L7 21/02 Flashcards
What is a survey?
A non-experimental quantitative research design. It is a study design that may use a questionnaire.
What form can a survey adopt?
A questionnaire or based on another means of gathering data from people
Why is a survey quantitative?
- Each participant has the same questions (the process does not change – it remains the same)
- It produces numbers that can be subjected to statistical analysis
- We can assign numbers to people answers by manipulating data
Open questions elicit what kind of data?
Qualitative (textual)
Limitation of survey
Often misused in health research
Useful survey applications in healthcare
- Service evaluation – good or bad within healthcare
- Patient experience/opinion studies
- Service redesign
5 things to maximise survey validity
- Use a clear research question – survey focused on research question
- Identify the population and a relevant, representative sample
- Choose a method of distribution to maximise response
- Plan the coding method: Variables labels for categorised answers
- Test the survey tool: Expert consensus groups, piloting
Survey advantages
- Simple measure of attitudes, beliefs, values, understandings
- Easy access to most human populations
- Can address sensitive topics – anonymous
- Effective at collecting large amounts of information
- Collects highly standardised data – provided not asking lots of open questions
- Efficient; cheap and non-resource-intensive
Survey challenges
- Response/completion rates are poor > around 30-40%: Particular issue with postal surveys, may need for ‘reminders’, must be short to ensure completion rate, small amount of data gathered
- No researcher control over sample: possible bias with who chooses to participate
- No control over data quality or accuracy: misunderstandings can affect accuracy
- Responses have to be retrospectively coded
Demographic data (background information about participants)
- Gender
- Age Group
- Ethnic group
- Height, weight etc.
- Length of time having health conditions
In order to maximise accuracy, data quality (and participation) closed response questions need to be what 4 things?
- Accurate (clearly linked to the point of the survey)
- Exhaustive (all possible options are covered)
- Mutually exclusive (so that only one answer is accurate)
- One dimensional (so that only one answer is possible)
What are the 5 question types?
- Fixed Alternative Responses / Multiple Choice Questions (MCQ)
- Likert scales (agree, strongly agree etc)
- Multiple responses
- Filter questions (yes/no and follow up if answered yes)
- Open questions
Other design considerations
- Document appearance
- Aesthetic and layout
- Use of white space
- Clarity of instructions for completion
- Grouping of questions into sections
What is coding?
Coding is the practice of assigning a numerical value to fixed response alternatives so that it can be easily managed.
Advantages of coding
- Useful in descriptive or evaluative designs
- Easy way to study attitudes, values, beliefs
- Useful in almost all populations
- Standardised data
- Gather large amounts of data
- Efficient
- Anonymous
Limitations of coding
- Potential for response biases
- Low response rates, particularly in postal surveys
- Risk of misunderstanding
- Incomplete responses
- Potential inaccuracy
- Often badly designed
Health designs often used validated scales such as..
- Hospital anxiety and depression (HAD) scale
- General Health Questionnaire (GHQ)
Indicators of validity are…
- Focussed research question and purpose
- Clear identification of population and sample
- Piloting and revising the survey tool and coding method
- Accuracy, ease of use, relevance to aims of survey
- Logical and appropriate recruitment method
- Reliable data collection method (Self report; Face to face; Postal distribution; E-survey)
- Sound data analysis methods (ideally conducted by 2 independent researchers0
Evaluating the survey tool
- Check that the survey tool relates to the research question
- Check what demographic data has been collected (NOT intrusive and always relevant to the issue)
- Consider if a previously validated tool has been used; if not then “why not”?
- Has the tool been piloted or validated
Evaluating Sampling and Recruitment
- Population and sample should be related to the question
- Sampling technique should be related to the aims and NOT just what is easiest to achieve
- Is a random sample needed or will convenience sampling suffice?
- Recruitment will always be related to the sampling strategy and invitations to participate
- Issues surrounding sampling and recruitment should be discussed in the write-up
Data collection methods (4)
- Self report
- Postal
- E-survey
- Face to face
Evaluation of data analysis
- Analysis should use the data to answer the research question(s)
- Check for rigorous coding (preferably repeated by independent researcher)
- Data should be entered into a data management system (Excel, SPSS etc.)
- Descriptive statistics should describe the data (look for diagrams to clarify)
- Analysis should identify patterns
- Inferential statistics may be used; check their validity
- Interpretation should relate to the survey aims
- Beware of “fishing” techniques
Advantages of surveys
- Useful in exploratory, descriptive or evaluative research designs
- Good for measuring attitudes, beliefs, values, understandings
- Can potentially access a wide population
- Useful for sensitive topics
- Provides anonymity
- Only appropriate if data can be standardised
Disadvantages of surveys
- No researcher control over sample
- No control over data quality or accuracy
- Incomplete responses/data sets
- Risk of misunderstandings or ambiguities
- No opportunity for clarification (except face to face)
- Poor response rates