Friday, March 11, 2011

De Vaus - Chapter 7: Constructing Questionnaires


 

Selecting Areas


-          Can't go back to subjects and ask additional questions

-          Think ahead about research problem

-          research problem à concepts to be measured.

-          What concepts mean

-          indicators devised for concepts / how variables are links

-          How we will analyse

 

Question Content


-          Five types

o   Behaviour – what people do

§  Difficult to extrapolate from behaviours to beliefs and attitudes

o   Beliefs – what's true / false

o   Knowledge – establish accuracy of beliefs

o   Attitudes – what they think is desirable

o   Attributes – characteristics – eg, age, gender

-          Direction, extremity and intensity of attitudes

o   Direction – question

o   How extreme – eg, likert scale

o   Intensity

 

Principles of question design

 

-          Reliability

o   Question should be answered same way on different occasions for same person

o   Avoid ambiguity etc

-          Validity

o   Measures what we think it does / measures what it is intended to do

-          Discrimination

o   Sufficient response alternatives

o   Need finer grain

o   Sensitive to measuring real & meaningful differences in sample

-          Response rate

o   Content

o   Construction

o   Method of administration

o   Length

-          Same meaning for all respondents

-          Relevance

 

Wording questions


-          Is language simple

-          Can question be shortened

-          Is question double barrelled – more than one question

-          Is question leading

-          Is question negative

-          Is respondent likely to have necessary knowledge – first filter

-          Will words have same meaning for everyone

-          Is there prestige bias – eg asking about pope's view on birth control / attitudes about pope vs attitudes about bc.

-          Is question ambiguous

-          Is question too precise

-          Is frame of reference for question sufficiently clear

o   Eg, time frame

-          Does question artificially create opinions

o   Where people don't have existing opinion

-          Is personal or impersonal wording preferable

o   How they feel about something

o   How people feel about something

-          Is question wording unnecessarily detailed or objectionable

o   Eg, precise income vs income band

-          Does question have dangling alternatives

o   Would you say that it is frequently, sometime, realy or never that you ....

-          Does question contain gratuitous qualifiers

o   Do you favour or oppose increasing number of uni places even if it leads to a decline in standards

-          Is question a dead giveaway

o   Uses all inclusive or exclusive words

o   These words allow no exception

 

 

Selecting Question Type

 

-          Open and closed formats

-          Closed / forced choice

o   Can create false opinions if not enough choice allowed

o   Speed up completion

o   Easier to code

o   Do not discriminate against less articulate

o   Useful prompts

-          Response formats and level of measurement

o   Level of measurement of variable

§  Based on response categories of variable

§  Relation of one response category to another

o   Interval level preferred à allows more powerful analysis techniques

o   choice of statistical methods

o   responses to similar question – can be nominal / ordinal / interval à depending on how question worded

 

Developing Question Responses

 

-           Exhaustiveness (or inclusiveness)

-          Exclusiveness – one and only one answer  à eg ask for most important / split into multiple questions

-          Balancing categories

o   Imbalance of responses in one direction

 

Developing Response Alternatives for Closed Choice Questions

 

-          Numerical rating scales

o   Likert scales

§  statement reflecting particular attitude / opinion : agreement / disagreement

o   Horizontal rating scales

§  Opposite attitude positions

§  Where along continum their position is

o   Semantic differential

§  Opposite adjectives – rather than attitude positions

o   Vertical rating ladder

§  Rank – eg, rank universities

-          Scores

o   Out of 10

o   Feeling thermometer

 

-          Common points

o   one and only one answer

o   responses high to low

o   way in which each item answered is not constrained by way in which other items in set are answered

 

 

-          Ranking


o   Rate importance / strength of agreement relative to the way other items in set have been rated.

o   relative (rather than absolute) importance

 

-          Checklists


o   only select items that apply

-          Binary choice formats


o   Dichotomous questions

o   Paired comparisons

-          Multiple choice formats


o   Multiple nominal categories – eg marital status

o   Ordinal categories – how often do you go to church

o   Ordered attitude statements – degree of belief about god

o   Numerical answers - income

 

 

Non Committal Answers

 

-          No opinion and don't know responses

-          Laziness

-          Inclusion of middle alternative

o   Neither for nor against

 

Number of Response Categories

 

-          Dichotomies

-          Five point scales

-          Longer scales

 

 

Response Sets

 

-          Aquiensence

o   tendency to agree with statement regardless of content

-          Social desirability

o   Look good

o   Techniques to reduce

§  Everybody does it

§  Use an authority

§  Build in excuse

§  Ask less specific question

-          Aquiescent response sets

 

 

Questionnaire layout

 

o   Answering procedures

o   Filter or Contingency questions

o   Instructions

§  General

§  Section

§  Question

§  Go to

o   Use of space

o   Order of questions

§  Easy to more difficult

§  Concrete to abstract

§  demographic at end

§  meaningful sections

o   Setting up for coding

o   Questionnaire length

-          Telephone questionnaires – additional considerations

o   Number of response options – memory

o   Layout

-          Pilot testing – evaluating questions and questionnaires

o   Question development declared or participating pre test

o   Questionnaire development

o   Polishing pilot test

-          Pilot Testing Items

o   Variation – mostly similiar answers – not as useful for analysis

o   Meaning

o   Redundancy

o   Scalability

o   Non-response

o   Acquiensent response set

§  See if people agree with 2 opposite questions

-          Pilot Testing Questionnaires

o   Flow

o   Question skips

o   Timing

o   Respondent interest and attention

-          How to pilot test

o   Who to pre test

o   How many

o   Who to conduct interviews

o   Code responses

§  Open ended

§  Other

o   Interviewer debriefing

-          Questionnaire Design Checklist

 

 

 

 

 

 

 

No comments: