|
Lecture 7:Questionnaires in Social Research
Structure of the session
Why use questionnaires?Designing questionnaires – issues of theory, variables, reliability and validityImproving your response rates
Why use questionnaires?
Questionnaires allow you to generate considerable amount of data to address your research question(s)They minimise the risk of bias, ensuring greater validity and reliability of dataThe answers generated are quantifiableCost-effective way of generating dataQuestionnaires can be web-based.
Questionnaires are not appropriate for exploratory research – too many open ended questionsThey are more suitable for descriptive or explanatory research You must have well-defined response categories, e.g. ‘Level of employee trust in management’A lotA littleNone at all
Questionnaires are not appropriate for exploratory research – too many open ended questionsThey are more suitable for descriptive or explanatory research You must have well-defined response categories, e.g. ‘Level of employee trust in management’A lotA littleNone at all Skill requirement - As long as the questions are worded correctly, it requires little skill to administer. Interviews on the other hand require a lot more skill on the part of the interviewer. Multi-method approach - It is usually better to combine questionnaires with other methods of data collection, e.g. interviews. Relying solely on one method may not be a good idea. Interviews allow you to ‘fill in the gaps’.Different types of questionnaires – self-administered: online/emails, postal interviewer administered: telephone interviews, structured interviews.
Factors that influence the choice of questionnaire
Time and cost, e.g. structured interviews would be very time consuming and costly. Type of questions you need to ask to collect your data, e.g. if the questions are too complex, standardized questionnaires may not workProfile of your respondents – their literacy levels, access to technology, availability etc.
Self-administered questionnaires – not sure who has answered the questions? This risk is minimized in interviewer-administered questionnaires – improves reliability of your data. But more expensive.Interviewer-administered questionnaires have better response rates compared to self-administered questionnairesContamination of responses – respondents may simply guess answers due to insufficient knowledge or experience.
Deciding what data needs to be collected through questionnaire surveys
You must be clear about your theory. E.g. mobilization theory – workers engage in collective action when a) they perceive an acute sense of injustice, b) there is social cohesion, c) they attribute blame to the employer for their grievances, d) they believe the union can make a difference. Identify the variables you wish to collect data on: e.g. ‘collective action’, ‘injustice’, ‘attribution of blame’ ‘union effectiveness’ etc.. Then formulate your questions accordingly: issue of validity. Asking irrelevant questions will reduce the validity of your data. #p#分页标题#e#
Issues of reliability and validity
Huselid (1995) asks managers to report what % of their annual company sales could be attributed to: a) cost reduction, b) innovation, c) focus. What do you think the problems are with this question?Charlwood (2004) measured UK employees’ propensity to join a trade union. He used the dependent variable as the employee’s current union membership status. Is this the right approach?
Types of data variables in questionnaires
Dillman (2000) differentiates between three types of data variables for questionnaires:Opinions, e.g. what do you think of the coalition government?Behaviour, e.g. how many times in the past 6 months has your branch union held workplace meetings? Attributes, e.g. gender, race, occupation, income etc.Open-ended questions – no predetermined response categories. Responses are coded after data collection. E.g. what are your views on global warming? Scale response categories – e.g. ‘Strongly agree’ to ‘Strongly disagree’ to a particular statement. The issue of ‘Neither agree nor disagree’ category. Ranked responses - e.g. please state the two most important reasons for you to quit your job. Quantity questions - e.g. what is your age, income, length of service...Clarity of questions and instructions – respondents should be able to understand and interpret the questions as intended by the researcherToo long or too short in length? Structured interviews v. self-administered or telephone interviewsSignificance of the research topic to the respondentVisual appeal of your questionnaireTiming of your survey – response rates will be typically lower during Easter or Summer vacations or if there is a major dispute like strike actionIs a web-based questionnaire likely to improve response rates?In addition to the previous factors, the following could also help improve your response rates:A covering letter clearly stating the purpose of your researchAssuring anonymity to respondents Promising access to results when the survey is completedGiving your contact details in case respondents wish to contact you for clarifications or suggestionsFollow-up reminders and questionnairesIncentives, e.g. lucky draws, gift vouchers, or monetary incentivesPre-paid return envelopes for completed questionnaires (problematic if international). Little subtlety or nuance – limited response categories do not capture the emotional response of the respondentsCould result in variations of responses – if respondents interpret the questions differentlyIf low response rates – difficult to chase non-respondents. You can do this in case of structured interviews, but it could be expensive. You must always pilot your questionnaire before using it for actual data collection
|
