# Research Methods Ch 8& 9 Darren Farr
question

What do Association claim describe?

The relationship found between two measured variables. It does this without using strong, causal verbs!
question

Bivariate correlation

an association that involves exactly two variables. The study may be done with more than two variables being measured, however this type of correlation will only look at two variables at a time
question

What are the three main types of association?

positive (both increase or both decrease together), negative (while one increases, the other decreases), and zero (no relationship shown)
question

How does one investigate association?

researchers need to measure the first variable and then measure the second variable–in the same group of people.
question

What do data tables of association claim studies show?

Each row shows one person’s scores on two measured variables.
question

What do researchers do after collecting data to test an association claim?

They use scatter plots and the correlation coefficient r to describe the relationship between the two measured variables
question

What are the two qualities of the correlation coefficient r

1. direction (positive, negative, zero) 2. strength (how close the value is to 1 or -1)
question

approximate strength of association (based on r)

.10 small/weak relationship .30 medium/moderate relationship .50 large/strong relationship
question

Categorical variable

the levels are categories and the values fall in either one category or another.
question

Quantitative variable

the levels are coded with meaningful numbers
question

How should one graph the data of an association claim study that has a categorical variable?

Researchers most often plot the results of a categorical variable as a bar graph. In doing so, the graph shows the mean value of the participants results to the quantitative variable for each level of the categorical variable. -You would use this information to examine the difference between the group averages to see whether there is an association.
question

What test do researchers use to analyze the significance of the difference between means (group averages) when at least one of the variables in an association claim is categorical

researchers occasionally use r, but it is more common to use a *t test* to look at the differences between means when one variable is categorical. They may also use other statistical tests
question

What is an association claim supported by?

it is supported by a particular kind of statistic or a particular kind of graph; it is supported by a study design in which both of the variables are measured
question

What are the two most important validities to interrogating an association claim?

Construct validity and statistical validity. You can also ask about external validity, but internal validity is not relevant
question

What do you ask to assess construct validity?

-How well was each variable measured? -Ask about the operationalization of (the way researchers measured) each variable -Does the measure for each variable have good reliability? -Is it measuring what it is intended to measure? -What is the evidence for its face validity (does it look like a good measure), its concurrent validity, its discriminant and convergent validity (making sure it doesn’t measure something its not supposed to)
question

What do you ask to assess statistic validity?

-How well do the data support the conclusion? -Check there were no outside factors that might have affected the scatterplot, r, bar graph ,or difference scores. Check for: 1) Effect size–larger effect sizes give more accurate predictions and are usually more important. Errors of predictions get larger when associations get weaker 2)Is the correlation statistically significant? Statistical significance calculations help researchers evaluate the probability that the result came from a population in which the association is really zero. (p value) 3) could outliers be affecting the association? 4) is there restriction of range? 5) is the association curvilinear
question

Effect size

Describes the strength of an association. -positive and negative associations can allow us to predict one variable from another, and the stronger the effect size, the more accurate, on average, our predictions will be.
question

Statistical significance

the conclusion a researcher reaches regarding how likely it is they’d get a correlation of that size just by chance, assuming that there’s no correlation in the real world. -depends on effect size and sample size
question

p value

statistical significance calculations provide a probability estimate (p) which helps researchers evaluate the probability that the sample’s association came from a population in which the association is zero. -indicated with a italicized p, \”sig\” or \” * \”
question

Nonsignificant result

If the probability of getting some correlation just by chance is relatively high( > p=.05), the result is considered in this way.
question

outlier

-an extreme score -a single case (or sometimes a few) that stands out far away from a pack. -can have a strong effect on the correlation coefficent r; can have a large impact on the direction or strength of the correlation -matters most in smaller samples
question

restriction of range.

if there is not a full range of scores on one of the variables in the association, it can make the correlation appear smaller than it really is. -When this is a suspected issue, one can use a statistical technique, correction for restriction of range, which estimates the full set of scores based on what we know about an existing, restricted set, and then recomputes the correlation. -can apply when one variable has very little variance. -is usually asked about primarily when the correlation is weaker than expected.
question

Curvilinear association

When the relationship between two variables is not a straight line (may be positive up to a point, and then negative) -r doesn’t describe the patter well because it is intended to show the slope of a straight line. -the statistically valid way to analyze this is to compute the correlation between one variable and the square of the other.
question

What do you ask to assess internal validity

Can we make a causal inference from an association? -assess this to \”guard against the causal temptation\” -remember the three causal criteria because \”Correlation is not causation. -it is not necessary to focus on internal validity as long as it is just an association claim
question

three criteria for causation

1. Covariance of cause and effect–must be an association between the cause and effect variable 2. Temporal precedence (a.k.a *directionality problem*)- causal variable must come first in time before the effect variable . 3. Internal validity (a.k.a. *third-variable problem*)- no confounds/alternative explanations
question

spurious association

when the association between two variables is only there because of some third variable, the original association is known as this.
question

What do you ask to assess external validity?

