Term
| Internal Validity Confound |
|
Definition
| a Nuisance Variable associated with IV & DV (3rd Variable associated with IV and DV, does NOT explains why IV causes DV) |
|
|
Term
|
Definition
| The extent at which a causal conclusion based on a study is warranted.(IV Caused DV, Not internal confounds) |
|
|
Term
| IV (Independent Variable) |
|
Definition
| Cause, what you might manipulate |
|
|
Term
|
Definition
| Effect, what ever it is you are trying to measure |
|
|
Term
|
Definition
| meaning dichotomous, did or didnt, yes or no. A Binary IV. |
|
|
Term
|
Definition
| 3rd variable that changes the strength or direction of the relationship between two variables. (In other words, moderators interact with the IV in Predicting the DV (An interaction effect) |
|
|
Term
|
Definition
| The extent to which we have evidence that the precise causal connection between the IV and DV is consistent with our hypothesis. |
|
|
Term
|
Definition
| 3rd variable associated w/ or causes both IV and DV, DOES explain why IV causes DV and is consistent with your specific causal |
|
|
Term
| Relationship between continuous variables is called? |
|
Definition
| A Correlation or Association |
|
|
Term
| Relationship between categorical (binary)variables is called? |
|
Definition
| Difference between groups in the Proportion or Rate |
|
|
Term
| Relationship between a Categorical (binary) variable and a Continuous variable is called? |
|
Definition
| a Difference between groups in LEVELS of the continuous variable (Ex: one Group has higher scores [Continuous] than the other group) |
|
|
Term
| Examples of Internal Confounds? |
|
Definition
| time and maturation confounds/ testing/ instrumentation |
|
|
Term
| ________ validity is strength of the evidence regarding causation |
|
Definition
|
|
Term
| Can Correlation studies have internal validity? |
|
Definition
| YES! thye are low in internal validity but they still can have some. |
|
|
Term
| Two types of internal validity confounds |
|
Definition
| person variables ( age, gender, …) and experience (things that happen to people other than IV (manipulation) during the course of assessment) |
|
|
Term
| Construct Validity Confound |
|
Definition
| 3rd variable, if it is not consistent with your specific causal hypotheses |
|
|
Term
|
Definition
| Extent to which we can generalize study’s findings to people, settings, times, measures and characteristics other than those in one particular study |
|
|
Term
| Participant vs context confounds- |
|
Definition
| differences in the types people vs. differences in their experiences |
|
|
Term
|
Definition
| Form of Bias in a study which has to dowith "Interpersonal Expectancy: The influence of the experimenter’s beliefs and desires about the results on how the subject performs. |
|
|
Term
|
Definition
| Randomized Control Trials to get high internal validity |
|
|
Term
|
Definition
| When you compare 2 naturally formed groups (like Dx, race, age) and you can manipulate it so you match people based on those case characteristics (the IV) to reduce the chances of that being a confound in your study. |
|
|
Term
| Quasi-experimental Design |
|
Definition
| When you are creating the IV, such as doing an intervention, but you do not randomize the groups.; groups are not equal. |
|
|
Term
| Construct Validity Variable/Confound |
|
Definition
| An intrinsic part of the IV, part of that construct, so therefore its a construct validity variable |
|
|
Term
| In a correlation study, with sexual abuse as the IV and violence as the DV socio-economic status could be... |
|
Definition
| an internal validity confound |
|
|
Term
| Correlation studies have... |
|
Definition
| strong internal validity, sometimes! |
|
|
Term
| Conceptually, a moderator is...... a mediator |
|
Definition
| different than (But not the opposite of) |
|
|
Term
| Conceptually, a moderator is...... an internal validity confound. |
|
Definition
| different than (but not the opposite of) |
|
|
Term
| Shared Variance between the promary IV and a 3rd variable (in predicting the DV) indicates the 3rd variable could be a... |
|
Definition
|
|
Term
| The graph where the line cross (at some point) indicate? |
|
Definition
| there could be a moderator (an interaction!) |
|
|
Term
| A study has high internal validity to the extent that ... |
|
Definition
| there are no significant moderator variables |
|
|
Term
| What methodological issues most directly address the reasons why an intervention leads to particular changes? |
|
Definition
|
|
Term
| What conclusions can be made from a demonstration of moderation effects in a tx study? |
|
Definition
| the Tx works differently for different SUBGROUPS of patients |
|
|
Term
| Which type of study usually produces the highers internal validity? |
|
Definition
| randomized experiments (Correlation/observational studies have better External Validity) |
|
|
Term
| Which type of study usually produces the highest external validity? |
|
Definition
| correlational /observational studies (randomized have better Internal Validity) |
|
|
Term
| In a randomized treatment study, it is possible that self-esteem can be... |
|
Definition
| a mediator or a moderator, depending on the other variables in the study |
|
|
Term
| In a clinical psychology correlational study it is possible that age can be... |
|
Definition
| a moderator. (Age or gender are common moderators) |
|
|
Term
| what are the two primary ways to control for confounds by "holding it constant" in a single-group correlational/observational study? |
