Questionnaires

Questionnaires

1. A form of self report data, which can be used in an experiment to asses the DV.

   • Open ended or Closed
     • Whilst open questions might provide more detail (with less chance for ambiguity) they are difficult to quantify.
   • Qualitative or Quantitative
   • Four elements to conducting a questionnaire
     • DISCRIMINATION- measures should be able to differentiate between individuals
       • more sensitivity if you have a scale rather than a simple 'yes' or 'no'
     • STANDARDISATION- everyone has to have the same experience, bias needs to be eliminated against groups of people
     • RELIABILITY- questions need to be externally reliable
       • Internal/ External
         • Methods for checking internal reliability
           • Split half- where you divide items in half, scores on each half should be comparable. Correlate individuals' scores on each half.
           • Alpha- Cronbach's alpha is an average of every possible half of the times correlated with every possible other half of the items. gives a better overall picture
           • Good reliability has a correlation of between 0.75- 1.
       • e.g. stability across time, measures should consistently produce the same results for the same people across time
     • VALIDITY- methods for checking external reliability
       • e.g. Test- retest. to correlate the scores later on in time
       • Methods for checking Validity
         • Face validity- do questions look like they are measuring the construct. This is simply achieved by 'eyeballing' the measuring instrument or by passing it to an expert to check.
         • Content validity- ask an expert in the field as to whether they think that the questionnaire neglects elements of the topic under investigation.
         • Criterion validity- applies questionnaire to a group who are known to have the characteristics under investigation.
           • Concurrent validity- whether findings are similar to those on a well established test.
           • Predictive validity- can measure predict of an outcome
           • Construct- long term process where one tries to show that proposed hypothetical construct measured in the questionnaire exists. Requires much experimental work.
   • What design is used for a questionnaire?
     • No set design
     • It is simply a tool to measure responses
     • You could have an experimental design and use a questionnaire to measure the DV

2. CONSTRUCTION ISSUES

   • Complexity- if the questions are too complex respondents may become confused and responses will be compromised
   • Technical- avoid specialised terminology
   • Ambiguity- questions should not have multiple interpretations
   • Double- barrelled answers- can be confusing
   • Double negative- can be confusing
   • Emotive language- can be leading
   • Leading questions- that encourage a particular response
   • Invasion of privacy- unless fundamental for the study, avoid very private questions
   • Balance of scale items- for a likert scale you should not have four positives and one negative option
   • Sensitivity of response format- dichotomous options versus a continuous scale

3. Demand characteristics and social desirability

   • Participants may provide the responses that they think you want to hear
   • Or with self report, participants may provide social favourable responses
   • Can imbed questions within the questionnaire that would asses social desirability
   • Or participants social desirability can be measured in a separate questionnaire

4. Factor analysis

   • Statistical procedure that identifies hidden factors (latent variables in the questionnaire) = factor analysis
   • Responses to items within the questionnaire which measure the same construct should be related
   • Factor analysis will identify 'clusters' of items/questions that are closely related: forming 'factors'
   • Is a statistical method used to describe variability among observed, correlated variables in terms of a potentially lower number of unobserved variables called factors