Research Methods

Aims - The purpose of the investigation.

Hypotheses - The formulation of a testable statement.

Directional or non-directional - Identifying a difference/correlation or not. One-tailed and two-tailed predictions.

Variables

IVs and DVs - IV is manipulated, DV is measured.

Levels of the IV - Experimental and control conditions.

Operationalisation - 'De-fuzzing' variables.

Extraneous variables - Nuisance variables but randomly distributed.

Confounding variables - Vary systematically with the IV.

Demand characteristics - Participants second guess the aims and alter their behaviour.

Investigator effects - The unconscious influence of the research on the research situation.

Randomisation - The use of chance to reduce the researcher's influence.

Standardisation - Ensuring all participants are subject to the same experiment.

Independent Groups - Particpants in each condition of an experiment are different.

Repeated Measures - All participants take part in all conditions.

Matched Pairs - Similar participants put in pairs and allocated to different experimental conditions.

Evaluation

Inderpendent groups - Less economical. No order effect. Participant variables not controlled.

Repeated measures - Order effects. Demand characteristics. No participant variable problems. More economical.

Matched Pairs - No order effects. Cannot match participants exactly. Time-consuming.

Lab experiments - IV is manipulated in a controlled setting.

Field experiments - IV is manipulated in a natural setting.

Natural experiments - IV has been manipulated naturally, effect on DV is recorded.

Quasi-experiments - IV based on an existing difference between people, effect on DV is recorded.

Evaluation

Lab experiments - High internal validity (control). Low external validity (low realism). Cause and effect. Replication. Demand characteristics.

Field experiments - Lower internal validity. Higher external validity (realism). Ethical issuses.

Natural experiment - Low internal validity (no random allocation). High external validity. Unique research. Opportunites may be rare.

Quasi-experiments - Low internal validity (no random allocation). High external validity.

Random sampling - All members of the population have an equal chance of selection.

Systematic sampling - Selecting every nth person of a group.

Stratified sampling - Sample reflects the proportion of people within different population strata.

Opportunity sampling - Choosing whoever is available.

Volunteer sampling - Participants 'self-select'.

Evaluation

Random sampling - No researcher bias. Time-consuming. May end up with bias sample.

Systematic sampling - No researcher bias. Usually fairly representative. May end up with bias sample.

Stratified sampling - No researcher bias. Representative. Cannot account for all sub-groups.

Opportunity sampling - Convenient. Researher bias. Unrepresentative.

Volunteer sampling - Less time-consuming. Attracts a certain profile profile of people.

Confidentiality - Protecting private personal data.

Deception - Telling the truth.

Consent - Advising participants of what is involved. May reveal research aims.

Debrief - May reveal research aims.

Right to withdraw - Allow participants to leave the research at any given time.

Protction from harm - Minimising psychological and physical risks.

Evaluation

Informed consent - Get permission. Presumptive, prior  general, retrospective.

Deception/Protection from harm - Debriefing.

Privacy and confidentiality - Maintaining anonymity. Use numbers not names.

Pilot studies - Checking procedures and materials. Making modifications.

Single blind - Participants aren't made aware of research aims until the end.

Double blind - Neither participants nor the individual conducting the research know the aim beforehand.

Control group/condition - Used as a comparison.

Naturalistic observations - Behaviour observed where it would normally occur. No control over variables.

Controlled observations - Some control over environment, including manipulation of variables to observe effect.

Covert and overt observations - Observing participants with or without their knowledge.

Participant and non-participant - To join the group or remain an outsider.

Evaluation

Naturalistic observations - Low internal validity (control is difficult). High external validity (especially when covert).

Controlled observations - Low internal validity-though some extraneous variables may be controlled. High external validity (especially when covert).

Covert and overt observations - Covert: Low participant reactivity but ethically questionable. Overt: Behaviour may be affected.

Participant and non-participant - Participant: Increased external validity but may 'go native'. Non-participant: More objective but less insight.

Unstructured and structured - Researcher records everything (unstructured) or controls what is recorded (structured).

Behavioural categories - Target behaviours broken down into observable components.

Sampling methods - Continuous. Event sampling: count events. Time sampling: count at timed intervals.

Evaluation

Unstructured and structured - Unstructured: more information but may be too much, qualitative data harder to analyse. Structured: May miss behaviours.

Behavioural categories - Must be observable. Avoid dustbin category. No overlap.

