Transcript Teaching Presentation

For This Class
For Next Class
 Beer, Puke, & Rubber Ducks
To do list






What is logistic regression?
Why would I use this?
Fun example
Student affairs example
Muddies point
Sneak peak at next time
 Factor Analysis
Know your
audience
Number of people who have math or
stats related scar tissue?
Number of people who have opened
SPSS in the past 3 months?
Online folks, take the poll via twitter
#SAAquantresearch
Kirsten R. Brown, Ph.D.
Logistic
Regression
UWL-SAA Teaching Presentation
What is logistic
regression ?
A statistical method for analyzing
data where you have one or more
independent variables that you want
to use to determine an outcome
(dependent variable). The outcome
is dichotomous.
 We can predict stuff!!! (And this is fun)
Why use
logistic
regression ?
 The stuff we want to predict is
dichotomous. There are only 2 possible
outcomes.
 The dependent variable is also mutually
exclusive, everyone fits into 1 of those 2
outcomes (you can’t be both).
 We are interested in the odds of a particular
outcome.
Let’s predict
something!
Future student
conduct case…
 Research question: What factors predict the likelihood that a
student will vomit in the residence hall elevator?
Research
question
&
Variables
 Dependent variable: Vomit ; Yes = 1 /No =0
 Independent or control variables:









How much did the student drink?
Did the student eat dinner?
Is the student Male or Female?
What is the student’s body weight?
Has the student vomited in the elevator before?
What is the student’s GPA?
How old is the student?
Did the student combine alcohol with other drugs?
What floor does the student live on?
1. Say hello & get to know your data
2. Clean your data
3. Make variables/ re-code variables (if needed)
4. Check assumptions of logistic regression
 Sample size , Multicollinerarity , Outliers
Identify
analysis steps
5. Run preliminary tests
 Bi-variable correlations between each IDV & DV
6. Go play
 Select variables based on- previous research, theory,
logic, and occasionally data mining
 Build some models (this is where quant is just as
“messy” as qual!)
7. Think about your findings
 Do they make sense?
 Do they fit with / extend the literature?
What are the odds that:
 a female
 125 lb
Fictional
findings
 with a 3.2 GPA
 who lives on the 4th floor
 and consumed 10 beers and 2 shots of vodka
 ate a small dinner at 5 pm
 has not vomited in the elevator before
 and did not use other drugs
Research Question
Round 2:
Peer review
journal style
What factors best predict if an
institution will offer Autism
Spectrum Disorder (ASD)
specific programs?
Independent & Control Variables
# of students with ASD
# of disability resource staff
Diagram the Problem
ASD
Yes
Sensory accommodations
Diagram
the
Residence hall accommodations
problem
Peer mentoring programs
Transition programs
Type of institution
Institution size
Model
ASD
NO
But we only have 30 minutes and
we are not in a computer lab… so
This is the
point in class
where you
would do
group work
with SPSS
The website has 2 videos that
show you how to run a logistic
regression in SPSS
Your handout has screen shots of
SPSS output with how to do a
write up.
Resources
@kirsten_brown
#SAAquantresearch
https://kirstenbrown.org/teaching/uwl
SPSS How to Run a Logistic Regression
 http://youtu.be/uwpjTFox9ko
 http://youtu.be/Q5CYDK-p8L8
#SAAquantresearch
Muddiest point
A look ahead
at next week
Why: Understanding the logic, helps you
understand what you are reading before
class.
Because: We often get lost in the little
picture (e.g., SPSS output) and forget
about the big picture.
Questions?
 Kirsten R. Brown, Ph.D.

* Cartoons are by Larson, Gary (n.d.). The Far Side. http://www.thefarside.com/