Statistics - created from Mind Map

Página 1

Types of data

Descriptive statistics

Inferential statistics

Methodological issues

Ethics and conduct

Nominal - used for categories (i.e. yes and no)

Ordinal - you can order the data but it isn't of equal intervals

Interval - can be ordered and has equal/standardised intervals between data

Qualitative - uses language

Quantitative - uses numbers

Primary - collected by researcher themselves

Secondary - not collected by the researcher themselves

MOCT

MOD

Other maths

Graphs

Mean - average of all results, only used with interval data

Median - only uses ordinal or interval

Mode - all types of data

Range - only for ordinal or interval data

Variance - for interval data only, how close to the mean values are

Standard deviation - interval data only, square root of the variance

Fractions

Percentages

Ratios

Pie chart - nominal data

Scatter diagram - for a correlation

Line graph - interval and continuous

Bar charts - nominal data

Histograms - only with interval data

Distribution curves

Non-parametric tests

Errors

Probability

Parametric tests

Normal - most scores are around the midpoint and there are less further away from it, true normal = all MOCT occur at highest point

Skewed - when there's an unusual data set

Negatively skewed curve - less people at lower end, mean is less than the mode

Positively skewed curve - more people at the lower end, mean is greater than the mode

Chi square - nominal, independent

Mann-Whitney U - ordinal, independent

Binomial Sign - nominal, repeated

Wilcoxon Signed Ranks - ordinal, repeated

Spearman's Rho - correlation

Type 2 - accept null hypothesis when alternative is true

Type 1 - accept alternative hypothesis when behaviour is due to chance

By decreasing p we increase the risk of a type two error, opposite for increasing p

Level of significance is usually 95% (p < 0.05)

p < 0.05 = 5% it happened by chance

Used for interval/ratio data

Has to have normal distribution

Similar variances

Generalisability

Reliability

Validity

Biases

Sample should be representative of target population

Need to be representative to generalise to the population

Internal - consistency of results of a test across items within that test

External - how test scores vary from one time to another

Inter-rater - two observers consistently rate the same behaviour the same, correlate them to see it (+0.8)

Split-half test - Test one half of questions and gain a score then do the same for the other half. Results should be consistent.

Test-retest method - redo the test to see if you get the same/similar results

Face - how good it looks to be at testing what it's meant to test

Internal - are effects due to the IV or extraneous variables?

Construct - whether it measures what it intends to (e.g. intelligence)

Concurrent - give same results as another test/study

Criterion - how much one measure predicts the value of another measure

Population - how much it measures the general population

Ecological - how true to life it is

External - generalisability to other situations

Demand characteristics - guess aim and modify behaviour to suit it

Researcher/observer bias - may observe different things due to knowing aim/hypotheses of the study

Social desirability - want to appear as a good member of society

Researcher/observer effects - behaviour influenced by their presence

Respect

Responsibility

Competence

Integrity

Informed consent

Withdrawal

Confidentiality

Protection of participant

Debrief

Deception