Basic Principles Re: Methods, Data, & Statistical Results Flashcards
Dependent variables (DVs) are known as what?
The outcome variables- values that constitute results of the study
Why are they called dependent variables?
Variation in these variables follows from or depends on other factors
True or False: The DV is measured, but not directly controlled
True
Independent variables are referred to as what?
Variable being controlled by the experimenter
What is the purpose of the IV?
To determine if the IV leads to variation in the DV
What are the two ways an IV can be created?
- Manipulation; Changing levels of the IV
- Selection; choosing people of certain ages
Do all variables have to be divided into IV and DVs?
No; some studies can study the variations in one measure which relate to the variations in others
Variables can also be referred to as
Factors
The values that variables take are called
Levels
When studies do fit the IV and DV mold, what can we use to describe them?
Factor terminology
The Brownell et al. (2009) study contained two independent factors (adult verbalization, child age), each of which had two levels (silent, verbalization) (18-, 25-month-olds). What would this design be called?
2x2 Factorial design
In the Brownell et al. (2009) study, it was a 2x2 factorial design. how many groups/conditions did this study have? This reflects the IVs and levels involved within the study.
4 conditions
Within a study, what is more important, the main effect or interaction?
The interaction effect
What is a main effect?
The overall effect the IV/predictor variable has on the Dv/criterion
The 18 months value was .53 and the 25 months value was .56. Was there a main effect?
no
the adult silent value was .51 and the adult verbalization was .58. Was there a main effect?
No
true or false: A main effect for a variable w/ 2 levels is NOT interpreted the same way as a main effect for a variable w/ 3 levels (or more).
True
If there are only 2 groups, what can we know in terms of effects?
We know if they differed via a main effect through t-test
If there are 3 groups, can you tell which groups differed from each other?
No, can only tell some differences, but not which specific groups. You must run post-hoc comparison after ANOVA test
When can an interaction effect occur?
When there’s more than one causal/predictor variable
Interaction effect involves the effect of
One variable differing depending on the level of the second variable
A main effect tells us there’s a _______ while an interaction effect tells us theres
simple difference; difference in differences
Interaction is often short for
two-way interaction
Can an interaction be observed?
Must be statistically tested; as it is more visible and interpretable through figural rather than table, representations (i.e. bar graphs)
Interactions are readily dectectable by what?
Non-parallel nature of graphs (bars/lines) across conditions
Groups/conditions should be represented by ______ but is usually presented in a ____ bc it is easier to see
bar graphs, line graphs
Describe the interaction from the Brownell,Svetlova, and Nichols (2009) study.
The younger group showed a slight decrease in sharing when adults verbalized, while the older group showed more sharing when adults verbalized.
true or false: interactions are limited to only experimentally manipulated IVs
False; it can occur among all IVs/PVs (including non-manipulated ones)
Are interactions only limited to group designs?
No; can occur among non-manipulated PVs in a correlational design
What does variance refer to?
differences in scores on the DV
Variance quantifies what?
how spread out the scores of a sample are around their mean
When computing the variance of a set of scores, we start by
calculating how far each score is from the mean.
Total variance accounted for” or “primary variance
those differences that are accounted for by the IVs/predictors (i.e., explained by the IV/predictor variables)
We always say that we want our studies to include a diverse group of people so that we can generalize it to other samples and settings. What kind of validity would this be focusing on?
external validity
When a study mainly emphasizes the strength of an effect (i.e., effect size) and the probability (statistical significance) that the results could have been obtained by chance (e.g., there is not really an effect/association), what kind of validity is this?
Statistical validity
In a study testing teacher autonomy support and school readiness, the study claimed that teacher A-S had an effect on a child’s school readiness, when in reality, it did not. What type of statistical error would this be?
Type 1 error; false positive
In a study testing school readiness and temperament, the study found that temperament DID NOT have an effect on a child’s school readiness, when it actually did. What type of statistical error would this be?
Type 2 error; false negative
What are the effect sizes and what can they be described as?
.20 -> small
.50 -> medium
.80 -> large
When a study wants to address the extent to which one variable (X) has an effect on another variable (Y) rather than some other variable (C), what kind of validity would this be?
Internal validity
To move from the language of association to the language of causality, a study must satisfy several criteria:
a) establish the two variables are correlated; correlation cannot be 0 (EASIEST)
b) study must show the causal variable came first (i.e. temporal precedence); HARDEST
c) establish no other/alternative exist for the relationship (i.e., study has internal validity); HARDEST
What does internal validity first ask?
if the study was able to achieve temporal precedence
What does internal validity ask after achieving temporal precedence?
whether the study controlled for alternative explanations (e.g., confounding variables)
What are ways to control for confound variables?
