Research Methods Exam 3!! <3

Unlike two-group designs, multi-group designs can test for ____ relationships between the independent variable and dependent variables.

Linear

Multi-group designs can allow you to test and control for the effect of multiple potential ____.

Confounders

Unlike two-group designs, independent variables in multi-group designs can be represented at the ____ level of measurement.

Interval

When researchers ____ levels of the independent variable, they intentionally select what level each condition will be exposed to.

Assign

A Bonferonni correction accounts for multiple comparisons by dividing your significance threshold by:

The number of comparisons you're making

One reason that multi-group designs are efficient at testing two-group hypotheses is:

You can sample fewer people

One way to avoid multiple comparisons is to use a(n) ____ test, like ANOVA or chi-squared.

Omnibus

Experimental designs involving three or more conditions are considered ____ designs

Multi-Group

In a multi-group design, when your outcome is measured at the nominal level, you can analyze your data using a(n):

Chi-Squared Test

When making multiple comparisons involving the same condition, we need to ____.

Correct for those comparisons

Multi-Group Designs

Experimental Research involving two or more conditions

This is a condition exposed to no manipulation

A "true" control condition

This control is exposed to some "neutral" manipulation

A "conceptual" control

This design involves manipulating a single IV across multiple levels

A "true" multi-group design

The most common way to create manipulations for multiple levels of an IV is to try and ______ each condition to a level

Assign

One way to better assess the relationship between our IV and DV is to randomly ______ levels to expose participants

Sample

Because p is never 0, there's always a chance that we're wrong when we reject the null hypothesis. This is called

Error of inference

When we reject a "true" null hypothesis, we have committed a

Type I Error

If we fail to reject a null that is actually false, we've instead committed a

Type II Error

Power

Our ability to detect a true effect, if it exists

Power is based on 3 factors:

The size of the "true" effect, the size of our sample, and our significance threshold

These can be made in two-group designs as well, but they're essentially mandatory in multi-group designs

Multiple comparisons

Pairwise comparisons

Comparisons involving only two groups

Familywise error rate

The probability of making a type I error in a given group, or family, of tests

False discovery rate

The proportion of significant findings across our study that are a type I error

Bonferonni correction

Divide your significance threshold by the number of tests in a family you're making

False discovery

A significant result that leads to rejecting a true null

We can correct for our false discovery rate through the

Benjamini-Hochberg correction

Omnibus test

A test that detects the presence of at least one difference between all groups tested

Analysis of Variance (ANOVA)

A test that compares the variance between conditions to the variance within conditions

A ________ can be used to see if the number of people falling in each category of your DV differs across levels of your IV

Chi-Squared Test (of independence)

Degrees of freedom

The amount of unique information present in a particular set of data, with regards to a particular test

We specify before our initial test (a priori) ) which contrasts we plan to make

Planned contrasts

Post-hoc tests

Tests made "after the fact" of your initial omnibus test

The two most common post-hoc tests

Fisher's Least Significant Difference (LSD)

Tukey's Honestly Significant Difference (HSD)

Within-subjects designs

Designs that involve measuring participants on a dependent variable multiple times

Pretest-posttest designs

Involves measuring some outcome before and after an intervention/manipulation

Repeated measures designs

Involve exposing participants to each level of the independent variable

One major issue with within-subjects designs is accounting for various ___ effects

Learning effects

Being exposed to research materials (eg. surveys) can change responses to those materials. This is called the

Testing Effect

Maturation effects

Occur when participants change over time for reasons other than the manipulation

Attrition effects

Occur when participants leave the study and your inference is affected as a result

The issues in which the order that participants are exposed to materials are called

Order effects

Practice effects

When participants change their behavior/responses due to familiarity with your measures

Fatigue effects

When participants get tired or bored over the course of a study, which introduces error into measurement

Carryover effects

When earlier manipulations affect responses to/engagement with later manipulations

Sensitization effects

When exposure to study materials at all lead participants to try and "guess" the hypotheses of the research, which can affect behavior

Oversampling

Collecting more participants than necessary for your target level of power, often by a specified percentage

Counterbalancing

Presenting your measures and/or manipulations in all possible orders to different participants

This design allows us to control for and analyze effects based on position, rather than the full order of methods

Latin Square Designs

These t-tests are used in two-group designs

Independent samples t-test

These t-tests are used for dependent data

Paired-samples t-test

If you have scores at multiple time points, or multiple manipulations, you can use a

Repeated measures ANOVA

Factorial design

A design that tests the effects of more than one independent variable, typically simultaneously

Up until this point, we've primarily been concerned with the effects of singular IVs on our outcomes of interest. These are called

Main Effects

Interaction

When the effect of one IV depends on the level of one or more other IVs

Interactions that involve at least two IVs

Two-way Interactions

Interactions represent the ___ effect of one IV on the outcome, given some level of the other IV(s)

Conditional/Dependent

Crossed design

A design wherein each level of your independent variables is paired with each level of every other independent variable

The "standard" ANOVA is a

one-way ANOVA

ANOVA with two IVs is often called a

two-way ANOVA

ANOVAs that let you test for main effects and interactions simultaneously

Factorial ANOVAs

Multiple regression

An analysis that assesses how well a set of variables predicts some outcome

One way to try and combat order effects is to _____ the order of your measures, wherein different participants are exposed to all possible orders of those measures.

Counterbalance

When people drop out from a study due to your manipulation, your results may become biased due to:

Attrition effects

Within-subjects designs allow you to control for individual differences because:

Participants serve as their own control group

One major advantage of within-subjects designs is that they are:

More powerful

When participants' responses to measures change as a consequence of having already been exposed to those measures, your results are affected by:

Carryover effects

Designs wherein you measure participants on some outcome before and after a manipulation are called:

Pretest-posttest designs

If participants give random responses to later measures in your study because they are tired or bored, your results will be unreliable due to:

Fatigue effects

When analyzing pretest-posttest data, the appropriate test that accounts for the dependence of your data is the:

Paired-samples t-test

An efficient method for controlling for order effects is the _____, wherein each measure or manipulation is presented at each possible position for different participants.

Latin square design

Research designs that involve measuring participants on a dependent variable multiple times are called:

Within-subjects designs

Mixed designs

Designs that incorporate both between- and within- subjects factors/designs

Between-subjects factors

Factors involving differences between people/groups of people

Within-subjects factors

Are factors involving differences within people at different points in time

Mixed models

Statistical analyses that can model and account for both dependent and independent data

Mixed-design ANOVA

An ANOVA that assesses main effects of between- and within- subjects factors, along with their interactions

Generalized linear mixed models

Regression analyses that can test both independent and dependent data