150X 9/30/03

Discussion of article – I did not read. Bloody embarrassing to have to admit. I have personal excuses, but the end result is the same; I am unprepared. Shit.

Observations:

• Sample size small
• Samples not random
• Brought up many questions about what is not known.
• Started to explain methods in introduction
• Gave a practical application for using predictors in actual course
• How recent should a study be to be considered "recent"?
• It can take up to 3 years to get an article published, so research for a 2000 article may have been conducted in 1997… 1994 is not invalid, but if all studies were from the 1960's, especially with such tech content, that would bring out questions.
• Poor explanation of sample population
• Questions are there, but statements of hypotheses are missing. This is not unusual, and you will frequently have to extrapolate the hypothesis by looking at the statistical techniques used. Frequently hypotheses will be stated as they get to each statistical calculation.

Basic issues of Research Design.

One week lesson doesn't even begin to touch the surface. As we read articles, will have to go back to previous articles and notes.

o Role of stats in behavioral research

o Classification of variables

o Quant of variables

Process of Empirical Research

o Hypotheses are a technical way of expressing our expectations about the result.

o Design research study to collect data bearing on questions. This should allow us to make valid, accurate statements (DRAW CONCLUSIONS) about our topic.

o What kind/how many subjects? Large or small group?

o What will we ask subjects to do? Survey? Test? Observe?

§ Just because you are doing observations doesn’t make it qualitative – we know how to count and code observations. Just because you are doing a survey doesn't mean it's only quantitative – we can ask open-ended questions. Our methods of data collection do not define the type of our research; it's what we do with the data that makes it quant or qual.

o How many comparison groups? Who is our control group? How do we define them?

o What dependent/independent variables? Limited resources demand that we narrow the focus of what we study.

o How and when will we measure? In what circumstances/conditions? Can't take snapshot to answer long-term outcome questions.

§ Pre-test sensitization; pretest can let people know what to say/study/look for

§ Subject thoughts about the study

§ Physical setting/conditions

o Where will we conduct study?

o Subjects:

o Where do they come from?

o How sample collected? What sampling methods were used? Does the sampling method bias the outcome?

o How are comparison groups formed?

o Attrition rate – how many subjects actually supplied data?

o Motivations – what motivated subject? How?

o Data

o "Data" is plural. "Datum" is singular. Therefore "The data are accurate" and "this individual datum is suspect."

o Instrument quality

o Question/data match. Are you measuring what you think you are? Is it related to your topic? Are other contributing factors constant?

§ Example: if you measure math ability using word problems and reading ability is not constant, you will not get an accurate result; you will have first measured reading ability, then math, and you won't be able to disaggregate the result.

o Independence of observations

§ Are subjects discussing or collaborating with one another? Are the comparison groups mingling outside the study?

o Who is collecting the data? What is their bias? Does it come through in the measuring instrument? How does that affect the data set?

Article should give you enough information to go out and REPLICATE the study.

If we have crystal clear questions and a poor plan, we get junk. How we collect data influences what we are going to be able to say in the end, and influences the statistical analysis we can use on the data in the end.

Descriptive statistics: Methods used to obtain indices that characterize or summarize data collected

• Motivation, learning style, gender, ethnicity, GPA, major, etc.
• Did they define their terms? (WTF is "cognitive motivation"?)
• About individuals or groups

Inferential statistics: Methods that allow the research to make inference froma set of data collected from a sample to a larger population

• About extrapolation/generalization of sample results to the larger population
• Basic minimum for education research (rule of thumb, not absolute) = 30

Terms:

• Research: A SYSTEMATIC approach to finding answers to questions.
• Research Design: a plan for gathering data for answering specific research questions
• Statistics: The methods we use on the data

Hypothesis: A tentative statement ("educated guess" about the expected relationship between two or more variables.

• State expected relationship between 2 or more
• Be worthy of being tested
• Be testable
• Be brief and clear

What's going on in the stats should tell you what the hypotheses are.

Group:

Nick Chan: Credentialed v. noncredentialed teachers:

Teachers with credentials are doing as well/not as well as teachers with emergency credentials/uncredentialed

Nara:

Problem: CA voters are not well-acquainted with the candidate positions in the CA governor's race

Question: To what extent does regular participation in the political process affect "name-recognition" voting?

Hypothesis: Registered voters with sporadic voting records are more likely to vote for Arnold Schwarzenegger than voters with regular voting records.

Nick Knudsen: Parents' perceptions of the term "moral education," whether they favor having the school involved in it.

Hypothesis: if gave control group survey and gave another group a working definition and a survey, hypothesis: the control group would see it as less favorable that educators involved in "moral education"

Variable: What is measured or varied. An attribute of a characteristic of a person or object that can change form person to person.

• Independent: a variable that is manipulated, measured, or selected by a researcher in order to observe its relation to the subject's response: an antecedent consition.
• Dependent: A variable that is observed and measured in response to an independent variable. Generally the dependent variable is the result we are interested in.
• Control: Any variable that is held constant in a research study by observing only one of its instances or levels.
• Intervening: a hypothetical variable that is not observed directly in the research study but is inferred by the relationship between independent and dependent variable

CORRELATION DOES NOT INDICATE CAUSALITY

Research question: "Why" or "How" does not determine qualitative or quantitative nature of study. The question is neutral; the method of data collection and analysis determine the nature of the study.

Measurement: the application of rules in assigning numbers to cases so as to represent the presence or absence of quantitiy of an attribute

4 scales of measurement:

1. Nominal Scale Measurement (Lowest)
1. Categories that have names: male/female, single/married. coke/pepsi/sprite. Often called "qualitative variables" but not to be confused with qualitative research.
2. Ordinal Scale Measurement
1. Ranked order of things: sizes, weights, populations, relative measurement.
3. Interval Scale Measurement (Continuous measurement)
1. achievement tests – where equal units between numbers mean the same thing. Zero has no absolute meaning (indicating absence of the variable) in interval scale: like temperature, zero degrees does not mean the absence of temperature.
4. Ratio scale measurement (Highest) (Continuous measurement)
1. in a ratio scale, zero has an absolute meaning, like number of children in a family or salary.

Variables measured at higher levels can be scaled down to lower levels.

LIKERT SCALE: Agree/Disagree (5 Question/Item survey)

Scale is continuous.