Summer 2001 - Psych 2311 (Elementary Psychological Statistics) - Dr. Burton
MTWR 11:20-2:15
Kozlowski 350
Office Hours: T 2:30-3:30 or by appointment
Office: Kozlowski 353 - Office Phone: 275-2701
e-mail: burtongr@shu.edu

There will be five quizzes, on July 12, 19, 26, August 1 and August 8. The Final will be on the last day of class, August 9. The quizzes will not take up the entire class period. The best five scores will be entered into the grading formula, below, meaning that there is one drop quiz. Thus, there will be NO make-up tests in this course, nor will there be any extra-credit. For this course, any effort a student is willing to put into extra credit would pay off more in additional studying for the quiz or homework. The final exam (see above) will be cumulative, and will count for about twice the value of a quiz. Students should bring calculators to all classes.

At least 12 homeworks will be assigned. Each homework is worth up to 5 points, minus one for each class session that it is late. There will also be a short take-home assignment due toward the end of the semester which will count for about half the value of a quiz.

Text:

Gravetter, F. J., & Wallnau, L. B. (1999). *Essentials of Statistics for the Behavioral Sciences (3rd ed.)*. Minneapolis, MN: West.

Students are to work on all home-works independently. Since you will encounter numerous statistical techniques this semester; it won't be possible to assign a distinct homework for every single technique you may have to demonstrate on a quiz. Thus, a good strategy would be to do extra homework on your own above and beyond what is turned it officially. There are plenty of examples available in the textbook, and it is also a good study technique to invent your own problems. This is optional, of course, but on each quiz I will ask you to report how many study problems you tried during your preparation for that quiz.

Thus, the grade is based on 4 quizzes, the Final, the take-home, and the best 20 homeworks.

There are three major questions in Statistics:

What should I measure?
What number can best represent my measurement?
What does this number mean?

*** Data**. Any description of the world, from a poem to an experiment, crucially involves deciding which aspects of the world must be included. Any researcher must decide to record some "things" and leave others unattended. If these "things" can take different values, they are variables; i.e., aspects of the world that can take different values and whose values can be measured. Statistics is most useful for those variables that range quantitatively (height, IQ, frequency, temperature, amount of agreement, etc.), in other words, variables whose values can be given ordered numbers that can be combined mathematically. A person who is 2 meters tall is really taller than a person 1.8 meters tall, and she is taller by .2 meters. We can assign numbers to variables that differ only by type, or qualitatively, but we can't do mathematics with these numbers. We could call Psychology Major 1, and Art and Political Science Major 2 and 3, respectively, but this doesn't mean that Art is two times Psychology or that Psychology + Art = Political Science; we may as well just keep the names. Records of the values of variables are data. For example, you might use scores on a math exam as an indicator of whether a new teaching strategy you tried in a class of second-graders has been successful.

*** Distillation of Data**. Even after you have decided to record some variables and not others, you may have to boil down the data collected to make it useful. For example, you may have 20 second-graders in your classroom, but sometimes you will want a single number that stands for all of them (an average is a good example). A statistic like this, that stands for a set of raw numbers, a measure of central tenden-cy. There are many situations where a single number is not very expressive. You may need a pair of numbers, so that you can compare one situation with another. (Perhaps you compare the average math test score of your 20 students to the average of 20 similar students who did not learn under the new method.) Or you may need to see the distribution of numbers, the overall pattern of scores rather than a single number that represents them all. A visual depiction of the behavior of scores is a graph. In graphing, there is more freedom of choice than in other realms of statistics; more judgment and less following of rules is called for. Nevertheless, there are guidelines for constructing graphs that are accurate, honest, and informative.

*** Evaluation of data**. With inferential statistics, the goal is no longer simply to describe the behavior of a group of people or animals, but to infer what changes can be made in an entire population. An experiment can only measure a small number of subjects; you may have 20 participants in your study even though it is the entire human race (4 billion strong) for which you need information. For example, some of the simplest inferential statistics (such as the chi-square test) are appropriate for frequency data. You may find that teachers in a private school pick the new math technique more often than do teachers in a public school. A chi-square test helps you decide whether this discrepancy is more than just a chance occurrence.
**ORDER OF TOPICS**

**Introduction
**

Anecdotal evidence vs. probabilistic evidence
The logic of inferential statistics: the dependent-samples t-test
Measures of tendency
Measures of dispersal
Probable outcomes
Probability [G&W ch. 6; pages 132-138]
Chi-square [G&W 16]
Critical values

**Descriptive statistics
**Types of measurement and variables [G&W 1]
Frequency distributions [2]
Graphing
Measures of central tendency [3]

**Variability
**Measures of dispersal [4]
The Standard Deviation
Z scores [5]
Correlation and regression [15]

**Hypothesis testing for quantitative data
**The normal curve [6]
The null hypothesis [7, 8]
Basic t-test [9]

**More inferential statistics
**T-test variations [10, 11]
Estimation [12]
Analysis of Variance [13, 14]