Week 8: cross tabulation

Page 1

Common tabulation:

compares 2 variables.

need to choose which variables and why.

Hypothesis:

H0 = null hypothesis ie. no relationship between the two variables

H1 = Alternate hypothesis ie. there is a relationship.

ERRORS

Type 1: (alpha) likelihood of rejecting the null and accepting h1 incorrrectly.

confidence level/coefficient is 1-alpha

(bigger is better)

Type 2: (beta) likelihood of accepting null and rejecting h1 incorrectly.

confidence is 1-beta

(bigger is better)

you can increase sample size (n) to reduce both errors.

Reliability measure (level of significance)

95% = a significance of <0.5

this means that a 95% confidence level of accepting the H1

use 99% if you want to be sure and 90% at a stretch.

Data Types

Univariate data- non numeric data

- Metric data - one sample - t test

- z test (sample <30)

- 2+ - related - paired t test (eg. employee training and customer satisfaction

- independent - two group test

- z test

- one way ANOVA

Univariate hypothesis test: z test

-determines if means are different when variance and sample size is known. Assumed normal distribution and SD ( and 1 respectfully)

Formula: mean of sample - hypothesised mean/ (population standard deviation/square root of n)

T Test

-two group or data sets you want to compare the mean score of a continuous (metric) variable

ANOVA:

-tests the equality of means of a dependent variable amongst independent variables.

dv = metric IV = non-metric

H0 = no difference between means.