Inferential Statistics Flashcards
Criteria for Mann Whitney U
independant groups
experiment
ordinal/interval data
criteria for Wilcoxon
repeated measures
experimental
ordinal/interval data
Criteria for Spearmans Rho
correlation
relationship
ordinal
Criteria for Chi square
Independant groups
difference
nominal data
criteria for Binomial
Repeated measures
difference
nominal
Observe value define
the number produced after many steps and calc
critical value define
a value taken from statistical test table which must be reached in order for results to be significant
How to know if my results are significant:
-observe value must be greater than critical value= spearmans rho + chi square
-observe value must be lass than critical value= man whit u + wilcoxon
name all the non parametric tests
mann whitney u
chi square
binomal
wilcoxon
spearmans rho
Name 4 assumptions of parametric tests
populations drawn from are normally distributed
variances of populations are equal aprox
has interval data
no extreme scores
-if has else then its non parametric
How to decide which test
level of measurement- is it nominal, ordinal or interval
experimental design- is it independent or repeated
relationship or difference- correlation or experiment
Describe nominal data
categories
weakest level of measurement
n. something occurs
closed questions usually
easy to collect
no degree of response shown since not linear
Describe ordinal data
rank
difference between each rank is unknown
likert scale
self reports
linear scale so mean mode calculated
median not assessed since gaps unknown
Describe interval data
known mathematical unit
most precise + reliable level of data
more informative
standard deviation can be calculated
no baseline to stop
no understanding of why
What is the standard deviation formula
σ = √(∑x−x̄)^2 /n)
What is variance
calculates the average difference between scores in the data set and the mean
the larger the SD the more spread out the results are
a small variance means scores are mostly similar and close to mean
SD is the square root of variance
evaluate use of mode
identifies most frequent
useful IF data is categories/nominal
not useful if several modes
evaluate the use of median
not affected by outliers since its marginilised
not as sensitive as mean bc all values are represented
evaluate the use of mean
most sensitive measure since all values are taken into account
can be inflated and deflated by outliers