Sampling methods - Event: Useful for infrequent behaviour, may miss complexity. Time: Less effort but may not represent whole behaviour.

Questionnaires - Pre-set list of written questions.

Closed and open questions - Fixed number of answers or not.

Evaluation

Questionnaires - Can distribute to many people. Easy to analyse. Social desirability bias. Acquiesence bias.

Closed and open questions - Produces quantitative or qualitative data, affected ease of analysis.

Structured interviews - Pre-set questions in a fixed order.

Unstructured interviews - No set formula, just a general topic. Questions developed based on responses.

Semi-structured interviews - Pre-set questions with flexibility to ask follow ups.

Evaluation

Structured interviews - Similar to questionnaires but fewer respondents.

Unstructured interviews - More flexibility. Analysis is more difficult. Social desiability bias may be reduced by rapport.

Semi-structured interviews - Advantages of both structured and unstructured.

Designing self-report

Questionnaires - Likert scale, rating scale, fixed choice option.

Interviews - Standardised interview schedule, to avoid interviewer bias. Awareness of ethical issuses.

Writing good questions

Overuse of jargon - Don't be too technical.

Emotive languages and leding questions - Replace 'loaded' words and phrases with neutral ones.

Double-barrelled question and double negatives - Ask one question only in a clear way.

Types of correlation - Positive, negative and zero.

Difference between correlations and experiments - No IV or DV. No manipulation of variables.

Evaluation

Strengths

• Useful preliminary tool.
• Quick and economical to carry out, using secondary data.

Limitations

• Cannot demonstrate cause and effect.
• The third variable problem (interviewing variables).
• Misuse and misinterpretation.

Qualitative Data - Written, non-numerical description of the participants' thought, feelings or opinions.

Quantitative Data - Expressed numerically rather than in words.

Evaluations

Qualitative Data

• Rich in detail.
• Greater external validity.
• Difficult to analyse.
• Conclusions may be subjective.

Quantitative Data

• Easy to analyse.
• Less biased.
• Narrow in scope.

Primary data - Collected first haand from participants for the purpose of the investigation.

Secondary data - Collected and analysed by someone other than the researcher.

Evaluation

Primary data

• High validity.
• Targets relevant information.
• Time and effort.

Secondary data

• Inexpensive and easy to access.
• Variation in the quality.
• Outdated and incomplete.

Measures of Central Tendency

Mean - Add them all up and divide by the number.

Median - The middle value.

Mode - Most frequently occuring.

Evaluation

Mean - Most sensitive and representative. Easily distorted.

Median - Not affected by extreme values. Less sensitive than the mean.

Mode - Easy to calculate. Crude, unrepresentative.

Measures of Dispresion

Range - Subtract the lowest from the highest.

Standard Deviation - Measures how many scores deviate from the mean.

Evaluation

Range

• Easy to calculate.
• May be unrepresentative of the data set.

Standard Deviation

• Much more precise than the range.
• Can be distroted by extreme values.

Presentation and display of Quantitative data

Tables - Raw scores are converted to descriptive statistics and summarised in a table.

Bar charts - Discrete categorical data represented for clear comparison. The frequency of each category is the height of the bar.

Scattergrams - Shows the strength and direction of a relationship between co-variables.

Distributions

Normal distribution - Bell curve. Mean, median and mode at same point. Tails never touch zero.

Skewed distribution - Negative skew leans right. Positive skew leans left.

Percentages and fractions - Convert one to ther and to decimals.

Decimals - Appropriate number of significant figures.

Ratios - Part-to-whole. Part-to-part.

Mathematical symbols - =,>,<,>>,<<

Funding - Approval of project proposal.

Validation - Quality check.

Improvements - Minor revisions or rejection of report.

Evaluation

Anonymity - May permit unjustified criticisms by rivals.

Publication bias - File drawer problem, creates false impression of current knowledge.

Burying ground-breaking research - Maintains status quo.

Attachment research

• Equal care from mother and father, means more effective contribution to economy.

Mental health

• Absenteeism due to moderate mental health (e.g. depression) issuses costs the economy.

Significance - Results have not occured by chance.

Probability - The 5% significance level. The more stringent 1% level.

Critical value - Cpm[arison with calculated value to determine significance.

The Sign Test

Criteria

• Testing for difference.
• Nominal data.
• Repeated Measures.

Steps

• Convert to nominal data.
• Add up pluses and minuses.
• S=Less frequent sign.
• Compare calculated value of S with critical value.