Lab 8: Partial Correlation Assignment

1
Lab 8 Partial Correlation Assignment
The purpose of this weeks lab assignment is to
give you some practice in conducting and
interpreting partial correlations. The following
4 pages contain correlation matrices among a
number of variables that are commonly studied
among personality psychologists. Please note
that these are real data! For Questions 1 5,
your goal is to test partial correlations for 5
sets of variables. These will be explained on the
pages that follow. For Questions 6 10 you will
again be asked to test partial correlations, but
youll be able to choose what to test. Please
note that the equation for computing partial
correlations is available in the lecture notes. I
encourage you to try at least one of these by
hand so you can get a feel for how the
computations are done. Once you have the hang of
it, I encourage you to use a simple web-based
program that will do the work for you
http//www.yourpersonality.net/psych437/fall2011
/partial.pl This web app will compute the
partial correlations for 3 variables. You have to
enter in the names of the variables and the
correlations among the three variables (i.e., r
(x, y), r (x, z), and r(y, z)). The app will then
compute all three possible partial correlations
for you (e.g., r(x,y . Z) ). Please note that you
will typically only be interested in one of these
three outputs, so use the variable labels instead
of X, Y, and Z so you dont get confused.
2
Data set 1 Leadership and Narcissism in Groups
The following correlations come from a study on
narcissism and leadership. Specifically, the
authors were interested in the question of
whether highly narcissistic people are more
likely to emerge as leaders in groups. One of
the key correlations they report is that
narcissism correlates .20 with group ratings of
one anothers expertise. In other words, group
members attributed more expertise to highly
narcissistic individuals. One potential
confound, however, is that highly narcissistic
people might be more outspoken or sociable in
groups (as measured with the Sociability variable
in the correlation table). And, in turn,
outspoken people might be judged as having more
expertise. Question 1. What is the partial
correlation between Narcissism and Expert Ratings
when statistically controlling Sociability?
(bottom table) Question 2. What is the partial
correlation between Narcissism and Desire to Lead
after controlling for Extraversion? (top
table)
3
Brunell, A. B., Gentry, W. A., Campbell, W. K., Hoffman, B. J., Kuhnert, K. W., DeMarree, K. G. (2008). Leader emergence The case of the narcissistic leader. Personality and Social Psychology Bulletin, 34, 1663-1676.
4
Data set 2 Early Maternal Sensitivity Subsequent
Social Development
The following correlations come from a
longitudinal study on social and emotional
development. Approximately 1000 children were
assessed from early childhood through Grade 6.
During the first 5 years of life, Early maternal
sensitivity was assessed (early sensitivity),
along with the sex of the child, the familys SES
(income-to-needs), the childs ethnicity
(ethnicity, coded here in a binary fashion), and
the mothers educational levels (maternal
education). At each assessment wave, Maternal
Sensitivity and Social Competence (as rated by
teachers) were assessed. One of the key
correlations the researchers report is that Early
Sensitivity correlates .28 with Social Competence
at Grade 6. In other words, children who received
higher levels of maternal support in the first
five years of life were rated as more socially
competent by their teachers at Grade 6. One
potential confound, however, is that maternal
sensitivity is relatively stable over time. Thus,
the correlation between Early Sensitivity and
Social Competence (Grade 6) might disappear if
Sensitivity at Grade 5 is statistically
controlled. Question 3. What is the partial
correlation between Early Sensitivity and Social
Competence (Grade 6) when statistically
controlling Maternal Sensitivity (Grade 5)?
5
Fraley, R.C., Roisman, G.I., Haltigan, J.D.
(2011). The legacy of early experiences in
development Formalizing alternative models of
how early experiences are carried forward over
time. Developmental Psychology.
G1 G6 grades 1 6. Social comp social
competence rated by teachers, income-to-needs is
a measure of SES. Early sensitivity is a measure
of observer ratings of maternal sensitivity
toward the child during the first 5 years of life.
6
Data set 3 Alcohol consumption and sexual
partners
The following correlations come from a
longitudinal study on alcohol consumption and
risky sexual activity. The researchers assessed
alcohol use from Age 15 to Age 28. They also
assessed the number of sexual partners the
subjects reported across the same assessment
waves. One of the key correlations the
researchers report is that the amount of alcohol
that people consume correlates with the number of
sexual partners they have. For example, the
correlation between alcohol consumption and
sexual partners is .47 at Age 15. (It is
noteworthy that the correlation appears each
assessment wave too.) One potential confound,
however, is Impulsivity. Namely, highly impulsive
adolescents might be more likely to consume
alcohol and to engage in sexual behavior. Indeed,
as can be seen in the table, both of these
correlations are positive. Question 4. What is
the partial correlation between Alcohol
Consumption and Sexual Partners at Age 15 when
statistically controlling Impulsivity? Question
5. What is the partial correlation between
Alcohol Consumption at Age 15 and Sexual Partners
at Age 21 when statistically controlling
Impulsivity?
7
Dogan SJ, Stockdale GD, Widaman KF, Conger RD.
(2010). Developmental relations and patterns of
change between alcohol use and number of sexual
partners from adolescence through adulthood.
Developmental Psychology, 46, 1747-59.
8
Assignment

• Turn in your answers for the first five
  questions. Also answer the following 5 questions.
• Questions 6 10
• For the last 5 questions I would like for you to
  examine 5 other correlations that seem
  interesting to you among these datasets.
• In each case, identify two variables that are
  correlated that you think are interesting. Then,
  think of a potential confound variable that might
  explain the correlation between those two
  variables that also exists in the same dataset.
  Conduct a partial correlation analysis to see
  what the correlation is between the original two
  variables when the third variable is
  statistically controlled. Finally, comment on
  what your findings suggest about the relationship
  between the original two variables.
• For each set, be sure to list the following
  information
• What is the original correlation between X and Y?
  What does it mean in words? (e.g., Narcissists
  are more likely to be judged as having expertise
  in groups. r .40)
• What is a potential confound and why? (e.g.,
  Sociability. Highly sociable people might be more
  narcissistic and might be more likely to be
  judged by others as having expertise.)
• What is the partial correlation between X and Y
  while controlling Z. What does it mean? (e.g.,
  The partial correlation between narcissism and
  expertise is .00 when holding sociability
  constant. This seems to suggest that the original
  correlation was due to the confounding influence
  of sociability.