A. how to take statements, some have an assumption, a # have a qualification, they all have strengths and/or an importance statements, and they also typically have weakness statement
*understand how you apply each of these probability sample design
Simple random samples (most basic, list of people)
- find SF from population that you want to generalize
- find a table of random #'s
- label each element on the SF with #'s from 1 to N in order they appear on the
list
-go to table and randomly select a # which represents a person in SF
- continue to write more #'s from the sample that your consistent from fashion from the table
- each # picked represents a person of the SF to be
Systematic samples (list of people)
- pick a # between 1 and K as the 1st SF members to be selected for inclusion in the sample
- then pick every Kth case/element after that until you go to the end of the list
K=1/sampling ratio
K=1/(sampling
size/population size)
5% sample of 100 ---K=1/(5/100)=1(1/20)=20=K
10% sample of 100 ---K 1/(10/100)=1(1/10)=10=K
20% sample of 100 ---K 1/(20/100)=1(1/5)=5=K
*the larger the designed sample the smaller K will be (the smaller K is, the bigger the sample)
**Sample of 160 ---- SF=16000
K= 1/(sample size/population size)=1(160/16000)=100=K
Stratified samples (list of people, time & effort)
- you would take a simple (or systematic) sample within each of the strata.
easiest way to think about this is to divide the SF into its relevant strata and then assign a sampling ration(%) that will be randomly taken from each category of the stratification variable
Cluster samples (list of grouping)
*randomly selet a # of grouping (called cluster) from a list with groups on it
*randomly select smaller sampling elements from the chosen group (clusters)
*they solve the problem of not having a list of elements for the population that you want to generalize
to
LIST OF GROUPING
Probability Proportionate to Size (PPS) Samples
*probability cluster picked = size of the cluster/size of population
*randonly chose the same # of elements from each of the chosen clusters
**you can use a computer generated formula to do this
Disproportionate Sampling
*you could analyze the small group data separately and then analyze large group data separately and compare the results then it doesn't matter if the small group is disproportionate or
you could use statistical weighting to weight down the disproportionate back to the correct % distribution found in the population.
Myers' Psychology for the AP Course
3rd EditionC. Nathan DeWall, David G Myers
955 solutions
Social Psychology
10th EditionElliot Aronson, Robin M. Akert, Samuel R. Sommers, Timothy D. Wilson
525 solutions
Myers' Psychology for AP
2nd EditionDavid G Myers
901 solutions
Social Psychology
10th EditionElliot Aronson, Robin M. Akert, Timothy D. Wilson
525 solutions
- Social Science
- Psychology
-
Flashcards
-
Learn
-
Test
-
Match
-
Flashcards
-
Learn
-
Test
-
Match
Terms in this set (50)
Internal Validity
- the degree to which the results are attributable to the independent variable and not some other rival explanation
External Validity
the extent to which the results of a study can be generalized
Population validity
refers to generalizations related to other groups of people
Do the results generalize for males as well as females?
Ecological validity
refers to generalizations related to other settings, times, contexts, etc.
Does the behavior modification technique work at school as well as at
home?
Internal and external validity are related reciprocally which means
Controlling internal validity decreases external validity
Controlling external validity decreases internal validity
Often both types of validity are shown by:
Using one experiment to first demonstrate an effect in a highly controlled environment
(i.e., internal validity), then
Using a second experiment to replicate the study in a more realistic, natural setting
(i.e. external validity)
The Internal Validity
of a study hinges on control of the extraneous variables.
A study has good ______________ if the manipulation of the independent variable is the only reasonable explanation for differences between groups on the dependent variable.
A study with good _________________supports the conclusion that there is a cause-effect relationship
Threats are sometimes referred to as
"rival hypotheses" - Any alternative explanation(s) for results.
These threats weaken support for a cause-effect relationship.
Common Threats to Internal Validity
History
Maturation
Testing
Instrumentation
Statisitcal
Regression
Differential Selection
Mortality
Attrition
Interaction
History
Refers to any external event or occurrence, not part of experimental treatment, that may have an impact on the dependent variable
The longer the duration of a study, the more likely it is that history can be a threat
For example: In the exercise study, if the participants were teenagers that were also getting ready for
their high-school prom....
Maturation
Refers to the physical, intellectual, and emotional changes that naturally occur within individuals over a period of time.
This is likely to happen when the participants are children or adolescents who are undergoing physical & emotional developmental changes
This threat is especially of concern in longitudinal designs.
