Independent Samples Test Worksheet

Statistics for the Behavioral SciencesSpring 2022
PSYCH-UA.10
Computer Assignment #3: Check syllabus for due date
Answer the following questions using SPSS and some computations by hand. Please make clear
which SPSS output goes with which question. Only include necessary output. Use the notes from
lab to help you. Remember, even if you have a large sample, you will be doing a t-test in SPSS,
not a z-test.
1) Open data labeled Cosmetic Surgery Revised in Brightspace (Content > SPSS
Assignments > SPSS Assignment Data > Cosmetic Surgery Revised B). In the US,
there was a 1600% increase in cosmetic surgical and non-surgical treatments between
1992 and 2002. According to researchers, there are two main reasons to have cosmetic
surgery: 1) to help with a physical problem and 2) to change one’s external appearance.
This data looks at whether there is a difference in Quality of Life After Surgery between
people who elect to undergo cosmetic surgery for a physical problem and those who want
to change their external appearance. The two variables you need are Reason (reason for
surgery) and Quality of Life After Surgery (Post_Qol); higher numbers mean participants
experience a better quality of life.
Only using SPSS, calculate a t-test and the 95% CI for the difference in Quality of Life
After Surgery between people who undergo the surgery for a physical problem and those
who want to change their appearance, just as you did in lab. Select the Group Statistics
box and the Independent Samples Test (t-test portion) for your output. Answer the
following questions:
a) In symbols and words, what are the null and alternative hypotheses for this
experiment?
b) What is the standard error of the difference between the means (assume equal
variances for this question)?
c) Using Levene’s test, do we have homogeneity of variance? How do we know?
d) What is the conclusion (assume equal variances for this question)? Write up the
results in APA format.
e) What is the 95% confidence interval for these data (assume equal variances for
this question)? How would you determine significance using this interval?
f) Explain what a Type I and Type II error would be for this study. Which one did
we possibly make?
Output should include the Independent Samples Test and Group Statistics
boxes. Just to make sure you’re on the right track, the mean difference
should equal .7825 and the sample sizes should be 36 people who wanted to
change their appearance and 84 people who had a physical reason for their
surgery.
2) Researchers would like to investigate the Baseline Quality of Life of these patients for the
different groups. Before they analyze their data, they would like to do a power analysis.
The descriptives for the variable “Baseline Quality of Life” are: M1 = 66.0, s12 = 45.0,
n1 = 36 for people who chose surgery to change their appearance and M 2 = 63.0,
s22 = 48.0, n2 = 84 for those who choose surgery for a physical reason.
a) Use these descriptive statistics to find the power for this study. Please show your
hand calculations: First, find s2pooled. Then find g (the sample estimate of d remember to take the square root of s2pooled). Always use the absolute value of g
(or d) for these calculations. Find δ, remembering to replace d with g. Find the
harmonic mean of the n’s because you have unequal sample sizes. Assuming
α = .05 (two tailed), use the table to find the power.
b) What is the probability of a Type II error for this study?
c) Using the formula to find n, what sample size would you have needed for a
power of .90? Remember to use g instead of d (absolute value).
3) Open data labeled Assignment NELS B in Brightspace (Content > SPSS Assignments >
SPSS Assignment Data > Assignment NELS B). This data consists of measures of
achievement outcomes in four core subject areas (English, history, mathematics and
science), in addition to personal, familial, social, institutional, and cultural factors that
might relate to these outcomes. We are specifically interested in the relationships
between tenth grade self-concept, teacher interest, socioeconomic status, and tenth grade
measures of academic achievement in math, social studies and science. The variables you
want are:
Slfcnc10 (tenth grade self-concept)
SES (socioeconomic status)
Tcherint (teacher interest) Even though this variable is technically an ordinal scale, we are
treating here as a ratio scale
Achsls10 tenth grade social studies achievement)
Achmat10 (tenth grade math achievement)
Achsci10 (tenth grade science achievement)
a) Calculate correlations in SPSS (Analyze>Correlate>Bivariate) for all of the
variables. Print out the box of correlations.
b) Using only SPSS, how many correlations are significant? Between which two
variables is the highest correlation?
c) Write up the results in APA format for the correlations between socioeconomic
status and 10th grade math achievement, teacher interest and 10 th grade math
achievement, and 10th grade science achievement and 10th grade math
achievement (even if they are not significant write them in APA format).
d) Out of all the variables for this problem, which is the best predictor of 10 th grade
math achievement? Why?
e) Create and print an interactive scatterplot (Graphs > Chart Builder > Simple
Scatterplot) of the variables socioeconomic status (on the X-axis) and 10 th grade
math achievement (on the Y-axis). Describe the pattern of data.
f) What happens if the range is restricted to only students that score at or above an
SES of 20: Go to Data>Select Cases>If condition is satisfied. Move SES from
the left-hand box to the formula box (using the arrow). Then put ≥ 20, so that
the selection looks like this: SES >= 20. Click continue, then OK. This should
select only those SES scores at or above 20.
g) Find the correlation between SES and 10th grade math achievement the same way
as before (Analyze>Correlate>Bivariate) for just these two variables. (Notice that
N will now equal 172). Print out this box of correlations. Also create and print
out a new scatterplot.
h) What is the correlation now? Why is it different from before?
Output should include two correlation boxes and two scatterplots. The N for SES and 10 th
grade math achievement should be 399 (before the restriction of range).