1) RA participants to groups/conditions; between-subjects experiment
2) assigning all conditions to every participant; within-subjects experiment
3) correlational study & statistically control for alternative explanations/confounds by using multiple regression
If your plausible alternative explanations for findings have NOT been removed/controlled for, what can you not do?
Be certain the variability recorded in the DV is due to the variability in the IV
True or false: it is impossible to control for all validities at once
True
Explain what happens between internal validity and external validity when trying to achieve both.
When you try to prioritize and control the experiment more, you achieve more internal validity, but compromise external validity due to the controlled environment.
If your goal is to test a causal claim, what validity would you prioritize?
Internal validity
if your goal is show that your claim can be generalized to multiple populations, what validity would you strive for?
External validity
In most research, what should we prioritize and why?
Internal validity; because if we don’t have valid data, we cannot even generalize it
What are the 3 ways to control?
1) control over the iv/causal variable
2) control over setting
3) control over preexisting diff among participants
For the 3 controls; correlational studies _______ what?
experimental studies ______ what?
don’t have these controls
mess up these controls
In Kliegel et al. (2007), they presented the instructions and stimuli for the familiar and novel conditions in the same way for all participants. What kind of control would this be?
Control over the IV
What must be true of control over the IV/causal variable?
No unintended deviations
Critical elements must be the same
How do you achieve control over setting?
1) hold all the factors constant for participants (same time, noise, etc)
2) disperse variations in other factors randomly (different times, experimenters, materials)
T or F: possible to control a variable by making it the same for all participants.
False; alternative is to vary acoss groups
In Brownell et al. (2009) they used the same testing room for all participants. What kind of control would this be?
Control over setting
In Kliegel et al. (2009), they varied the time of testing randomly across groups. What kind of control did they achieve?
control over setting
How do you control over pre-existing differences among subjects?
1) random assignment (between groups)
2) matching variables
3) giving each participant every condition (within subjects)
In Kliegel et al. (2009), they randomly assigned half of the participants to the familiar condition and the other half to the novel condition. what kind of control is this?
Control over pre-existing differences among subjects
What are subject variables aka “classification” variables?
variables that are NONMANIPULATED; taken as they naturally are
How do you control for subject variables aka classification variables?
Through selecting people w/ that characteristics
Age, income/SES, gender, temperament, IQ are all examples of what kind of variable?
subject variables
How do we create “causal variables” for subject variables?
Through selection
For research w/ non-manipulated variables, what is the limit for construct validity?
a potential problem; we are unsure what the true construct is bc each variable is multiple factors
For research w/ non-manipulated variables, what is the limit for internal validity?
You have no control over the IV/PV; therefore you cannot assume equivalence
For research w/ non-manipulated variables, what is your solution to controlling for pre-existing differences among subjects?
1) control through selection of participants w/ only one level of characteristic (not really feasible)
2) measure for the confound and see if it correlates w/ your causal variable (IV/PV) and your DV
When is the C variable a confound?
ONLY if it systematically varies w/ your PV and your DV
True or False; A variable is a confound if it does not vary
False; only a confound if it varies
If the c variable does vary w/ your DV & causal variable, what do you do?
Control for it using multiple regression
Examining ONLY 5 year old toddlers in a study would be what kind of comparison?
Implicit age comparison
True or False; finding a “genuine change with age” means literal age produces the change
False; Only means that variables that are regularly & naturally associated w/ age produce the change
True or False; researchers study ALL the people they’re interested in.
False; test only samples
How can researchers ensure that a sample is representative of the population to which they wish to generalize?
1) define the scope of generalization (i.e. the population of interest)
2) decide how to select from the population
If all members of the population really are equally likely to be selected, what method is this?
random sampling
For a study of U.S. undergraduates, researchers draw 5% of each class year. What kind of sampling method is this?
Stratified sampling
For a study of high schoolers in which comparisons among ethnic groups are of interest, suppose that Asian Americans constitute 3% of the high school population in your city. Researchers choose to sample 6% of the population to get a larger subgroup. What kind of sampling method is this?
oversampling
Where do researchers tend to draw their samples from realistically?
Within their communities in which they live/work and may only sample a few.
Selection of samples primarily on the basis of availability or cooperation is referred to as
convenience sampling
Samples obtained via ___________ will not be perfectly representative of the broader population with respect to variables like race, ethnicity, region, education, social class, or even age.
convenience sampling
Although many samples depart from representation; what is not threatened?
external validity bc the deviations fr/ representation do not have any plausible effect on generalizability
Finding a representative pool of potential participants is a good starting point, but… the REAL question is how well the final sample reflects the initial pool?
1) initial pool solicited
2) % agreed & how they differ from initial pool
3) % retention of people & how they initial from initial pool
