P2: Research Methods Flashcards

Question

Describe and explain what makes a good questionnaire

Answer 1

-Avoid Jargon -Avoid Double-Barrelled Q(2Qs in 1) -Avoid Leading Q CLOSED Q: Limited choices(Quantitative data) +Easier to analyse(can make graphs) -Restriction(lack of representation= low validity) OPEN Q: Provide own answers(Qualitative data) +No restriction(Detailed=accurate) -Hard to analyse(maybe forced to reduce data to stats)

Answer 2

- Interview schedule: Standardised list of Q to cover - Quiet Room: More likely to open up - Rapport: Begin with neutral Q= relax participant - Ethics: Reminders= ANS will be treated in confidence

Answer 3

Used in all types of research -Trial Run: small scale trial run done before hand to determine any errors ∴ you can fix before the real thing ∴ save money/time

Answer 4

``` Quantitative: Numerical data +Easy to analyse(make graphs) -Oversimplifies(Individual meanings lost) Qualitative: non-numerical data +Represents complexities(Detailed) -Hard to analyse(hard to summarise) ```

Answer 5

Primary: First hand collect data for study +Bespoke(all relevant data) -Time/effort(money) Secondary: Collected by someone else conducting same study +Cheap(save money/time) -Quality control(outdated/incomplete) Meta-Analysis: Combing data from large NO. of studies +Good Validity(large sample=generalisable) -Publication bias(researcher may not select all relevant studies)

Answer 6

``` MEAN: Arithmetic Average +Sensitive -Unrepresentative(outliers=distorted) MEDIAN: Middle Value +outlier proof -Less sensitive MODE: Most Frequent Value +Relavent to categorical data -too simple ```

Answer 7

Range: difference of biggest/lowest (+1) +Easy to do -Do account of distribution between scores Standard Deviation: AVG spread around mean +More precise -Misleading(hides outliers)

Answer 8

USE CONDITION: Analyse difference between scores for NOMINAL DATA METHOD: Scores for test B - test A = tally total of +/-. -Neutral(e.g.5-5=0)= ignore TALLY up frequency of both signs -'S' VALUE = total of less frequent sign If S is equal to/less than CV ∴ S is significant/ experimental hypothesis is retained

Answer 9

Before publication, all aspects investigated by peers. AIM: validate/suggest improvements/funding +Protects quality(minimises fraudulent research) -Used to critics rivals(∵lack of funding=fight) -Publication bias(Need 4 eye-catchy=remove things) -Ground-breaking study=buried(more critical of studies which contradicts their views)

Answer 10

Represents strength of correlation(+1 or -1) Close to 1= strong, 0= none, +=pos,-=neg SIgn= direction of correlation e.g. +1 = perfect positive

Answer 11

Longitudinal.Can involve gathering data from friends/family. Often involve unusual individuals/events. Can involve concentrating on typical cases.e.g.old guy recalls childhood. DATA collection: Interviews/observations/questionnaires (QUALITATIVE)/Psychological tests(QUANTITATIVE) +Detailed(= more validity) +Enables study of rare behaviour(e.g. HM) -Researcher bias(conclusion subjective) -Participants accounts biased(prone to memory decay)

Answer 12

Indirect study via communications(observational research)e.g. spoken interactions, written forms,examples from media Coding first stage .e.g.counting NO. of times word x appears (QUANTITATIVE DATA) Thematic analysis .e.g.mental health theme reoccurring. (QUALITATIVE DATA) +Ethic(material=public usually(TVS) ∴no consent issue) +Flexible(can produce QUANTITATIVE/QUALITATIVE) -Communications outta context(may add motivations not intended ∴reduce validity of conclusions -Lack Objectivity(bias(specifically descriptive ∴ validity affected)

Answer 13

RELIABILITY (replication with same results) = consistency ASSESSING RELIABILITY -Retest: test same person twice -Inter-observer: Compares different observers watching same observation -Correlation: coefficient should be above +.80=reliable IMPROVING RELIABILITY -Questionnaires: rewrite Questions(e.g.less ambiguous) -Interviews: improve training(no leading Q) + use same interviewer -Standardised Procedures: good control of aspects -Observations: operationalisation of behavioural categories(shouldn't be overlapping)

Answer 14

TYPES OF VALIDITY = is the result legitimate(of the IRL world) Data can be reliable but not valid(e.g.IQ test) Ecological Validity: If findings generalise to other settings Temporal Validity: Do findings remain true over time ASSESSING VALIDITY Face Validity(eyeballing): Does test look like it measures what it should Concurrent Validity: Whether findings= similar to (older)well-established test(correlation should exceed +.80 for validity) IMPROVING VALIDITY -Control groups: allows comparisons ∴ more confident changes in DV due to IV -Questionnaire: lie scale= controls for effects of social desirability bias (assured of confidentiality) -Observation: Categories. Well-defined, operationalised, not overlapping -Qualitative research: Triangulation= using NO. of different sources

Answer 15

-Purpose: to test if results are due to chance 3 CRITERIA FOR CHOOSING STATS TEST: 1) Looking for Difference/Correlation 2) Experimental design related(Repeated/matched pairs)/unrelated(Independent group?) 3) LV of measurement CONDITIONS FOR STATISTICAL TESTS -Nominal data: (Categories) No order, only appear in 1 category -Ordinal data: (Ordered, subjective intervals) Based on subjective opinion VS objective measure.e.g.Rate how much you like x on scale of 1-10 -Internal data: (Units of equal size, based on numerical scales)

Answer 16

Null Hypothesis= no difference/correlation. If stat test is not significant the null hypothesis is accepted. 0=statistical impossibility, 1= statistical certainty USING STATISTICAL TESTS Usual Lv of significance=/less than 0.05(5%) Check for statistical significance. Calculated V compared with CV. -To find CV: 1)find if hypothesis is one-tailed/two-tailed 2)(N)No. ppl, or degree of freedom(df) 3)Lv of significance (p value) TYPE I/TYPE II ERROR -Type I error: Null rejected, when null= true Optimistic error= sig difference found when 1 doesn't exist. Likely = if sig Lv is too lenient -Type II error: Null accepted, when ALT = true Pessimistic error Likely = if sig Lv is too stringent

Answer 17

- Abstract: Summary of the study with all major elements - Introduction: Literature review. look at relevant theories, concepts, studies (that are relevant) - Method:State. Design, Sample, Apparatus/materials, Procedure, Ethics(∴ others can replicate study) - Results: Descriptive(graphs, charts),Inferential(reference of stat test). raw data collected appear in appendix. - Discussion: Summary(findings), relations (comparison research), consideration of limitations/weakness, IRL implications - Referencing: Sources.e.g.journal articles, books, websites

Answer 18

PARADIGMS/PARADIGM SHIFTS Paradigm(scientific disciplines)= shared set of assumptions/methods. Paradigm shifts(scientific revolution):ppl start accept paradigm ∵ too much contradictory evidence to ignore THEORY CONSTRUCTION set of general laws= explain particular events Make clear predictions based on theory. Hypothesis tested ∴determine support/refused Deduction: deriving new hypothesis from existing theory FALSIFIABILITY Proof impossible. Real theories =challengeable= prove false/not Pseudoscience = cannot be challenged ∴ not proved false yet REPLICABILITY Theory=trustable= replicable = extend to which we can generalise OBJECTIVITY researcher must keep critical distance = reduce bias EMPIRICAL METHOD Direct experience= importance of data collection via direct sensory experience.e.g.observational/experimental methods(overt/repeated G)

P2: Research Methods Flashcards

(42 cards)