Week 9

Page 1

metric data is cumulative

T test:

one sample: group v a hypothetical mean OR price v competitors price OR comparing two groups from the same sample eg. a before and after result.

ANOVA

more than one thing that are interrelated.

H0 = no difference between the means.

NANOVA: 2+ DV to 1 IV

Association between variables (shows direction/strength/type of relationship)

-chi square: shows scatter diagram, pearson correlation co-effictient and spearman correltion between 2 non metric variables.

-scatter plot

-correlation

pearson correlation is r

the coefficient of determination is r^2

a crosstab with 'a' at the bottom means it is invalid due to not enough data

a correlation can be meaningful or spurious and this is determined by the surrounding literature.

Mulitcollinearity

structural: x^2 values

Data based: poorly designed experiment

when all variables effect each other eg. age and income so you make them into one variable as a weighted score.

Discriminant analysis:

Dv: categorical/nominal

IV: interval data eg. a likert scale

This helps to predict group membership.