1
A One-Stop-Shop for Nailing Your SPSS Assignments
2
SPSS Grading Guide
●
We are using Gradescope for all assignments. All grading rubrics will appear on Gradescope as
you submit your assignment
●
Not typed (-1 off overall grade) (this does not apply to formulas or hand calculations, although
typed is preferable)
o
●
Although typed makes it easier for us to read, the actual reason why we require the assignment to
be typed is because the intention of this course is to prepare students to be able to conduct and
report statistics in the future, should they need to, and in the real world they would never
handwrite a report (for statistics or any other course).
Output printed directly from SPSS; it should be copied to a word document (-2 off overall grade)
o
This is because the SPSS assignments are also assessing students’ ability to differentiate output,
so including all output (even if not requested) does not demonstrate this ability – it would be akin
selecting all multiple choice answers.
▪
●
Extra output is included (-1 off overall grade)
o
●
Similar to previous points about how you present output in reports
Critical value is included in addition to test statistic (-.25 each occurrence)
o
●
Similar to above point for typing; they should be properly reported and so tables shouldn’t be
wrapped or split within/across pages (unless the amount of output justifies it, which, for this
course, it doesn’t)
Graphs not properly titled (-.1 for each occurrence)
o
●
See above
Split tables (-.2 for each occurrence)
o
●
We usually only indicate -2 on the first assignment (which is where points aren’t actually
taken off), and tend to only end up subtracting up to 1pt since usually there’s not an
egregious amount of additional output (although there are the occasional cases)
This is essentially like circling two multiple choice answers – as if you don’t know which is
correct and are hoping to get credit based on one being right.
Hand-calculating a value that should have been retrieved from SPSS output (-.2 for each
occurrence)
o
These are SPSS assignments after all, so unless the question specifies to calculate a value by
hand (which some do) you should be finding the answers via SPSS
▪
Grading this way is also consistent with exams in which some questions require answers
only using SPSS output or else points are deducted
3
The output is clearly
sectioned by question.
There’s no extraneous output
such as syntax or
unrequested tables/charts.
SPSS Output
4
Question 1
Question 2
Both graphs have informative
titles that include the relevant
variable(s).
Output has been resized so that
there are multiple tables and
graphs on this one page. Narrow
margins are acceptable and may
help you fit more output
efficiently.
However, all output must
remain legible, with nothing cut
off of the page.
Independent Samples Test
Levene’s Test for
Equality of
Variances
F
Perceived
Choice
subscale of
Intrinsic
Equal
variance
s
assumed
.08
2
Sig.
.775
t-test for Equality of Means
t
2.9
12
df
Sig.
(2taile
d)
245
.004
Mean
Differe
nce
-.4628
7
Std.
Error
Differe
nce
.15897
95%
Confidence
Interval of the
Difference
Low
er
Upper
-.77
599
-.149
75
5
Motivation
Inventory
Equal
variance
s not
assumed
2.9
00
236.
631
.004
-.4628
7
.15961
-.77
732
-.148
43
Written Responses
Question 1
a) I am now answering the question in sentence form. Blah blah blah blah blah. Words and words and words
and words and words.
b) This is my answer to another part of question 1. Text text text text.
Question 2
a) I’m answering this question, now, also in sentence form. Blah blah blah blah blah. Words and words and
words and words and words.
b) This is my answer to part b of question 2. Text text text text.
As you can see, this organization follows a logical
and consistent pattern, with question #s clearly
labeled.
All written responses are typed—never handwritten—so as to resemble a formal report.
Other options for organizing your assignments:
You could lead with the written responses, then include your SPSS
Output section.
Instead of separating out the SPSS output, you could simply organize
by Question #:
Question 1
Written Response
SPSS Output
Blank space to hand-write in
formulas/hand-calculations, if applicable
Question 2
Written Response
SPSS Output
Blank space to hand-write in
6
Academic Integrity
Each assignment must be your original work!
• Absolutely NO sharing of outputs
• ALL written work must be done independently
Any plagiarism of another’s assignment will result in a ZERO for the assignment and appropriate
disciplinary action at the instructors ’discretion.
Your understanding of this is attested to in the header included in every assignment (see page 2).
What counts as plagiarism?
• “Collaborating between two or more students who then submit the same paper under their individual
names.”
• “Submitting another student’s work with your name on it.”
• “Paraphras[ing] or restate[ing] someone else’s facts, analysis and/or conclusions.”
Bottom line: It’s not worth the risk!
● Grading is done by experienced TAs who are able to spot tell-tale signs of copying.
7
Miscellaneous Tips
•
Due dates are listed on the syllabus, so plan ahead
•
Lab worksheets guide you step-by-step through the analyses you’ll need to run for assignments
•
Check that you entered data correctly
•
Writing in black ink is preferred
•
Make sure you answer ALL parts of ALL questions
•
Do SPSS problems in the order on the assignment and label each answer
•
Title & label graphs
•
Box or highlight every answer to hand-calculations
•
Only include relevant output
•
Resize output & arrange in Word document to reduce excess pages and avoid splitting charts/tables
between pages. Save this as a pdf for Gradescope.
•
Double-check output with a classmate’s if possible. This way, you can troubleshoot in case of
discrepancies and avoid major point losses. (However, see info on academic integrity.)
•
See “Rules” on pages 3-4 for additional tips!