Testing
refers to the situation in which a pretest is given and scores on the posttest improve just because subjects have prior experience with the test
This effect exists even when no treatment is administered
This is especially true when the time interval is short, and items can be easily recalled
Instrumentation
Occurs when changes on the dependent variable between pretest & posttest are due to a lack of consistency in the instrument being used to measure the dependent variable.
This occurs when using unreliable measures
Using non-equivalent forms of a test
Form-B is more difficult than Form-A
Using untrained raters
Different raters interpreting the measure differently at each time
The threat of Statistical Regression
regression is most likely to be present when subjects are selected for a study based on extreme scores (very high / very low) on some measure.
When a test that is not perfectly reliable is administered two times, a test taker's score on the second administration is likely to be closer to the mean of the group than the score on the first administration
Regression to the Mean
Differential Selection of Subjects
Occurs
when a researcher is not able to randomly assign subjects to groups
In this case, participants in the experimental group are somehow systematically different from the people in the control group (Sampling bias)
For Example: In the exercise study, if people who regularly go to the gym are selected as the experimental group....
Mortality (or attrition)
refers to the concern that subjects who drop out of a study may systematically differ from subjects who remain
This is also of concern in longitudinal studies where researchers can lose track of participants over time.
Interaction
Differential selection of participants may interact with another of the threats to internal validity (typically maturation, history, or testing) to cause another concern.
This occurs when using groups that are already formed.
One group could have a systematic advantage (or disadvantage) over the other(s).
Maturing at different rates
Different histories
External validity
is concerned with how well results can be generalized to populations and settings beyond those used in the study.
If the results of the exercise experiment showed that exercising was an effective means to weight loss, can these results be generalized to a larger population?
Population validity
refers to generalizations related to other groups of people
Do the results generalize for males as well as females?
Ecological validity
refers to generalizations related to other settings, times, contexts, etc.
Does the behavior modification technique work at school as well as at
home?
Common Threats to External Validity
Pretest Sensitization
Sequencing Effects
Selection-Treatment Interaction
Specificity of Variables & Treatment Diffusion
Participant Effects
Experimenter Effects
Pretest Sensitization
In some situations, the pretest may "sensitize" participants to the treatment.
In these cases it could make participants respond more strongly than they would if they had not been pre-tested.
It is most likely to occur when the dependent variable is related to emotion or attitude.
It makes the effect of treatment appear more different than it actually is.
It affects generalizability to populations that are not pre-tested
Carryover Effect
occurs when the effects
from a prior treatment make it difficult to assess the effectiveness of the new treatment.
E.g. many reform initiatives have been implemented in the last 10 years in the public schools. Because these initiatives remain in place until another one takes its place, it is difficult to determine which, if any, program is improving student achievement.
Order Effect
occurs as a result of the order in which the
treatment(s) is given.
If you want to examine the effects of caffeine on learning, you might give the participants caffeine on the first day and observe the learning. The second day you give the placebo and observe the learning. The second day the participants are more familiar with the study which could effect the outcome.
Selection-Treatment interaction
occurs when participants are not randomly assigned to
treatment groups
Non-representative groups are used and so treatment effects apply only to that group and not to the population in general
This threat can also occur when participants are randomly selected
For instance, suppose you have a list of ten randomly selected schools for a new science initiative. Nine schools on the list turn you down, but the tenth school accepts. It is possible that this tenth school is fundamentally different in some way from the other nine on your list.
Specificity of Variables
Occurs when variables (independent and/or dependent) are not operationalized sufficiently.
Treatment Diffusion
Participants from different treatment groups communicate and share aspects of the treatment with each other
Participant Effects
A researcher often creates an artificial environment to control for extraneous variables, this can affect the way participants react to the treatment.
Participants know they are involved in an experiment and may react differentially
Hawthorne effect
- participants react positively to the 'special' attention given to them and perform better than would be expected of the general population.
John Henry effect
members of the control group feel threatened or challenged by treatment group and try to outperform them.
Novelty effect
participants react with heightened interest or motivation because the treatment is something new and different. This effect tends to wane as the experiment continues.
Placebo effect
Using a 'sugar pill' so it appears that all groups are being treated the same way
Experimenter Bias Effects
When the researcher knows which participants are assigned to which group
Previous knowledge of the participants might also affect
experimenter behavior or attitude
For example, if a teacher knows that Mary has a particularly difficult home environment, this may unconsciously affect the way the researcher evaluates or rates her on the dependent variable
Researcher Expectations
For example, the researcher developed a particular instructional strategy and has a strongly held belief that it to be superior to other strategies before experiment begins.
Experimenter personal-attributes
The characteristics of the experimenter affect participant reaction to treatment
Age, gender, race, anxiety level
For example, studies have shown that participants react differently to the same questions depending on the gender and/or race of the interviewer.
