SPRING 2010 Class Information:
Classes: Thursdays 12:00-3:00pm
Location: E14-525
Units: (3 0 3) G;
Staff | Description | Announcements | Assignments | Syllabus | Policies | Schedule
Dr. Matthew S. Goodwin
Office: E14-374K
Phone: 617-253-6341
mgoodwin ( @ media dot mit dot edu)
Elliott Hedman
Office: E14-274
Phone: 617-225-9611
hedman ( @ media dot mit dot edu)
Coco Krumme
Office: TBD
Phone: 617-225-9611
kak ( @ media dot mit dot edu)
Dr. Rosalind Picard
picard ( @ media dot mit dot edu)
Staff | Description | Announcements | Assignments | Syllabus | Policies | Schedule
Course Objectives:
In order to design effective experiments and be critical consumers of research, investigaotrs need to be able to generate relevant hypotheses, select appropriate research designs, compute proper statistics, and effectively communicate findings. By the end of the course, students will:
1. Have a thorough understanding of a variety of common research principles, designs, and statistical approaches employed in behavioral research;
2. Be able to address assumptions and limitations of various research designs and statistical approaches;
3. Understand how to formulate and apply hypothesis testing;
4. Get exposure to both qualitative and quantitative analyses. Methods will include ethnography, interviews, surveys, power analysis, effect size, correlation, regression, analysis of variance, repeated measures, and times series analysis; and
5. Communicate statistical results, both written and orally.
Motivation:
For a creative concept to gain traction it must be validated. We intend to introduce students with little or no prior training in behavioral research methods and statistics to evaluate whether a technology, idea, or experimental outcome differs from what would be observed randomly. We also strive to provide tools to better understand the assumptions and limitations of current research. Finally, we hope to promote a learning approach to research design and statistics in line with the Media Lab way of thinking: by drawing examples from real data and providing students with opportunities to learn concepts using their own technology and data.
Staff | Description | Announcements | Assignments | Syllabus | Policies | Schedule
Fri 01/29 Our first introudction class will be held February 4th, 12:00 - 1:00pm. A general overview will be given.
Fri 01/29 The second class begins February 18. This delay will provide you time to order books for the readings.
Staff | Description | Announcements | Assignments | Syllabus | Policies | Schedule
NOTE: With the exception of the first day of class, the readings and assignments are due 12:00 pm (Noon) on Wednesday before class.
2/18 (Answers due 2/17 at Noon)
| Read | |
|
Write |
1.Define the following concepts in your own words: internal validity, external validity, construct validity, and conclusion validity. 2. Kline describes several threats to internal, construct, conclusion, and external validity. Where applicable, describe three threats to each type of validity you anticipate confronting in your area of research. 3. Kline reviews a variety of experimental and quasi-experimental designs commonly used in behavioral science research. Pick a design that's applicable to your area of research, briefly describe how you would design a study that uses it, and describe strengths and weaknesses. 4. Kline highlights a number of common misconceptions about statistical significance testing, what he calls the "Big Five." Briefly describe when statistical tests are appropriate and what you can do to overcome their limitations. 5. List anything you didn't understand in the readings and/or concepts you'd like to discuss further in class. |
|
| Apply | |
| We will Be applying the definitions you used to your own research during class. |
2/25 (Answers due 2/24 at Noon)
| Read | |
| Write | |
1. What did you like about the ideas in the reading? 2. What did you not like about the presented ideas? 3. What questions came up for you through the readings? 4. What qurestions remain unanswered (or were raised) in the field of qualitative research? 5. Design a small qualitative study for your research study (150 words or less). What would the strength and weaknesses of this design be? |
|
| Apply | |
| We will be using answers to number 5 to lead discussion in class |
3/4 (Answers due 3/5 at 11:59 PM)
| Read | |
|
Write |
1. What is power? 2. What are the different types of power analysis? 3. Which of those types would apply to your research and why? 4. Give an example of a null hypothesis for your research. 5. What did you find interesting or confusing in the readings that you would like to go over more? |
|
| Apply | |
| We will be using answers to number 3 and 4 to lead discussion in class |
3/11 (Answers due 3/10 Noon)
| Read | |
|
Write |
1. Why would you use regression instead of using a T-test? (See chapter 10 of online statbook if you are unsure about T-tests). 2. On page 79 of Choen and Lea, six uses of correlation are given. Give an example of how you could use each of these for your own research. 3. On page 91 there are 4 assumptions of linear regressions. Explain in your own words these rules. 4. Create a regression formula that you might use in your experiment, use as many independent variables as you think appropriate. For example, for my research: Child's EDA = B0 + B1*(Boy/Girl) + B2*(Time of Day) + B3*(Day of Week) + B4*(Does child have ADHD) + B5*(Does child have autism) + B6*(How active was session) + B7*(How many sessions did this child have) + B8*(Parent's income level) + B9*(How long child has been doing therapy) + B10*(Child's age) 5. Of all the variables you listed above, would any of them be highly correlated with one another? (For example How many sessions of therapy the child has done and their age would be correlated together for me). 6. (bonus/optional) If two independent variables are highly correlated, using the 2nd one does not increase r^2 by much. Can you explain why?
