YouTube Presentation: http://bit.ly/GradTrackStatistics2018
Dr. Gary Burns, Professor, School of Psychology, Florida Institute of Technology, Evans Library GradTrack Workshop
11. Why does Scott spend so much
more time playing Dominos
than his other games?
6
6.5
7
7.5
8
8.5
9
9.5
10
10.5
Dominos 7th Continent Terraforming Mars
HoursSpentPlaying
12. What game should I buy next?
Title Avg.
Rating
Num.
Voters
Deep Madness: Rise of
Dagon(2017) 10 2
Twilight Imperium: Fourth
Edition(2017) 8.80 3899
Wander: The Cult of
Barnacle Bay 8.80 27
Pandemic Legacy: Season
1 (2015) 8.498 27,385
13. Lessons from Board Games
We need to…
be specific with the questions that we are asking.
make sure that the data can answer our questions.
be careful not to misinterpret or misrepresent the
data.
remember the limitations of our data.
15. The 5 Questions Data Science Answers
Question 1: Is this A or B?
Question 2: Is this weird?
Question 3: How much? Or how many?
Question 4: How is this organized?
Question 5: What should I do now?
(a Microsoft perspective for folks
less obsessed with board games)
16. Gary’s Steps to Move
Forward with Statistics
Learn about data!
Understand sampling error and
corresponding standard errors
Grasp statistical significance
Know what statistical test to use
Search for the
^
tests
17. Keeping it Simple
What do we want to know about?
Type of data?
How is our data structured?
18. Keeping it Simple
What do we want to know about?
Sample (descriptive statistics)
Population (inferential statistics)
Type of Data
How is our data structured?
20. Keeping it Simple
What do we want to know about?
Type of Data
Nominal
Ordinal
Interval
Ratio
How is our data structured?
21. Keeping it Simple
What do we want to know about?
Type of Data
How is our data structured?
One group of objects with one score per object
One group of objects with two or more variables
measured for each object
Two or more groups of objects with the same
variable(s) measured
22. One group of objects with one score per
object
Typically describing something
Descriptive statistics include frequencies, proportions, percentages, mean,
median, and mode
Example Inferential Statistics Questions
Is this sample different from a known population value? (single-sample t-test)
Do these frequencies adhere to a uniform distribution or do they follow a normal
curve distribution? (chi—square good of fit)
23. One group of objects with two or more
variables measured for each object
Describe and evaluate the relationship between variables
Descriptive statistics include (previous slide) + correlation coefficients and
regression coefficients
Example Inferential Statistics Questions
Does the distribution of frequencies depend on the level of another variable? (chi-
square test of independence)
Can we predict the value of one variable from another variable? (regression)
Are these numerical scores related to a dichotomous variable? (point-biserial
correlation coefficient)
24. Two or more groups of objects with the
same variable(s) measured
Describe and evaluate differences between groups of scores
Descriptive statistics will primarily be means, median, mode, and categories
Inferential statistics
Are two groups different from each other? (independent-measures t-test)
Did two groups develop differently over time? (repeated-measures ANOVA)
Is the rank order of one group different than the rank order of another group?
(Mann-Whitney U test)
25. Two or more groups of objects with the
same variable(s) measured
Describe and evaluate differences between groups of scores
Descriptive statistics will primarily be means, median, mode, and categories
Inferential statistics
Mann-Whitney U – evaluates rank order differences between two groups
Kruskal-Wallis Test – evaluates group differences for more than two groups
Friedman Test – evaluate repeated measurements across groups
Chi-square Test for Independence - test if frequencies on one variable depend on
the level of a another variable
(^ these are for nominal or ordinal data)
26. Lots of tools like this online
(the link just googles “choosing the right
inferential test” and goes straight to the pictures)
29. Common Data Analysis Programs
Excel
SPSS or programs like SAS, Minitab, MATLAB,…
JASP – Developed as a free, open-source alternative to SPSS
R – free software environment for statistical computing and graphics
Platform independent
Consistently in the top programs languages identified by IEEE
Data scientists report Python and R are the most common programs they use
Survey of technology professionals name R as the highest-paying skill
30. R you interested?
R was designed by statisticians
Freely available and has add-ons
to meet your needs
R Studio – makes R easier to use
Shiny – web applications for
visualizing data
R Packages – users develop
packages to meet their needs
But it does have a steep learning
curve
Not only is Dr. Joe Houpt
really tall, he also authored
the sft package (Functions for
Systems Factorial Technology
Analysis of Data)
35. Campus Resources for Getting Started
with Statistics Building 405
Academic Quad, West
entrance. See
on Google Maps.
36. MAC
Open M-F 9-5 (MTF)/6:30 (WTh)
Walk Ins Welcome!
Priority does go to students in math classes!
Offers tutoring for
Algebra, Precalc, Calc 1, 2, and 3
Differential Equations
Probability and Statistics
Complex Variables
Intro to PDE
Discrete Math
Models in Applied Math
Functions and Modeling
37. Campus Resources for Getting Started
with Statistics
Bleakley
Bolton
Bostater
Burns
Carney
Conradt
Converse
Mingareev
Nezamoddini-
Kachouie
Park
Smith
Wang
Deaton
Dshalalow
Edkins
Gallo
Gates
Jensen
Mesa Arango
38. Special Thanks To!
(click for links)
Facebook’s The Board Game Group users who
shared their BG Stats screen shots with me.
Local game shop at 3020
W. New Haven Ave. –
they have a library of
games that you can stop
by and play – open late
Thursdays!
My Wife for
Supporting My
Hobby! (no link)
JASP