Please enable JavaScript.
Coggle requires JavaScript to display documents.
Research methods - Coggle Diagram
Research methods
data analysis
-
-
meta analysis
-
evaluation
large sample, high validity. publication bias , file drawer problem.
-
-
-
Experiments
experimental method
-
Hypothesis: Testable , operationalised statement.
Directional or non directional: identifying a difference/ correlation or not. the choice depends on previous theory or reserch.
variables
Iv's and Dv's: Iv manipulated, Dv measured.
-
-
research issues
-
-
Demand characteristics: Participants second guess the aims, and change their behaviour as a result.
-
-
-
experimental designs
-
-
Matched pairs: similar participants paired on participant variables, allocated to different conditions.
evaluation
independent groups: participants variables aren't controlled (use random allocation). less economical. no order effects.
Repeated measures: order effects (use counterbalancing), demand characteristics. participants variables controlled. more economical.
-
types of experiment
-
-
-
Qazi experiment: Iv based on an existing difference between people , the effect of the Dv is recorded.
evaluation
Lab experiment: high internal validity ( control over CVs/EVs, cause and effect established) replication more easy (support for findings) . however low external validity (generalisability, mundane realism) , low internal validity (demand characteristics).
Field experiment: Higher external validity (more authentic, realism) . lower internal validity (less control). ethical issues (consent not possible).
natural experiment: only option for practical/ethical reasons, high external validity (real world problems). limited opportunities, no random allocation (Cv's), low realism in a lab, not manipulation of IV (cant claim cause and effect).
Quasi experiment: if in a lab, issues for a lab experiment. no random allocation, no manipulation of the Iv (cant claim cause and effect).
the sign test
-
the sign test: difference, repeated measures, nominal data.
the concept of probability: likelihood that the sample could occur if the null hypothesis is true. usually use a significance level of 5%, sometimes 1% for precise studies.
-
steps: 1) convert to nominal data if needed. 2) add up pluses and minuses.3) s= less frequent sign. 4) compare calculated value of S with critical value . if S is smaller than the critical value , the difference is significant.
self report
questionnaires
-
-
evaluation
questionnaires can be distributed to many people. fixed choice , easy to analyse. social desirability response bias.
closed and open questions: quantitative or qualitative data, affects ease of analysis.
interviews:
structured interviews : pre set questions, fixed order , face to face.
unstructured interviews: no formula , just general topic. questions based on responses.
-
evaluation
-
unstructured interview: Flexible. increased interviewer bias. Analysis more difficult. social desirability bias reduced by rapport.
Designing self-report
questionnairs: likert scale, rating scale , fixed choice.
interviews: Standardised schedule, avoids interviewer bias , comfortable setting for rapport. ethical issues.
-
ethical issues
-
how to deal with it
informed consent: signed consent form. presumptive , prior general , retrospective.
deception/ protection from harm: debriefing , right to withdraw /withhold data, counselling.
Privacy and confidentiality: use numbers and not names, Data not shared with other researchers.
observation
types of observation:
naturalistic observation: behaviour observed where it would normally occur. no control over variables.
-
-
-
evaluation
-
naturalistic obervation: low internal validity (control difficult). high external validity (every day life)
controlled observations: high internal validity- extraneous variables may be controlled . low external validity (except if covert).
covert and overt observation: covert - low demand characteristics, ethically questionable.
overt: behaviour may be effected
participant and non-participant: participant- increased external validity but may lose objectivity.
non-participant- more objectivity (increased internal validity) less insight.
observational design:
-
-
sampling methods: continuous. event sampling: count events.
time sampling: count events at timed intervals.
evaluation
structured vs unstructured : structured - numerical, easier to analyse. unstructured - may just be eye- catching information, qualitative data harder to analyse. observer bias.
-
sampling methods: event - useful for infrequent behaviour, misses complexity.
time- less effort but may not represent whole behaviours.
peer reveiw
aims: 1) allocate funding 2) validate the quality of research 3) suggest amendments or improvements.
evaluation
-
publication bias: file drawer problem, creates false impression of current knowledge.
-
sampling
population sapling
random sample: Equal chance of selection, like a lottery.
-
-
-
-
evaluation
random sample : potentially unbiased, control Cv's/Ev's . time consuming , may not work.
systematic sample: objective method , but time consuming , those selected may refuse ( = volunteer sample)
-
-
volunteer sample: easy and participants engaged. volunteer bias, responsive to cues.
-
correlations
the method
-
differences between correlations and experiments: no manipulation of variables, no cause and effect.
evaluation
strengths: useful starting point. quick and economical , using secondary data.
-
-
mathematical content
what you need to know:
persentages and froactions: out of 100, part of a whole
decimal places : digets to the right of the decimal point 10ths, 1000ths etc.
-
-
-
-
mathematical symbols =,= ,<,>,<<,>> etc.