|
Definition
1. Covariance - analyze covariance by controlling for it statistically.
2. Selecting samples such as only having "females" or "young people" |
|
|
Term
| What conclusions CAN be made from an experiment with high internal validity (when differences between conditions are found)? |
|
Definition
| The manipulation of the IV really did cause change (Lead to the DV) |
|
|
Term
| What conclusions cannot necessarily be made from an experiment with high internal validity (when differences between conditions are found)? |
|
Definition
-You do not know External Validity.
-We do not know WHAT about the IV was helpful
-You would not know reasons WHY a Tx worked |
|
|
Term
| What are the main threats to Internal Validity of RCT's or other true experiments? |
|
Definition
1) Subjects may differ (selection bias) at different levels of the IV
*Baseline levels of DV
*Demographics
*Differential Drop outs
2)Subject's experiences in study differ
*Demand/expectancies (Control group is demoralized or competitive and Experimental group is more hopeful)
*Amount of Tx received |
|
|
Term
|
Definition
1) Person Moderator (Age, Gender, Ethnicity)
2) Context Moderator (Weather, University/lab setting rather than real world setting) |
|
|
Term
| How can you clearly describe the influence of a confound? |
|
Definition
| C could explain the association between X and Y bc ppl who are high in X and Y are also LIKELY to be high in C. |
|
|
Term
| When a Moderator is NOT part of the hypothesis, it is a |
|
Definition
| Construct Validity Confound |
|
|
Term
| What methods allow a researcher to prevent or reduce the impact of confounds? |
|
Definition
**Make it constant (BEST SOLUTION)increase the homogeneity of participants using inclusive/exclusive criteria)
*include it as a covariate (ex: Gender)
*Random assignment
*Standardize study procedures (automating instructions/prevent
*exclude it (exclude men)
*match non-randomized grps on gender |
|
|
Term
| What conclusions can be made based on RTC (high in internal valididty) showing differences in change between two treatments? |
|
Definition
*whether or not one Tx is better than the other
*Can NOT tell that the Tx worked. |
|
|
Term
| What conclusions can be made based on an RTC (high in internal validity) showing dirrences in change between Tx and no Tx conditions? |
|
Definition
*that the Tx worked (or is efficacious)
*Can NOT tell if WHY and we do not know if it is more efficacious than other Tx's (bc we only compared it to no Tx) |
|
|
Term
| What are the ways that subject attrition can compromise interval validity? |
|
Definition
1) Differential drip out rates
2) Differential Pattern for drop out /reasons |
|
|
Term
| if you did not find significant differences between your two conditions on any pre-treatment measure could there be selection bias? if so Why and how you could you find out? |
|
Definition
Yes.
why? Lack of Power/small sample
*You might not have measured confound variables that really did differ between grps. |
|
|
Term
| In a RTC with 100 subjects in each of your conditions and there were no significant differences on any pre-treatment measure, could there be a selection bias? |
|
Definition
| Yes, there could be a selection bias here. |
|
|
Term
| Is it necessary to check pre-treatment differences if you could guarantee that your randomization procedure was truly random? Why or Why not? |
|
Definition
| Yes, it is necessary because you may not have measured other confounds which might be irreverent in the 2 conditions. Even if you can guarantee randomization you still want to check for pre-treatment confounds (Like severity of depression) |
|
|
Term
| Random Sampling vs. Random Assignment |
|
Definition
Random Sampling: has to do with attempting to increase external validity (generalizability)
Random Assignments: has to with attempting to increase Internal Validity; creating groups of people on a random basis to the IV (Like Tx groups) |
|
|
Term
| Conceptually a Moderator is _________ to the IV and DV. |
|
Definition
unrelated.
