Research Design Part 2 Flashcards
What are the types of research design?
Experimental primary research designs :
- Correlational
- Experimental
- Quasi Experimental
Cross sectional/ survey
What is correlational research design?
Investigate relationships or measures of association between variables e.g. correlational designs into relationship between smoking and lung
Stats techniques include chi squared
Why can we not assume causation?
Causal relationships but difficult to establish this as we simply observe and record changes in variables
Correlation not causation
What is Experimental Research Design?
Manipulation of the IV to see what effect this has on the DV
Looking for differences between conditions
Participants are randomly allocated
What is a hypothesis?
A prediction of how specific variables might be related to one another
What is a Quasi experimental research design?
Differences on the DV between conditions of the IV
There is no random allocation
Stats techniques include the t-test
What is the structure of an Experimental Design?
Obs - observation made in relation to the DV - pre and post test
Exp - experimental treatment
No exp - absence of exp - control group
T - timing of observations are made in relation to DV
How do we allocate participants?
Within participants - repeated measures on the same people
Between participants design - independent or unrelated
What are the adv and disadv of within?
Adv - Control many inter individual co found variables, don’t need many participants
Disadv - Order effect, but can be counterbalanced by varying order of conditions , demand effect
What is order effect?
Person knows what is happening next so adjusts their performance accordingly or may be fatigued
What is demand effect?
Participants realise the purpose of the experiment and conform to it
What are the adv and disadv of between?
Adv - No order or demand effect
Disadv - Need more people, lose certain degrees of control over cofounding variables
What is a cross sectional (social survey) design?
Collection of data on more than one case at a single point in time in order to collect a body of quantitative data in connection with 2 or more variables that can then be examined for relationships/ patterns
What is the role of surveys?
Experimental design not always feasible
Experiments develop from studies
Experiments have to be done in an artificial environment
Some variables can’t be manipulated
Expensive
What do questionnaires and surveys involve?
Large sample
Self administered by the researcher
Open and closed questions
Reduced bias
Less intrusive
How can we pre-code questionnaires?
Set format closed response
Numbers can be entered or scanned
How do we design the Self Completion Questionnaire?
- Easy to follow
- Identifying response sets in a Likert scale
- Clear instructions
- Keep it short
- Use simple language, word questions well and vary the format
- Ensure questions provide variability
- Use a logo
- Dummy run (test in person)
What are the adv of self completion questionnaires?
- Cheap
- Quick
- No interviewer effect
- Convenience
- no interviewer variability e.g. tone
What are the disadv of self completion questionnaires?
- Can’t probe
- Questionnaire can be read as a whole
- Don’t know who answers, may not answer at all
- hard to ask lots of questions
- low response rate
- may not take research seriously
- may give the desirable answers rather than real
How to improve response rate
- Good covering letter
- Return with clear instructions
- Follow up - 2nd copy may not ensure anonymity
- Must capture the interest of the participant
- Provide an incentive
What should a covering letter include?
- Friendly but short
- Describe why the study is being done
- tailor letter to the audience
- Mention incentives
- describe confidentiality policy
Why might we have missing data?
- Lose the paper it was recorded on
- illegible handwriting
- equipment stops working
- participant doesn’t consent to all measures
What is a pilot study?
Tests measures you are using, identifies the feasibility
Important with humans as they are unpredictable
Can have peers comment on the questionnaire
What are the guidelines for creating data for analysis?
- Decide what variables you need and document them (Data dictionary)
- Design your data set with one subject per line
- Name variables - short, no duplicates or special characters
- Descriptive labels for variables
- Type for variables
- Tips for categorical variables
- Define missing value codes - 999
- Consider the need for a grouping variable
