Research Methods lecture

This lecture is a standard lecture for all sections of the course I'm taking.  Frankly, it seems pretty dry.  In an attempt to focus myself, I'm going to summarize the sections here.


Surveys:  3 components:  1) sample (representative of larger population), 2) standardized questionnaire (clear, unambiguous), 3) method (self-completed, telephone, in person, online, etc.) 

Guidelines for formulating good questions: 

• Consider the Form:  Will you use statements, questions, or both?  If questions are used, will they be open- or closed-ended?  Both types provide challenges.
• Questions must be clear and unambiguous.  They should also be relevant to the users' knowledge--in other words, the people taking your survey should know enough to understand your questions.
• Keep it short and clear.
• Avoid negative statements and bias statements ("don't you think").
• Provide clear instructions in the introductory comments (and perhaps before specific questions).
• Pre-test it and correct any ambiguities or biases.

Send a cover letter to introduce respondents to the survey topic and obtain informed consent.  Parts of the cover letter should include:

• brief statement informing respondent that study contains research and a simple description of the project,
• name and phone number of principle researcher,
• clear statement of research purpose,
• clear instructions on procedures to be followed,
• risks/benefits/costs to participants when applicable,
• confidentiality guidelines,
• study is voluntary,
• for minors, informed consent note for parents/guardians.

Experimental and Quasi-Experimental Design:   focus on examining the effect of a specific action.  Can be controlled (ie, in a "lab" with something being done to subjects) or natural (ie, reactions after an event, such as a presidential assassination or natural disaster). 

Key components of classical experiment:

• Independent and Dependent Variables.  Independent variable can be present or not present.  Researchers observe what happens when the Independent Variable is present or not.
• Pre-testing and Post-testing.  The dependent variable is measured before the independent variable is introduced (or removed) and after.
• Experimental and Control group.  Populations of each group should be similar.

 However, this can be difficult in social experiments, so we have Quasi-Experimental Design:

• Time series design:  looks at behaviors over time, such as speeding rates before and after a new traffic light is installed.
• Non-equivilant control group: control group is similar to experimental group, but not determined by random assignment.
• Multiple time series design:  looks at behaviors over time in several areas or with several groups, in effect creating a non-random control group.

I am falling asleep as I attempt to download the readings in the lecture, so I will continue another time.