You are on page 1of 9

Name:

Amber Costantino
Project #3 Psychology 240T
Dr. Yockey Fall 2015

Project 3 Psychology 240T






Complete each problem below, answering each question as completely as possible. You will
need to use SPSS to answer some of the questions on the project. For this project, you are
allowed to work with others, as long as each person is making significant and meaningful
contributions to the work. You may also access the class videos and any other class materials as
needed (as well as the internet). If you would like, you may hand-write your answers just
please make sure they are legible, or you can type in the answers, as a copy of the project in
Word is also available on blackboard. SPSS output is not required to be turned in with the
project. The project is due on Wednesday, 12/16, at 4 pm. For turning in the project, emailing
me a copy of your project is preferred (either a Word document with typed answers or a
scanned copy of your written answers is preferred). However, if you are not able to do this, you
are welcome to turn it in to the Psychology department office by 4 pm on the 16th. (It will be
time and date stamped when you turn it in. There is no in person class meeting on the 16th.)

1. For the following problem, access the SPSS file PASS data.sav in blackboard (in the project 3
folder). In the file, there are the following eight variables:
Variable
FF1
FF2
FF3
FF4
FF5

AT1
AT2
AT3

Meaning
Question 1 fear of failure procrastination subscale
You were concerned the professor wouldn't like your work.
Question 2 fear of failure procrastination subscale
You were worried you would get a bad grade.
Question 3 fear of failure procrastination subscale
You didn't trust yourself to do a good job.
Question 4 fear of failure procrastination subscale
You were concerned you wouldn't meet your own expectations.
Question 5 fear of failure procrastination subscale
You set very high standards for yourself and you worried that
you wouldn't be able to meet those standards.
Question 1 aversiveness of tasks procrastination subscale
You really disliked writing term papers.
Question 2 aversiveness of tasks procrastination subscale
You didn't have enough energy to begin the task.
Question 3 aversiveness of tasks procrastination subscale
You felt it just takes too long to write a term paper.

The above items come from a popular measure of academic procrastination developed by
Solomon and Rothblum (1984). On this scale, participants are asked to recall a time when they
procrastinated on a term paper and then are asked to answer a number of questions about that
event (see above). In their original factor analysis, Solomon and Rothblum reported a two-
factor solution, which they named fear of failure and aversiveness of tasks. Your job as the
factor analyst is to take the data file (collected on a new sample) and assess the factor structure
of the scale. In other words, do you find support for the Solomon and Rothblum solution? In
performing the analysis, be sure to answer each of the questions below. (30 points total).
a. Perform Bartletts test of sphericity. Is the test significant? Report the df, chi-square value,
and p-value below (and assess whether or not the test is significant).

Bartletts test of sphericity was significant, 2 = 933.44, df = 28, p < .001




b. What does the result in (a) above indicate? Be specific.

Results from the Bartletts test of sphericity indicate that the correlation matrix
created by our variables is significantly different from an Identity Matrix.
Therefore, the correlations between our variables are significantly different
from zero.
c. Perform a principal components analysis on the data file in SPSS. Include all 8 items in your
analysis. As research indicates that the subscale can be correlated to a certain degree, I
recommend performing an oblique rotation to the data. After performing the analysis, how
many components should be retained according to Kaisers eigenvalue > 1 rule?

Two components should be retained according to the Kaisers eigenvalue > 1


rule

d. How many components would you retain using a Scree plot? Describe how you arrived at
your decision.

Three components would be retained using the Scree plot because the slope of
the line looks fairly level (no slope) after component 3, indicating a scree.
Whereas the slope between components 1, 2, and 3 is steep, indicating it is
above the scree


e. Report and Interpret the communalities for FF1 and FF2 below.

FF1: .604, 60.4% of the variance in FF1 is accounted for by the component
FF2: .635, 63.5% of the variance in FF2 is accounted for by the component


f. How much total variance was explained from your solution (i.e., the components retained)?

61.67% of the total variance was explained by the two components retained

g. Using a component loading cutoff of .40 (i.e., interpret loadings of .40 and above), interpret
the component solution. That is, which items load on which components? Be specific.

FF1, FF2, FF3, FF4, and FF5 all load on component 1 (loadings between .75-.82).
AT1, AT2, and AT3 all load on component 2 (loadings between .70-.89)


h. Interpret the components. That is, give them names. It is fine to use the names provided by
the original authors as long as it is defensible from the solution.
Although all factor loadings for FF questions and component 1 were high, the highest factor
loading for component 1 was FF2 (You were worried you would get a bad grade.) or FF4
(You were concerned you wouldn't meet your own expectations.). I think the authors
original name of Fear of Failure procrastination subscale is a defensible name because
both of those questions seem to be measuring a concern for the consequences of failing to
meet a deadline.
Again, although all factor loadings for AT questions and component 2 were high, the highest
factor loading for component 2 was AT3 (You felt it just takes too long to write a term
paper.). I think this subscale should be renamed fear of commitment procrastination
subscale because this question seems to be measuring a concern for the amount of time
they have to commit to the task. Further, questions such as energy expenditure (AT2) of the
task and dislike (AT1) of the task also may be measuring fear to commit to the task.


i. What is the correlation between the components? Is this a large value?

The correlation between the components is .229, this is a small value



j. Overall, based on the results, would you consider this a fairly clean solution or not? Why or
why not?

Yes, I would consider the 2 factor solution to be a clean solution because the
factor loadings are high, the percent of total variance is moderate, and the
correlation between the components is low.


2. Shown below is an edited table provided from an analysis in SPSS for a principal components analysis.
While the output is somewhat edited below from SPSS, all components from the original solution are
shown (16 points).
Component

Eigenvalue

2.25

1.10

.70

.55

.40


Based the above table, answer questions (a) through (d) below.
a. How much variance did component 1 account for? (Show your work.)

