In-class notes for 02/27/2020

CS 263 (ESD), Spring 2020

• Quiz

• Brief presentations next time
  ORC exercise to appear

• Surveys

Brief team presentations

• 5 minute team presentations on your project's progress so far.

• Each team member should speak.

• Use slides (to be submitted)

Survey method

Surveys may be one of the first research methods people think of, but writing/editing a good survey is laborious.
The payoff is better and more useful data.

Identifying topics - what you want to know

Make a list of topics for survey questions before trying to write the questions.

• Be specific.
  For example, "Find students' opinion of the SIS" may be too broad; topics "Find students' opinion of privacy in the SIS" and "Find students' opinion of ease of use in the SIS", or other questions with narrower context, would give more useful information.

• Reconsider the topics before trying to write questions.
  For example, troubleshoot to see if a topic might be misinterpreted or too complex to ask effectively.

  • It's more efficient to refine topics before proceeding to questions, than to waste effort editing questions you won't use.

• Note that some topics may require multiple questions.

  • For example, a complex topic may require multiple questions to get the respondents in the right frame of mind.

Advice on writing and refining questions

The reading gives helpful guidelines for writing and editing the questions themselves.

• Questions should be constructed to be as clear, simple, specific and relevant for the study’s research aims as possible;

• Questions should focus on current attitudes and very recent behaviour;

• More general questions should precede more specific questions;

• Vague quantifiers such as “frequently”, “usually”, and “regularly” should be avoided. Instead, carefully pre-tested response options should specify the number of times per appropriate period (e.g. day, week, month, year) of an event or behaviour;

• A desirable Likert-type response scale length ranges from 5 to 8 response options;

• The inclusion of a middle option increases the validity and reliability of a response scale slightly;

• The numerical scale should be unipolar with matching verbal labels as anchors at both ends of the scale;

• All numeric labels should be shown to respondents;

• “Extremely” and “not at all” can serve as most effective verbal intensifiers;

• Numeric and verbal anchors (=endpoints) should be mentioned explicitly;

• A “don’t know” option should be recorded if volunteered whereby interview instructions should be such that interviewers are not to encourage respondents to choose a substantive response options if they hesitate;

• Demographics questions should be put at the end of the questionnaire.

< >