Ch 7: Effective Measuring Flashcards
(28 cards)
Ordinal example
Example: A new online grocery store called yuorlocalgrocer.com is concerned about its inability to bring back repeat customers. Most customers use the site one time, with only 7% of customers being returning users. As such, the business decides to incoporate a series of items evaluating the customer experience using Likert-type items. For instance, they ask users how much they agree or disagree using a 7-point scale with the usability of the site, the prices, and the convenience.
What type of measurement?
Self-report
The most common approach to measurement involves asking questions. Measures that rely on study participants to provide answers
ex: how many min do you spend a day on using facebook, linkdin and twitter
Others’ reports
As an alternative to self-reports, you might ask questions of people other than the ones you are studying.
ex: parents reporting on childs behavior
Rapport
A positive connection and interaction between the researcher and the research participant
Use participants’ names, share goals, build trust.
Example: Using this to improve response rates in panel studies.
Questionnaire Structure
All survey research involves presenting participants with a standard list of questions, known as a questionnaire or survey instrument.
Questionnaire Design
How a questionnaire is organized can have implications both for the quality of the data and for the likelihood of completion.
indicate clear ways to answer (1-5) and include a period of time (last 3 months)
Creating an effective questionnaire
-clear directions
-start strong
-avoid response set (reverse items and restate directions)
-Leave space (espcially for open-ended questions)
-pre-test
-Proofread
-Streamline and organize
-End strong (thank them for time)
Filter questions
used to direct participants to only those questions that are relevant to them
ex: do you use facebook? if yes then direct them to pertaining questions
Contingency Questions
are the subsequent questions that are answered by only a portion of the participants
Response set
Patterns in responses where participants select similar options (e.g., always “agree”).
Acquiescence bias (agreeing with all statements).
Extreme responding (only choosing endpoints like “strongly agree” or “strongly disagree”).
solution: use reverse-coded questions to disrupt patterns
ex: mix positive and negative statements about satisfaction
Order effects
Influence of question order on responses.
Respondents’ opinions on “leadership” may change if asked after questions on “management failures.”
open-ended questions
limited set of responses (Categorical, Likert, Semantic Differential)
Must be:
1. Exhaustive
2. Mutually exclusive
ex: Did you experience good customer service?
Strongly agree, etc…
open-ended questions
No predefined answers., Allow participants to elaborate.
Usually begin with words like “how,” “why,” or “what.”
ex: How was your customer experience today?
Exhaustive
if the response choices include all possible answers or options for a given item (directly tied to close ended questions)
mutually exclusive
two or more categories cannot be true at the same time
ex: what is your highest level of education?
Nominal level
labels variable categories (indicates whether cases are the same or different)
ex: what is your class standing at UCSB? What is your zip code?
Ordinal Level
Numbers indicate a rank order (data points can be ordered or ranked, 1st, 2nd, 3rd but the exact difference is not known
What is your highest level of education?
-high school, associates, some college, bachelors,etc
Interval level
numbers indicate distances between cases with no meaningful zero or absolute zero point
What year were you born?
Ratio level
all aspects of intervals (meaningful distance) but also include an absolute zero point (allows for meaningful distances between values)
ex: height, income, time
Leading questions
that is, one answer or type of answer appears to be intended or more ideal than others
ex Do you support or oppose the new health care legislation that will hurt middle-class families and destroy jobs?
Social Desirability
Research participants feel pressure to answer items in a particular way based on perceptions about how their answers will reflect on them
Clairity
avoid overly complex items, jargon, highly technical language, vague descriptions, and very long questions.
ex For each of the following items, please indicate how frequently you have engaged in those behaviors during the past week using a scale from not at all to every day
Double-barreled questions
Individual items that try to simultaneously measure more than one indicator, dimension, or concept with a single closed-ended response
ex Do you have a good relationship with your partner and enjoy time together?”
Likert-type items
ask respondants to choose from a range of answers to indicate how they feel about statement or question (common for attitude and perception studies)
ex: I am not able to accept what is happening
(1) totally disagree –> (5) totally agree