Confounding
refers to the intertwining of the effects of the independent variable and the extraneous variables
There are several ways to control extraneous variables in the context of experimental designs
Randomization
Matching
Homogeneous sub grouping
Using participants as their own controls
Some of these methods are used to control for extraneous variables in non-experimental designs as well.
Randomization
if used, there is no reason to believe that the groups differ systematically on any extraneous variable.
Random Selection (external validity)-
selecting participants at random from the population - random sampling
Random Assignment (internal validity)
assigning participants to the treatment groups randomly
Matching
Each participant is matched or paired with another participant who is the same (or similar) on the variable to be controlled
Once a pair is identified, one is randomly assigned to one treatment group and the other to the other treatment group
Alternately, you can rank all members based on control variable and then use two highest as a pair regardless of match similarity and so on.
Limitations of Matching
Participants with no match have to be excluded from the study
Very difficult to match participants on more than two extraneous variables
May not be possible to find a match for participants with "extreme scores"
Homogeneous Subgroups
Does not involve one-to-one matching
The goal is to make groups the same with respect to representation of categories or ranges of the extraneous variable
After forming the subgroups, randomly assign half of the participants to one treatment group and the other half to the other treatment group
Limitations of homogeneous sub grouping
It restricts generalizability
It is not really different from stratified random sampling
You can simply design the study to account for the variable.
Using the Subjects as Controls
For this technique, use a single group of participants and expose them to different treatments one at a time
The same participants get all of the treatments.
However, you must guard against carryover effects and order effects.
To do this, divide the group into two smaller groups and give the treatments, but in different order.
This is called counterbalancing.
Threats to internal validity
Internal validity is always a concern with single-subject research
Two major threats
Instrumentation
Specificity of variables
Controlling threats
Baselines are multiple measures of pretest
performance so that stability of the behavior of interest is established.
By repeating baseline measures over a period of time specific threats to internal validity (history and maturation) can be controlled.
Threats to external validity
Lack of external validity is the major concern with single-subject designs
Generalizability is addressed through multiple replications of the same treatment and design
that produce similar results for a number of different participants.
Replication
is, therefore, an important aspect of single-subject research
The more one's results are replicated the more confidence one has in the procedures that produced the results
Ch. 16 Other Descriptive Research Methods
13 terms
reandel
Psych B5 Ch 2 quiz
10 terms
amdelarosa91
Connect Psych: Chapter 3 Quiz
27 terms
lolmichelle
Research Methods Test #6
13 terms
mmonto62
Sets found in the same folderInvestigating the Social World Chapter Three, Ethi…
31 terms
Ed_S8Plus
Sampling
25 terms
Ed_S8Plus
Investigating the Social World Chapter One/ The na…
36 terms
Ed_S8Plus
Nonexperimental and experimental research part one…
75 terms
Ed_S8Plus
Other sets by this creator
Savicka's career construction, Linda Gottfredson's…
220 terms
Ed_S8Plus
Linda Gottfredson's Developmental Theory, Erik Eri…
210 terms
Ed_S8Plus
Erik Erickson stages of development, Trait & Facto…
198 terms
Ed_S8Plus
Trait & Factor Theory, Chapter 6: Myers-Briggs Typ…
182 terms
Ed_S8Plus
Verified questions
question
In most cases, in order to recover damages for emotional harm, a person also must show physical harm. Why would society want to establish an additional barrier to someone who has been harmed by the commission of a tort?
Verified answer
question
Examine a foreign culture of your choice, and propose a marketing plan for a brand of *cosmetics* made in your country.
Verified answer
algebra
Subtract. $$ \$7.19 - \$5.54 $$
Verified answer
psychology
If the recurrence of a behavior lessens because performing the behavior no longer leads to something, the process involved is (extinction/negative punishment). If the frequency of a behavior decreases because performing the behavior leads to the removal of something, the process involved is (extinction/ negative punishment).
Verified answer
Recommended textbook solutionsSocial Psychology
10th EditionElliot Aronson, Robin M. Akert, Samuel R. Sommers, Timothy D. Wilson
525 solutions
Social Psychology
10th EditionElliot Aronson, Robin M. Akert, Timothy D. Wilson
525 solutions
HDEV5
6th EditionSpencer A. Rathus
380 solutions
Myers' Psychology for AP
2nd EditionDavid G Myers
901 solutions
Other Quizlet setsMi Amiche Breakfast Menu
45 terms
hannah_fernandez9
rates of reaction
16 terms
laetitia_j05
Physics 2 Tests
22 terms
barnesa2018
Frankenstein
62 terms
ellen1568