# What is that answer to this statistics question?

label Statistics
account_circle Unassigned
schedule 1 Day
account_balance_wallet \$5

What are the assumptions underlying the analysis of variance?

Jul 1st, 2015

Assumptions can pertain to:

• Measurement scale

• Method of sampling and/or assigning subjects to treatments

• Selection of factor levels

• Etc.

Assumptions in analysis of variance:

1. The normality assumption

2. The homogeneity of variance assumption

3. The independence assumption

What can be done?

• In general: –

Analysis of variance is robust.

• Nominal α level ≈ actual α level – No normality – Heterogeneous variances (cell frequencies large and equal) – But: small violations of the independence assumption can have dramatic effects.

• This is a requirement of the design

Normality of variances and homogeneity:

• Normality of variances – The distribution is skewed – The dependent variable is not an interval variable – Kolmogorov-Smirnov and Shapiro-Wilk normality tests

• Homogeneity of variances – Levine’s test => Welch’s test

Example:

• Dyslexia

• Communicative Developmental Inventory

• Closed-class words

• 17m, 23m & 29m

Jul 1st, 2015

...
Jul 1st, 2015
...
Jul 1st, 2015
Nov 19th, 2017
check_circle