(almost anything can be a moderator. They Change the magnitude or direction of the relationship between two other variables) |
|
|
Term
| A Mediator is a 3rd variable that can _______ why the IC causes the DV. |
|
Definition
| explain. A Mediator is NOT part of the IV but it effects how the IV relates to the DV. |
|
|
Term
|
Definition
| tells you the direction of the causal pathway |
|
|
Term
|
Definition
| tells you whether or not your variable is a mediator. |
|
|
Term
| Moderation vs. Generalizability |
|
Definition
they are opposites.
Generalizability is:
*Breadth (Mod. is specific)
*High External Validity (Mod. Low Ext Validity)
*Low Internal Validity (Mod. High Internal Validity) |
|
|
Term
| threats/confounds to internal validity: factors or influences other than the IV that could explain the results |
|
Definition
| History, maturation, attrition, placebo/expectancy effect |
|
|
Term
| threats to external validity |
|
Definition
| characteristics of the experiment that may limit the generalizability of the results |
|
|
Term
| Moderation and ________ validity are opposites. |
|
Definition
|
|
Term
| If we have strong confidence that the IV caused the DV this means we have strong ________ validity? |
|
Definition
| internal (IV caused the DV) |
|
|
Term
| If we have minimal evidence about WHY the IV caused the DV we say that the study has poor _________ validity, and research probably needs to study more ___________ variables. |
|
Definition
|
|
Term
| If we have evidence that the causal relationship between the IV and the D V only occurs on a very limited basis(specific ppl or settings) then the study has poor ________ validity and research has probably found many __________ variables. |
|
Definition
|
|
Term
| Shared variance can be different things including: |
|
Definition
| confounds= construct validity, mediators |
|
|
Term
| How do you cause an interaction? |
|
Definition
| force it by multiplication then add it in as a predictor |
|
|
Term
| Moderation and ________ validity are opposites. |
|
Definition
|
|
Term
| Conceptually a moderator is ______ an internal validity confound. |
|
Definition
| different but not opposite |
|
|
Term
| Conceptually a moderator is _______ a mediator. |
|
Definition
| different but not opposite |
|
|
Term
| Decrease variability (increasing homogeneity) within groups improves _________ validity and statistical _____. |
|
Definition
|
|
Term
| T/f - Correlation studies have strong validity sometimes |
|
Definition
|
|
Term
| what are the primary ways to minimize bias in a two-group experiment to reduce the chance of bias-related internal validity confounds? |
|
Definition
1)Automation of protocols (scripting, etc)
2) Make the experiments naive (to eliminate bias) |
|
|
Term
| Primary ways to control for internal validity confounds in a randomized experiment? |
|
Definition
Holding constant or control by:
-Making them homogeneous
-Put in as a covariate (statistically)
-Stratify or Randomize |
|
|
Term
| Name a variable that could be a moderator but never a mediator |
|
Definition
|
|
Term
| When an experiment maximizes internal validity, there is often a loss of |
|
Definition
|
|
Term
| what conclusion can be made when you have a LACK of moderation effects? |
|
Definition
| Generalizability (of the effect of the IV on the DV) **remember: external validity/generalizability is opposite moderation |
|
|
Term
| what conclusion can be made from a study in which the age distribution in the study closely matches the age distribution in the local community? |
|
Definition
| generalizability cannot be known until more analysis are done |
|
|
Term
| what conclusion can be make from a study with one one ethnic group in the sample? |
|
Definition
| generalizability would not be known |
|
|
Term
| When an experiment maximizes construct validity, there is often a loss of... |
|
Definition
|
|
Term
|
Definition
-increase sample size
-increase group differences (effect size)
-within-subjects
-increase alpha (10%type I error)
-one-tailed tests
-be parsimonious
-decrease variability (standardize procedures, use scripts, homogeneous sample (narrow inclusions, use reliable measures, clean your data) |
|
|