2.25 / 5 = .45, 45% of the variance is accounted for by component 1




b. How much variance did component 2 account for? (Show your work.)

1.10 / 5 = .22, 22% of the variance is accounted for by component 2




c. Using the eigenvalue > 1 rule, how many components would you retain?

Two components should be retained according to the Kaisers eigenvalue > 1 rule


d. How many components would you retain with a scree plot? (You may find this easier to answer if you
plot the values on paper.) Please justify your answer with a few comments as to how you made your
decision.

Two components would be retained using the Scree plot because the slope of the line would
look fairly level (no slope) after component 2, indicating a scree. Whereas the slope between
components 1 and 2 is steep (.85 difference between 1 & 2), indicating it is above the scree.


3. Suppose someone brought to your desk the following results from a PCA they ran. For the PCA, they
analyzed five variables, with the eigenvalues (as reported by them) provided below. (10 points).
Component

Eigenvalue

3.45

2.25

1.70

.40

.30


a. What is wrong with the analysis? Be sure to justify your answer.

The Eigenvalues should add up to the number of components (5), but they do not in this case
(8.1).




4. Using the article excerpt from Yockey (2016) where an EFA was used, answer the following questions
(24 points).

a. How many variables were analyzed in the EFA?

Five


b. How many components were retained?

One component


c. What methods were used for component extraction (i.e., retaining components)? Did the methods all
agree on the number of components to be retained?

eigenvalue > 1, scree plot, and parallel analysis were used and they agree on
the number of components to be retained (1)



d. What were the component loadings for each of the items? (Provide the loadings and item numbers
below.)

Item
1. I put off projects until the last minute.

Component Loading
.82

2. I know I should work on schoolwork, but I just dont do it.

.78

3. I get distracted by other, more fun, things when I am


supposed to work on schoolwork.

.73

4. When given an assignment, I usually put it away and forget


about it until it is almost due.

.85

5. I frequently find myself putting important deadlines off.

.86




e. Interpret the component loading for item #5 of the APS-S. (In your description, be sure to note what a
component loading is, i.e., what statistical measure is it equal to.)

The Component Loading is the Pearsons r correlation coefficient of the variable w/ the
hypothetical component. Item #5 is strongly correlated (r = .86) with component 1.



f. As an alternative to the rule cited in the article by Nunnally, another approach for establishing the
reliability (or stability) of components is provided in your text on page 333 (section 11.7). Take a look at
the 3 rules summarized from Guadagnoli and Velicer (1988) and apply them to Yockey (2016). Does the
component solution provided in Yockey (Table 1) meet any of the three recommendations provided in
your text on page 333 (either 1, 2, or 3)? If so, which one does it meet and why?

Yokey (2016) meets the first recommendation Components with four or more loadings
above .60 in absolute value are reliable, regardless of sample size because all five loadings
are above .60.
Yokey (2016) does not meets the second or third recommendation.


5. A marketing researcher investigated the location of web page ads on both the number of clicks made
and the time spent on the page. For simplicity, the locations have been characterized as top, middle,
and bottom of the page. Both clicks made and time spent on page are expressed on a 1 to 10 scale,
where higher values indicate more clicks made and more time spent on the page, respectively. The data
for the study are provided below. (20 points)
Page Location
Top
Top
Top
Top
Top
Top
Top
Middle
Middle
Middle
Middle
Middle
Middle
Middle
Bottom
Bottom
Bottom
Bottom
Bottom
Bottom
Bottom

Clicks
5
4
6
5
4
3
4
6
7
7
8
9
8
7
2
1
2
3
4
6
4

Time
5
4
3
4
5
6
5
8
7
4
3
7
8
7
4
3
5
3
5
3
1

Perform the appropriate test in SPSS to assess whether location makes a difference when considered
jointly on the variables clicks and time. Using = .05 for the overall test. Perform follow-up tests as
needed, with each test conducted at the appropriate Bonferroni adjusted level required to maintain
overall alpha at .05.
a. Report the results of the overall test below. Is the test significant at = .05?

There was a statistically significant difference in the number of clicks made and the time
spent on the page based on add location, F (4, 34) = 10.98, p < .001; Wilk's = 0.190, partial
2 = .564.






b. Report the Bonferroni adjusted alpha level that each follow-up univariate test should be conducted at
below.

.05 / 2 = .025
Bonferroni adjusted alpha = .025


c. Report the results of each of the ANOVA follow-up tests below (F, df, etc.) and interpret whether each
test is significant or not. (Tukeys is not required here.)

Ad location has a statistically significant effect on both number of clicks (F (2, 18) = 21.55; p <
.001; partial 2 = .71) and time spent on the page (F (2, 18) = 6.38; p < .01; partial 2 = .42).



d. If relevant, report the results of Tukeys test below. For each significant dependent variable, describe
both significant and non-significant differences between the groups (i.e., be sure to describe all groups
on each sig. DV for full credit).

Add Location and Clicks: The number of clicks between adds on the bottom of the page and
adds on the the top of the page did not significantly differ from each other, but adds in the
middle of the page had significantly higher clicks than both the top of the page and the
bottom of the page.
Add Location and Time Spent on the Page: The amount of time spent on the page did not
significantly differ between adds on the top of the page and adds on the the top of the page
or the middle of the page. However, the amount of time spent on the page did significantly
differ between adds on the bottom of the page and adds on the the middle of the page.



This concludes the project. Congratulations on being finished with the course!
It has been a pleasure having you in the class this semester. I hope you all have a great break!

You might also like