ounterbalancing is unwise.
When is a between-subjects (independent t-test) preferred over a within subjects (dependent t-test)?
When an experiemnt involves tow conditions and is either a matched-groups or repeated measures design.
When is a dependent sampls t-test preferred?
Counter balancing of the order in which participants perform the conditions must be done.
If repeated measures desingn s used what else must be done?
Within subjects (Dependent t-test)
matched groups design is used in what kind of experiment?