2 (2) Research Design Flashcards
What are surveys?
Measure variables (questionnaires, interviews) and then examine relationships between different variables.
Pros of surveys
- Most effective data gathering technique (internet)
- Can address a broad range of questions
- can address 1+ research questions w 1 survey
Cons of surveys
- inferring cause + effect diff ➡️ correlation ❌ causational
- subject to bias
What is the biases responding to measures and questionnaires?
1 SATISFICING: giving a satisfactory answer w/o properly considering the question. Answering how u think someone want you to.
2 SOCIAL DESIRABILITY (& ‘faking good’): changing answers to appear better.
3 DEVIATION (& ‘faking bad’): answering to appear to be in a worse condition ➡️ don’t get discharged, or get ++ services
4 ACQUIESCENCE: subconscious. Tendency to give same/positive answers ➡️ work around by switching qu orders.
5 CENTRAL TENDENCY vs EXTREME RESPONSE: tendency to choose middle response option or extreme one. Get rid of middle option.
6 HALO EFFECT: feelings on whole vs specific. Person may base response on overall judgement about a person or questionnaire rather than specific items.
What is field research?
It’s another type of qualitative research. It resembles natural inquisitiveness. It takes place in the field, and a range of methods are used:
- Observation (participant, non-participant)
- Unstructured interviewing
- Drawing from documents to build up a picture
It’s good for the study of issues/phenomena we know very little about; it’s also good for classroom situations.
List three experimental designs.
- Between subjects or independent groups
- Match pairs
- Within subjects or repeated measures
What are independent groups?
Several groups differ by a variable:
- Different therapies
- Different diagnostic groups
- Environmental factors (mainstream versus special school)
What are matched pairs design?
Individuals are matched based on factors such as gender, age, education. They tend to have the same level of skill. An independent variable is then applied to them.
What are repeated measures design?
All factors of the independent variable are applied to all participants.
Use the same group but may change the independent variable such as time. Basically, you are looking at and comparing results within the same group.
Pros and cons of repeated measures design
Use if you can:
✔️ needs few participants.
✔️ The variability of subjects is non-issue because we compare subjects with themselves.
❌ must counterbalance to avoid fatigue and learning effects.
What’s an experimental design?
We manipulate the independent variables; we choose subjects, intervention etc.
We measure the dependent variable.
Try to control other [extraneous] variables.
What could increase our confidence that a different exists (in our results).
- The difference between means increases
- The variability between subjects decreases
- Get the same results with more subjects
What a confounding variables?
Variables that that can affect the dependent variable. Need to be controlled.
- Situation variables
- Subject variables
- Habituation effects
Situational variables
inconsistencies in the way the experiment is carried out. Same condition/instructions carried out.
Subject variables
Individual variables; assigned to groups randomly
Habituation effects
Learning effect/fatigue
- counterbalance
Two different time lengths of assessments
CROSS SECTIONAL: subjects are test at a single point in time. Use a single group. Two more independent groups. Easier.
LONGITUDINAL: repeated measures. Following the same individuals over time. Very powerful design. Helps to see development or change.
Why use single or small case studies?
Great for when a new technique of behaviour has been reported.
When something is not predicted has happened.
When data is very detailed and in-depth.
What is nonprobability sampling?
Chances are selecting any particular case from the population are unknown. Population undefined, laws of probability not defined.
What’s probability sampling?
All cases in the population have a known probability of being included. You know to which population the sample may generalise.
Subsets of probability sampling?
Simple random sampling: pps are randomly selected from population; every person has an equal chance.
Stratified sampling: divide population in mutually exclusive groups (strata) and sample from each e.g.socio-economic status
Cluster sampling: sample from natural groupings
Systematic sampling: sample every nth person from population
Types of non-probability sampling
Convenience sampling: collect sample (when possible), okay for small studies, preliminary studies, large is better.
Purposive something: the accepted standard in qualitative methodologies. Requires expert knowledge of target population to identify cases that are representative of the population.