Research Methods

• Systematic study through the physical and natural world of experimental observation and research
• Ask a question - existing literature - state a hypothesis - test the hypothesis - analyse results - draw conclusions
• Psychological research does not prove anything
• Human behaviour never deals with absolutes - if absolute proof existed the scientific method would not loop - this enriches our understanding
• Describe - explain - predict - change
• Pure - expanding the bound of knowledge, not immediate application
• Applied - expanding the bounds of knowledge, can be applied to the real world
• Psychology is an epistemological(study of knowledge) paradigm(pattern or model)
• Empiricism - all knowledge acquired through the 5 senses - brain is sealed - David Hume
• Research methods/practices are rooted in philosophy
• Rene Descartes - Rationalism

• Research Dissemination - journals, conferences, media, teaching
• Title - relevant, creative, clear, concise, not too short, not too long
• Abstract - summary, 120-150 words, written last, not including citations
• Introduction - introduce, previous research, why, compare existing literature, defining key terms, state hypothesis
• Method - participants, research method(type of research, design, variables), procedure(step by step, chronological, replicable), materials
• Results
• Discussion - answers hypothesis, implications, future research, strengths and weaknesses
• References and citations (APA)

Types of Data:

• Nominal - Categories for data. No Order. 
• Ordinal - Categories with an order, differences between points may not be equal.
• Ratio - Impossible to go below zero
• Interval - Ordinal data with equal differences, can go below zero

Variable: A measurable factor or anything that can change

IV: Antecedent to dependent variable, manipulated to observe the effect on the DV

DV: Dependent on the effect of another variable, the thing we measure

Extraneous variable: IV not related to the purpose but may affect the DV

Experimental error: The effect of the extraneous variable on the DV

Control: Minimising the effects of the extraneous DV

Confounding relationship: IV and DV is not free from the influence of extraneous variables

Lab experiment - artificial situation, purpose to study relations between many variables

Field Experiment - An intervention or event in the real world and naturally occurring environments

True experiment - Quasi-experiment - Nonexperimental

True experiment: Manipulation(treatment or conditions) and randomization

IMD/Between Participants design:

• Two separate groups of people take part in different conditions of the experiment
• Advantages - no order effects, easy to randomly assign people
• Disadvantages - more individual differences error

RMD/Within participants design/Related pairs:

• Single group of people take part in different conditions of the experiment
• Advantages - controls for individual differences, statistically powerful, less participants
• Disadvantages - Practice effects, order effects(use counter-balancing to eliminate), wait time required, design becomes transparent to the participants

Matched samples design: Participants matched on specific criteria

Quasi-experimental designs:

• Compares non-equivalent preexisting groups
• Not possible to randomly allocate participants (IV naturally occurring)
• Used when we want to investigate the DV in different types of participants

Controls: Trying to keep track of one variable and trying to ensure nothing produces the results

• Practice effects - same participants in two conditions - randomise order in which participants take part in the conditions
• Fatigue - randomise order in which participants take conditions, rest in between, get different participants for both conditions
• Samples - representative, avoid bias, sample is big enough
• Order effects - counterbalancing
• Blinding - concealment of information - single or double
• Complex true experiment - Randomised, increased levels of IV's, multiple DV's
• Non-experimental designs - Correlation, case study, survey

• Type of Non-experimental design - relationship between two variables
• No IV's, no control groups, no manipulation or randomization
• Variation in one variable is related to the variation in another variable
• Describes both research design and statistical test - can use other statistical tests in correlations
• Developed by Francis Galton in 1888 - during work on hereditary
• Later refined as a mathematical formulae by Karl Pearson
• Bivariate Method (2 variables) - may have multiple bivariate correlations
• Correlations presented as scatter plots with lines of best fit
• NO IV or DV, just variables
• Predictor variable and criterion variable - typically doesn't matter which is which
• HOWEVER, if one variable could be an IV in future research we call this the predictor variable
• Direction and strength, can be positive negative or nonexistent
• Scatter plots do not reveal the whole story
• Correlation coefficient - direction and strength - represented by the letter R - no zero before decimal place! - Weak - moderate - strong (Positive or negative)
• Coefficient of determination - times the coefficient by itself - e.g. .6 x .6 = 36%
• Assumptions of linearity - curvlinear relationships

• Descriptive (Organisation and analysis of data to yield meaningful information) or inferential (drawing conclusions from samples to be able to generalise to the population - testing hypothesis)
• Results can only be ever probabilistic - describe results and probability of results due to chance
• Measures of central tendency - mean, median, mode - top of bell curve - average of scores
• Bimodal distributions
• Measures of variability/dispersion - range, variance, standard deviation - spread of scores
• Quartiles - interquartile (3rd-1st quartile)
• Variance - deviation from the mean(each score subtract the mean) - you have to get rid of the minus signs - square the deviations - equation
• Standard deviation - square the variance - spread diagram
• Histograms
• Illustrative statistics - scatter plots - bar graphs with error lines - line graph with standard error bars - pie charts

• Population is an entirety of cases under consideration
• A sample is a representative subset of a population
• The higher the number of scores in a sample the closer the mean is likely to be the population mean
• If a sample is taken to represent or generalise a population - the bigger the sample the better - more closely represent the population
• Parameters and sample statistics symbols
• Probability sampling - simple(equal chance)/systematic random(first person chosen at random, then use fixed intervals until desired sample)/stratified(independent subgroup sampling)/cluster random(random samples of clusters then random samples of individuals)/Multi-stage(similar to cluster, several stages, large populations)
• Non-probability - Quota(sampling frame defined in advance, sample chosen from list)/ purposive(selecting participants in relation to aims of research question)/convenience(basis of opportunity)/snowball
• Representativeness - The extent to which a sample bears the characteristics of a population
• Sampling bias - the error resulting from taking a non-random sample of a population
• Self-selection(volunteer) - people interested in a subject are more likely to take part but not as representative as the population

• A branch of correlational research - measures same variables over a long period of time - time becomes a variable which is used in analysis - mostly used in developmental psychology
• The Caerphilly Heart Disease study - The 1970 British Cohort Study - Child of our lifetime study
• Issues: retention, cause and effect, needs a very large representative sample
• Comparative designs - compares across cultures and countries - innately done a lot of the time
• Comparative designs require massive coordination across the world - often cultural language differences raises issues with validity 

• Questionnaires - self-report, variety of questions, usually one topic, open or closed, sets of scales, can refer to both scale and questionnaire
• Scale - 'standardised instrument' - closed question(same type) - often has calculable totals or subtotals - rigorously studied(reliability and validity) - used by nurses and GPs - measures a single psychological construct
• Open questions - space to write, think and reflect, opinions and feelings, they have control of the conversation
• Closed questions(forced choice) - measured nominal, ordinal(ranks), interval or ratio(numerical), facts, easy quick, they keep control of conversation as respondent
• Open answers can be a direct answer or sentence completion
• Closed questions can be a choice of categories, rating scales(Likert or semantic differentials), visual analogue scales, checklists or ranking
• Conventional language, Purposeful questions, concrete questions, use time periods based on importance of questions to obtain valid information 
• Avoid double-barreled questions, negative questions, biased/leading questions 
• First page is a consent form

• Questionnaires can be self-administered (+'s - cheap, easy convenient, preserves confidentiality, can get large samples) (-'s - may forget issues, misinterpret questions, open questions generate large amounts of data, low response rates, not suitable for some groups, nearly 90% of communication is visual) , face to face (+'s - allows clarification of ambiguity, allows participation of illiterate or disabled  individuals), phone or on the internet