Chapter 12 - The Correlational Research Strategy

Define the goal of correlational research

To demonstrate the existence of a relationship between two or more variables 

Define the goal of experimental research

to demonstrate a cause-and-effect relationship between two variables

(correlation/causation) offers high external validity but cannot imply causality

correlation

in the line-of-best-fit (or regression line), the (closer/farther) the points are to the line, the greater the association between variables 

closer

what is the range of the correlation coefficient r and what do the extremes mean

r ranges from -1 to 1, -1 indicates a perfect negative, 0 indicates no correlation, and 1 indicates a perfect correlation

what are the two types of correlation coefficients r, when are they used, and what are their notations

1. spearman rho (r𝑠) - ordinal values

2. pearson r - ratio or interval values 

for both types of correlation coefficients, in what 3 characteristics are we interested

1. form (linear/nonlinear)

2. direction (-/+)

3. strength (absolute value between 0 and 1)

in a (linear/nonlinear) correlation, a change in one variable is not always consistent with a change in another

nonlinear

define a monotonic relationship

relationship between two variables when each of the two variables has values that continue in one direction or stay the same 

define a non-monotonic relationship

relationship between two variables where on variable (or both) can reverse direction

when should spearman’s rank-order correlation be used 

with at least 5 pairs of data (>8 preferred)

define the numerical values of correlation coefficients for the following degrees:

1. no relationship

2. weak relationship

3. moderate relationships

4. strong relationship

1. no relationship 0-0.10

2. weak relationship 0.10-0.30

3. moderate relationships 0.30-0.70

4. strong relationship 0.70-1.00

a spearman correlation of 1 will result when ______, even if the relationship is not linear

the two variables are monotonically related

when the data are roughly elliptically distributed, there is (variance/margin/deviation) on both factors and there are no outliers, the (pearson correlation is greater/spearman correlation is greater/the pearson and spearman correlation are equal)

1. variance

2. none, the spearman and pearson correlations are roughly (but not exactly) similar

this correlation coefficient offers no assumption on shape of distribution other than variables holding a monotonic relationship

spearman

this correlation coefficient is more sensitive to outliers

pearson

this correlation coefficient is appropriate for ordinal and non-normally distributed data

spearman

this correlation coefficient only features linear relationships

pearson

this correlation coefficient ranges from -1 to 1

spearman and pearson

this correlation coefficient will yield values close to 0 for nonmonotonic relationships 

spearman and pearson

why is one correlation less sensitive to outliers than another

the spearman correlation reassigns outliers to a rank and ranks cannot be outliers therefore the correlation coefficient is less affected by outliers 

for p< _____, the relationships is unlikely to be the result of chance

0.05

what is the calculation for the degrees of freedom of correlation

df = sample - 2 (numebr of variables)

to be significant, r must be ________

equal to or larger than the value corresponding to the df and p level

a small r can be statistically significant if _____

the sample size is big enough

True or false? with n=2, the correlation will always be r=0.5

false, r will always equal -1 or 1

correlation values r are (ordinal/interval/ratio)

ordinal, do not increase in equal increments

define shared variance in the context of a venn diagram

middle section covered by both variables representing the proportion of variability shared by them (x and y)

True or false? in correlation venn diagrams, the greater the degree of overlap, the greater the strength of the correlation

true

r² is the coefficient of ______

determination

what does the coefficient of determination r² measure

percentage of variability in one variable that is determined by its relationship with the other variable

True or false, a disadvantage of correlational methods is that they are not sensitive to outliers, falsifying results

False, they are very sensitive to outliers

what are the 3 types of time-point correlations

1. cross-sectional correlations

2. cross-lag correlation

3. autocorrelations

define the significance of the cross-sectional correlation

test of whether 2 variables measured at the same timepoint are related to each other

define the significance of the lag cross-correlations

tests whether a variable at an earlier timepoint is associated with another variable at a later timepoint 

what is the phenomenon of a variable at an earlier timepoint being associated with another variable at a later timepoint called

temporal precedence 

define the significance of the autocorrelations

tests whether a single variable at one timepoint is related to the same variable at another timepoint

a parent’s soothing reducing over time is an example of of a (cross-sectional correlation/cross-lag correlation/autocorrelation)

autocorrelation

if a ______ correlation is significant, then one variable has temporal precedence over the other 

lagged cross-correlation

if a ______ correlation is significant, then one variable covaries with the other

cross-sectional

if a _____ correlation is significant then on variable shows regular repeated change over time

autocorrelation

True or false? of the cross-sectional, lagged cross-sectional, and autocorrelations, the autocorrelation is the only one which can assert causality

false, neither the cross-sectional, lagged cross-section, not the autocorrelation can assert causality