CHAPTER 5 - ALTERNATIVES TO EXPERIMENTATION: CORRELATION & QUASI-EXPERIMENTAL
QUASI-EXPERIMENTAL
What does a scatter plot show?
➔ Quasi = “seeming like” ➔ graphic display of pairs of data points on the
➔ meaning: superficially resemble x and y axes.
experiments, but lack their required ➔ A scatter plot illustrates the linearity, sign,
manipulation of antecedent conditions and/or magnitude, and probability (indirectly) of a
random assignment to conditions correlation
➔ may study the effects of preexisting
antecedent conditions—life events or subject Range truncation
characteristics—on behavior.
➔ In experiments, researchers randomly assign
subjects to antecedent conditions that they ➔ an artificial restriction of the range of X and Y
create. An experiment might randomly that can reduce the strength of a correlation
assign subjects to either daily ibuprofen or coefficient
aspirin use, and then measure their
incidence of Alzheimer’s Outliers
Why should we use quasi-experiments instead of ➔ extreme scores (extra??)
experiments? ➔ They usually affect correlations by disturbing
the trends in the data. Range truncation
➔ We should use quasi-experiments when we removes outliers
cannot or should not manipulate
antecedent conditions.
COEFFICIENT OF DETERMINATION (R2)
➔ Quasi-experiments could study the effect of
spouse abuse on the frequency of child
abuse ➔ estimates the amount of variability that can
be explained by a predictor variable.
PEARSON CORRELATION COEFFICIENT
➔ For example, Chaplin et al. (2000) showed
that handshake firmness accounted for 31%
➔ used to calculate simple correlations of the variability of first impression positivity.
(between two variables) and may be
expressed as: r (50) = +.70, p = .001.
WHY DOESN’T CORRELATION PROVE
➔ Correlation coefficients have four properties.
CAUSATION?
linearity, sign, magnitude, and probability
➔ correlational studies do not create multiple
PROPERTIES OF CORRELATION
levels of an independent variable and
LINEARITY randomly assign subjects to conditions, they
➔ how the relationship between x and y can be cannot establish causal relationships.
plotted as a line (linear relationship) or a ➔ additional reasons
curve (curvilinear relationship) 1. casual direction
SIGN 2. bidirectional causation
➔ refers to whether the correlation coefficient is 3. the third variable problem
positive or negative
MAGNITUDE
➔ the strength of the correlation coefficient, CAUSAL DIRECTION
ranging from -1 to +1
PROBABILITY ➔ correlations are symmetrical, A could cause
➔ the likelihood of obtaining a correlation B just as readily as B could cause A.
coefficient of this magnitude due to chance ➔ e.g. does insomnia cause depression or
does depression cause insomnia?
QUASI-EXPERIMENTAL
What does a scatter plot show?
➔ Quasi = “seeming like” ➔ graphic display of pairs of data points on the
➔ meaning: superficially resemble x and y axes.
experiments, but lack their required ➔ A scatter plot illustrates the linearity, sign,
manipulation of antecedent conditions and/or magnitude, and probability (indirectly) of a
random assignment to conditions correlation
➔ may study the effects of preexisting
antecedent conditions—life events or subject Range truncation
characteristics—on behavior.
➔ In experiments, researchers randomly assign
subjects to antecedent conditions that they ➔ an artificial restriction of the range of X and Y
create. An experiment might randomly that can reduce the strength of a correlation
assign subjects to either daily ibuprofen or coefficient
aspirin use, and then measure their
incidence of Alzheimer’s Outliers
Why should we use quasi-experiments instead of ➔ extreme scores (extra??)
experiments? ➔ They usually affect correlations by disturbing
the trends in the data. Range truncation
➔ We should use quasi-experiments when we removes outliers
cannot or should not manipulate
antecedent conditions.
COEFFICIENT OF DETERMINATION (R2)
➔ Quasi-experiments could study the effect of
spouse abuse on the frequency of child
abuse ➔ estimates the amount of variability that can
be explained by a predictor variable.
PEARSON CORRELATION COEFFICIENT
➔ For example, Chaplin et al. (2000) showed
that handshake firmness accounted for 31%
➔ used to calculate simple correlations of the variability of first impression positivity.
(between two variables) and may be
expressed as: r (50) = +.70, p = .001.
WHY DOESN’T CORRELATION PROVE
➔ Correlation coefficients have four properties.
CAUSATION?
linearity, sign, magnitude, and probability
➔ correlational studies do not create multiple
PROPERTIES OF CORRELATION
levels of an independent variable and
LINEARITY randomly assign subjects to conditions, they
➔ how the relationship between x and y can be cannot establish causal relationships.
plotted as a line (linear relationship) or a ➔ additional reasons
curve (curvilinear relationship) 1. casual direction
SIGN 2. bidirectional causation
➔ refers to whether the correlation coefficient is 3. the third variable problem
positive or negative
MAGNITUDE
➔ the strength of the correlation coefficient, CAUSAL DIRECTION
ranging from -1 to +1
PROBABILITY ➔ correlations are symmetrical, A could cause
➔ the likelihood of obtaining a correlation B just as readily as B could cause A.
coefficient of this magnitude due to chance ➔ e.g. does insomnia cause depression or
does depression cause insomnia?