|
|
| Apply | |
| While this class will be more conceptually based, we will be discussing why you would apply a regression model to your experiment. |
3/18 (Answers due 3/17 Noon)
| Read | |
Optional: If you want a simpler introduction to ANOVA you should also read: Chapter 13, Onlinestatbook.com - Analysis of Variance |
Write |
(If you struggle to answer these questions, please read onlinestatbook.com Chapter 13 for further clarification!) 1. Why would you use ANOVA or a TTest? Why would you use ANOVA over regression? 2. Cohen poses the question on page 107: "Why not skip the ANOVA and proceed directly to the t-tests?" Why shouldn't you use multiple t-tests in place of ANOVA? 3. Give an example of when you would use one-way ANOVA in your research? Clearly specify the dependent variable and factors. 4. Give an example of when you would use two-way ANOVA in your research? Clearly specify the dependent variable and all factors. 5. In your two-way ANOVA would you expect any interaction effects? Why? 6. Does Murphy's methods stand up to the 4 "sins" pointed out in Cairns article: reporting, checking assumptions, over-testing, using inappropriate tests? Why?
|
|
| Apply | |
| We will be discussing the included articles for 30 minutes and we will be discussing your research in terms of ANOVA for the last hour. |
Staff | Description | Announcements | Assignments| Syllabus | Policies | Schedule
(subject to adjustment)
Course Objectives:
1. In order to design effective experiments and be critical consumers of research, investigators need to be able to generate relevant hypotheses, select appropriate research designs, compute proper statistics, and effectively communicate findings. By the end of the course, students will:
2. Have a thorough understanding of a variety of common research principles, designs, and statistical approaches employed in behavioral research;
3. Be able to address assumptions and limitations of various research designs and statistical approaches;
4. Understand how to formulate and apply hypothesis testing;
5. Get exposure to both qualitative and quantitative analyses. Methods will include ethnography, interviews, surveys, power analysis, effect size estimates, correlation, regression, analysis of variance, repeated measures, and times series analysis; and
Communicate statistical results, both written and orally.
Motivation:
For a creative concept to gain traction it must be validated. We intend to introduce students with little or no prior training in behavioral research methods and statistics to evaluate whether a technology, idea, or experimental outcome differs from what would be observed randomly. We also strive to provide tools to better understand the assumptions and limitations of current research. Finally, we hope to promote a learning approach to research design and statistics in line with the Media Lab way of thinking: by drawing examples from real data and providing students with opportunities to learn concepts using their own technology and data.
Guiding Structure:
Following the National Science Foundation’ s (NSF) Guidelines for Assessment and Instruction in Statistics Education College Report 2005, we hope to:
Class Structure: Each class will consist of 3 components (approximately 1 hour each):
Guest Speaker
When applicable, we will invite guest speakers who have expertise in each week’s topic area to provide insight on relevant research designs, statistical methods, and common errors in their everyday work.
Discussion
Each class will involve a discussion of the readings and homework assignments, including addressing any questions students have. We will also highlight how the week’s concepts have been applied to real data, and review common mistakes and/or biases when applying these methods.
Application
We will take student’s research projects from class and discuss their work in the context of current readings and guest lectures.
Required Texts (additional weekly readings are noted in the schedule):
Cohen & Lea (2004). Essentials of Statistics for the Social and Behavioral Sciences.
New Jersey: Wiley.
Kline (2009). Becoming a Behavioral Science Researcher: A Guide to Producing Research That
Matters. New York: Guilford Press.
Recommended Texts:
Shadish, Cook, & Campbell (2002). Experimental and Quasi-Experimental Designs for Generalized
Causal Inference. New York: Houghton Mifflin.
Online Resources:
Lane, D., Lu, J., Peres, C &. Zitek, E (2008). Online Statistics: An Interactive Multimedia Course of
Study. http://onlinestatbook.com/
Staff | Description | Announcements | Assignments | Syllabus | Policies | Schedule
Grades:
Weekly reading assignments 25%
Weekly application assignments 25%
Class participation 50%
Total 100%
Academic Honesty:
We expect the highest standards of academic integrity, in keeping with MIT’s policy: "Fundamental to the principle of independent learning and professional growth is the requirement of honesty and integrity in conduct of one's academic and nonacademic life….Cheating, plagiarism, unauthorized collaboration, and other forms of academic dishonesty are considered serious offenses for which disciplinary penalties can be imposed.” All referenced work should be appropriately cited using American Psychological Association (APA) format. Students should feel free to contact course instructors with any questions or concerns regarding these policies.
Weekly Assignments:
In order to participate in class discussions, you are responsible for completing readings prior to each class meeting. To ensure productive discussion, we also ask that you critically apply reading topics to your areas of research interest, and complete assigned homework. All homework assignments, unless otherwise noted, are due by 12:00pm (noon) the Wednesday before class.
February 4
Introduction, Course Overview, Getting to Know Each Other
February 11
No Class. Please read and prepare for next class meeting.
February 18
Basic Research Design Principles and Statistics Overview
Readings:
February 25
Qualitative Methods
Readings:
Guest Speaker: Karen Brennan, MIT Media Lab
March 4
Power Analysis and Effect Size Estimation
Readings: TBD
Guest Speaker: Dr. Joe Rossi, University of Rhode Island
March 11
Correlation and Multiple Regression
Readings:
Guest Speaker: TBD
March 18
Analysis of Variance
Readings:
Cohen & Lea (2004), chapter 5
Guest Speaker: Dr. Kristopher Thornburg
March 25
Spring Break
April 1
Repeated Measures and Time Series Analysis
Readings:
Guest Speaker: Dr. Wayne Velicer, University of Rhode Island
April 8
In Depth Construction of your Research Design
Readings: