Book 2

• Case Studies


  Detailed and in-depth analysis of an individual, group, institution or event

  Usually unusual individuals individuals or events

  Qualitative data

  Case history using interviews, observations, questionnaires

  May be subject to experimental or psychological testing to assess what they are capable of

  Longitudinal data

  Gather data from those around them
• Content Analysis 


  Observational research where people are studied indirectly via the communications they have produced

  Spoken interaction, written forms, media

  Aim is to summarise and describe this communication in a systematic way so overall conclusions can be drawn
• Coding


  Initial stage of content analysis

  Some data sets may be extremely large so needs to be categorised into meaningful units

  Qualitative data
• Thematic Analysis


  Form of content analysis

  Outcome is qualitative

  Identification of themes, any idea explicit or implicit that is recurrent

  May then be developed into wider categories

  Once reader satisfied the themes cover most aspects of the data they are analysing may collect new set of data to test the validity of themes and categories

  Write up final report using direct quotes from the data to illustrate each theme
• Reliability


  Measure of consistency

  Measurement made twice and produces the same result then that measurement is described as being reliable

  Same measurement every time object measured
• Ways of Assessing Reliability

  Test-retest

  Inter-observer reliability
• Test-retest

  Administering the same test or questionnaire to the same people on different occasions

  Results should be the same or similar

  Commonly used with questionnaires and psychological tests but can be applied to interviews

  Long enough between you forget answers

  Not too long that attitudes, opinions or abilities changed

  Questionnaire or test can be correlated

  If correlation significant measuring instrument assumed to be good
• Inter-observer reliability

  One observer perspective different to another causing subjectivity, bias, unreliability

  Should conduct research in teams of two<

  May involve a pilot study to check observers applying behavioural categories in the same way

  Observers watch the same events but record data independently and correlated

  Content analysis - Inter-rater reliability

  Interviews - Inter-interviewer reliability
• Measuring Reliability

  Correlational analysis

  Correlational coefficient should exceed +0.80 for reliability
• Improving reliability - Questionnaires

  Test-retest

  Correlation should exceed +0.80

  Low test-retest require some items deselected or rewritten

  Replace open questions with closed fixed-choice alternatives to reduce ambiguity
• Improving reliability - Interviews

  Same interviewer each time

  All interviewers properly trained

  More easily avoided in structured interviews where behaviour is more controlled by fixed questions
• Improving reliability - Observations

  Behavioural categories have been operationalised, measurable and self-evident

  Categories should not overlap

  All possible behaviours should be covered on the checklist
• Improving reliability - Experiments

  To compare participant performance procedures must be the same

  Standardised procedures
• Types of Validity

  The extent to which an observed effect is genuine or legitimate

  Broken scales consistent (reliable) but not right amount (valid)
• Internal validity

  Extent the researcher managed to measure what they intended to
• External Validity

  Extent to which findings can be generalised beyond the research setting in which they were found

  Ecological validity
• Ecological Validity

  Type of external validity

  Generalising the study to other settings

  Natural setting should make it more generalisable, high ecological validity

  Lab low ecological validity

  Low mundane realism of the dependant variable = low ecological validity
• Temporal Validity

  Type of external validity

  The extent to which findings from a research study can be generalised to other historical times and eras

  Asch - high conformity due to 1950s America in the conformist era

  Freud - penis envy outdated, sexist and reflection of patriarchal Victorian society
• Face Validity

  Basic form of validity in which a test, scale or measure appears to measure what it claims to
• Concurrent Validity

  The extent to which a psychological measure relates to an existing established similar measure

  Intelligence test compared to IQ

  Close agreement between the two sets of data would indicate the new test has high concurrent validity and close agreement if correlations exceeds +.80
• Improving validity - Experiments

  Control groups

  Standardise procedure, minimise participant reactivity and investigator effects

  Single-blind, reduce demand characteristics

  Double-blind, reduce demand characteristics and investigator effects
• Improving validity - Questionnaires

  Lie scale, assess the consistency of responses and control effects of social desirability bias

  Assuring participants data is anonymous
• Improving validity - Observations

  Minimal intervention from the researcher

  Covert observations, undetected, behaviour likely to be natural and authentic

  Broad, overlapping or ambiguous behavioural categories negative effect on validity
• Improving validity - Qualitative Research

  Higher ecological validity as less interpretive

  Depth and detail better represent reality

  Demonstrate interpretive validity

  Triangulation, use of a number of different sources as evidence
• Interpretative Validity

  The extent to which the researcher's interpretation of events matches that of their participants

  Can be demonstrated through coherence of researcher's narrative and the inclusion of direct quotes

  Case studies
• Choosing a Statistical Test

  1 - Difference or correlation

  2 - Experimental design: unrelated design (independent groups) and related design (repeated measures or matched pairs)

  3 - Level of measurement: nominal (categories), ordinal (ordered) and interval (numerical scales)
• Nominal data and unrelated design

  Chi-Squared
• Nominal data and related design

  Sign test
• Nominal data and correlation

  Chi-squared
• Ordinal data and unrelated design

  Mann-Whitney
• Ordinal data and related design

  Wilcoxon
• Ordinal data and correlation

  Spearmans rho
• Interval data and unrelated

  Unrelated t-test
• Interval data and related design 


  Related t-test
• Interval data and correlation

  Pearson's r
• Nominal Data

  Categorical data

  Discrete, one item can only appear in one of the categories

  Mode
• Ordinal data

  Ordered data

  Doesn't have equal intervals between each unit

  Based on subjective opinion, lacks precision, unsafe data so raw scores converted to ranks used in calculations

  Median and range
• Interval data

  Numerical scales that include units of equal, precisely defined size

  More detail is preserved

  Public scales of measurement that produce data based on accepted units of measurement

  Most precise and sophisticated form of data

  Necessary for parametric tests

  Mean and standard deviation
• Null Hypothesis

  States there is no difference between the conditions

  Statistical test determines whether we accept or reject the null hypothesis

  H0 - null hypothesis

  H1 - alternative hypothesis (directional or non-directional)
• Probability

  Statistical tests work on the basis of probability

  Significance level 0.05

  Probability the observed effect occurred when there is no effect in the population is equal to or less than 5%

  When researchers claim to find a significant difference or correlation there is still 5% chance it is not significant

  Psychologists can never be 100% certain as they cannot access all members of the population under all circumstances