Research Methods Continued Flashcards
What is reliability?
Reliability is the measure of consistency - getting the same results each time
What is test-retest reliability?
Test-retest reliability is the same person/group undertaking the research measure - potential demand characteristics- time has to be taken into account between testing
Understanding test-retest results
Test-retest results-> scores correlated after measure has been taken on two occasions - significant = high reliability (+0.8-1)
Inter-observer reliability
Inter-observer reliability = extent to which two observers are observing and recording behaviours in a consistent way -> useful in ensuring reliability
Inter-observer reliability and behavioural categories
Inter-observer reliability - used in measuring behavioural categories - inter-observer reliability ensures categories correct - as psychologist observe same situation together or seperately -scores are then correlated
Improving reliability and questionnaires
Can identify questions that has the biggest impact on reliability
Improving reliability and interviews
Adjusting interviews:
-same interviewer (creates researcher bias)
-training interviewers - eg the way questions are asked, how open they are
Improving reliability and experiments
Experiments:
-level of control researcher has over variables
-lab = high reliability = control over IV - standardised procedures
-control over extraneous variables
Improving reliability and observations
Observations:
Can look objectively- relies on observer interpretation
Operationalise behavioural categories (clear and specific)
Difference between related and unrelated designs in psychology
Related designs = designs such as repeated measures and matched pairs where there are connections
Unrelated designs = designs such as independant groups where there is no connection between people in each group
What are the 3 levels of data?
The three levels of data are norminal, odinal and interval
What is meant by nominal data?
Norminal data refers to categorical data -> numbers referring to people in categories -> discrete data as each part appears in only one category
+ = Easily generated from closed questions
- = lacks depth
What is meant by ordinal data?
Data ordered in some way - used to rank data on a numerical scale from high to low
+ = More detail than nominal
- = intervals not equal value so cannot use a mean
What is meant by interval data?
Interval data: data on a numerical scale with units of equal size eg temp, objective
->Ratio data = type of interval data = fixed 0 = cannot have negative - eg weight and height
+ = more informative and reliable
- = can be arbitrary
Considerations that researchers have to take into account when deciding on an appropiate statistical test
Considerations that researchers have to take into account when deciding on an appropriate statistical test:
-Wether you are investigating a difference (one control and one experimental condition) or a relationship (of two co-variables)
-Experimental design (when looking for a difference
Statistical tests used for nominal data (for related or unrelated)
Nominal data:
Test of difference for related designs = Simon Sign Test
Test of difference for unrelated designs = Cowell Chi-Squared
Stastical tests used for ordinal data (for related or unrelated
Ordinal data:
Test of difference for related designs = Wilcoxon (Wants)
Test of difference for unrelated designs = More Mann-Whitney Test
Test of association for ordinal data = Singers Spearman Rho
Stastical tests for interval data
Interval data:
-Test of difference for related designs = Related T-test (parametric, receiving)
-Test of difference for unrelated designs = Unrelated T-test (parametric, unanimous)
-Test of association = Pearson’s R (parametric, praise)
3 Parametric assumptions that have to be applied to use statistical tests
3 Parametric assumptions that have to be applied to use statistical tests:
-Data should be interval
-Data should be drawn from normally distributed population (inverted U)
-Should be homogeneity of variance between conditions (the deviation of scores is similar between conditions)
What is the sign test?
The sign test is used when looking for a difference between paired data ie repeated measures or matched pairs - generates nominal data
How to carry out the sign test
How to carry out the sign test:
-Each pair of data scored with either + or - (if there is no difference = 0)m
-Value of S is calculated (adding total + and - and selecting smallest value)
-Calculate N value (total - nil scores)
-Use against critical value to determine significance (equal or less to be significant)
Chi-squared test
Chi-squared test = Test of either difference or association - used on nominal, unrelated data
How to carry out the chi-squared test
Chi-squared test:
-Frequencies recorded and put into contingency table
-Degrees of freedom calculatred before determining significance of data ( df = ((rows-1))x((columns-1)) )
-To be significant = results equal or more than critical
Calculating degree of freedom for chi-squared test
Calculating degree of freedom for chi-squared test:
df = (rows-1) x (columns-1)
What three things should science have?
Science should have:
-Assumptions (paradigm - shared set of collective assumptions
-Scientific methods (empiricism)
-Terminology
What was peppers idea on science?
Peppers idea = falsification - if it is false, then this can be shown by observation or experiment - cant be fully proven right but can be proven wrong - no theory in science final
What is predictive power?
Predictive power is theories should take nomothetic approach (general laws should be applied)
What is empiricism?
Empiricism - Locke - gathering information through observation/experiments rather than reasoned argument/opinion
Types of internal validity?
Types of internal data:
Face validity - does it look like what is should be measuring
Concurrent validity - comparing to established test
Construct - using specific group with predictable results
Predictive - can performance be used to predict future behaviour
Types of external validity
Types of external validity:
Ecological - reflective of everyday behaviour
Temporal - reflective of time periods
Population - reflective of target populations
What is standard deviation?
Standard deviation is a measure to show to what extent the values in data deviate around the mean
What is counterbalancing?
Counterbalancing is a technique used to deal with order effects when using a repeated mesures design - splitting up sample
Designing experiments in psychology
Designing experiments in psychology:
-Self-report methods (interviews, questionnaires) (open/closed)
-Experimental design (RM, IM, MP)
-Procedure (sample size, type of sampling)
-Consent from participants