To whom can the association be generalized? (other people, places, and times) -Ask who the participants were and how they were selected. random selection supports stronger. -just because a bivariate correlation study may not have used a random sample doesn’t mean you should automatically reject the association for that reason. look how it stands up across other validities. -moderators can inform external validity–it allows us to see how the association may not generalize from one situation to the next.
question

moderator (variable)

when the relationship between two variables changes on the level of this other variable. -EX: \”Gender moderates the association between extroversion and group conversations
question

Multivariate design

correlational study designs. involving more than two measured variables.
question

Multivariate design Correlational techniques used to get closer to making causal claims

The variables are still measured, not manipulated 1. Longitudinal design 2. multiple-regression analyses 3. Pattern and parsimony approach
question

Longitudinal design

-measures the same variables in the same people at several points in time which allows researchers to evaluate temporal precedence in their data -the researchers measure the same variables in the same group of people across time -gives several individual correlations: cross-sectional correlations, autocorelations, and cross-lag correlations -can provide some evidence for a causal relationship by means of the 3 criteria for causation (but they don’t always help rule out third variables/assess internal validity).
question

Cross-sectional correlations

correlations that test to see whether two variables, measured at the same point in time, are correlated -Examples of such a correlation: variable A vs variable B at time 1. Also variable A vs Variable B at time 2.
question

Autocorrelations

evaluation of the associations of each variable with itself, measured on two different occasions. Example: variable A at time 1 vs variable A at time 2
question

Cross-lag correlations

show whether the earlier measure of one variable is associated with the later measure of the other variable -Most important in multivariate studies measuring across time. -address the directionality and help establish temporal precedence. -allows us to investigate how people change over time. -can have three possible patterns: (1) where results show A leads to B over time, (2) results show B leads to A over time, and (3) may show both correlations are significcant such that A predicts B across time.
question

Multiple-regression analyses

-assess the two variables in question, and then measure several other variables that could be potential confounds to look at their effect on the correlation in question. -by conducting a multivariate design, researchers can evaluate whether a relationship between two key variables still holds when they control for another variable. (\”controling for\” is similar to identifying subgroups. So we ask if the relationship still holds up when we analyze the relationship at all levels of the variable we are \”controling for\”) -the statistical technique of multiple regresson can tell us what scatterplot best describes the relationship between our two key variables. -help researchers rule out certain third variable explanations/address internal validity -doesn’t necessarily account for temporal precedence
question

What is the statistical technique of regression used for when looking at a multiple-regression analyses?

this is used to test whether some key relationship holds true even when a suspected third variable is statistically controlled for.
question

criterion variable

the dependent variable the researchers are most interested in understanding or predicting (in a multiple-regression analyses)
question

Predictor variables

independent variables, the variables measured in a regression analyses that are not the criterion variable.
question

Beta/even standardized beta

used when comparing the criterion and predictor variables of a regression analysis. Beta is similar to r in that is direction is indicated by sign (+/-) and the larger it is the stronger it is (and smaller is weaker). Beta’s are relative to each other in the chart -computed from predictor variables that have been changed to standardized units.
question

Interpreting beta

-The beta that is associated with a predictor variable represents the relationship between that predictor variable and the criterion variable, when the other predictor variables in the table are controlled for. -As far as statistical significance, p values will indicate if a beta is statistically significantly different from zero
question

coefficent b

sometimes called the standardized coefficent. similar to beta in that the sign(+/-) indicates the direction but the values can not be compared to each other within the same table. That is because the values are computed from the original measurements of the predictor variables.
question

What are the benefits of looking at more predictors to a regression analysis?

it helps answer two kinds of questions: (1) it helps control for several third variables at once and (2) by looking at the betas for all the other predictor variables, we can get a sense of which factors most strongly affect the criterion variable.
question

\”telltale\” phrases for regression analysis designed experiments

– \”Controlled for\” …\”a number of variables\” or \”factors, such as\” – \”Taking into Account\” …\”after ‘ ‘ these variables, ___ found that…\” -\”Correcting for\” or \”Adjusting for\” Ex: \”after correcting for factors that may effect scores\” Ex: \”the researchers adjusted their results for a number of variables\”
question

\”pattern and parsimony\” approach

there is a pattern of results that is best explained by a parsimonious causal explanation -process exemplifies the theory-data cycle -cigarette ex: because all of the diverse predictions are tied back to one central principle, there is a strong case for parsimony. Also, the diversity of the findings makes it harder to raise third-variable explanations. -the results of a variety of correlation studies all support a single causal theory.
question

parsimony

the simplest explanation of a pattern of data.
question

mediator/mediating variable

a variable that helps explain the relationship between two other variables. -a variable that may be proposed as a potential reason for the association between the two key variables.
question

Mediators vs third variables

-third variables are often a nuisance to researchers and are external to the two variables in the original bivariate correlation. -Mediators are not a nuisance but often of direct interest to the researchers and are internal to the causal variable. researchers propose mediators when they are interested in isolating which aspect of the causal variable is responsible for the relationship. -BOTH involve multivariate research designs and can be detected with the same statistical tool (multiple regression)
question

Mediators vs. Moderators