Lecture 2 - Analysis of Variance (ANOVA) Flashcards Preview

Statistics > Lecture 2 - Analysis of Variance (ANOVA) > Flashcards

Flashcards in Lecture 2 - Analysis of Variance (ANOVA) Deck (25)
Loading flashcards...
1
Q

What is variability?

A

It is to do with the range/spread of data. Data can have the same mean but different variabilities, or the same variability but different averages

2
Q

What do parametric tests assume?

A

They assume a normal population distribution

3
Q

What is the normal distribution?

A

a mathematical function that defines the distribution of scores in a population. It does this with regard to two ‘population parameters’ - Mu/µ and Sigma/σ

Most (but not all) data is normally distributed

4
Q

What does Mu/µ represent?

A

the population mean

is involved in calculating the normal distribution

5
Q

What does Sigma/σ represent?

A

the population standard deviation (variance)

is involved in calculating the normal distribution

6
Q

What does this show in terms of a normal distribution?

A

The same µ (pop mean) but different σ (std. dev./variance)

7
Q

What does this show in terms of a normal distribution?

A

The same σ (pop std. dev /variance) but different µ (pop mean)

8
Q

What does this show in terms of a normal distribution?

A

A different σ (pop std dev) and the same µ (pop mean)

9
Q

What do variances look like when:

H0 is true

H1 is true

A

H0 is true - scores almost completely overlap

H1 is true - very little overlap between score distribution

10
Q

What are the two kinds of experimental error?

A

individual differences error

measurement error

these are the two main kinds, there are others but these are the main ones. In a real experiment, both would contribute to each subjects’ score

11
Q

Why do we look at experimental error?

A

It contributes to the differences found between treatment conditions

So we must try to find out how much of the difference can be attributed to experimental error

12
Q

What can tell us about experiemental error?

A

The variability of subjects in the same condition ought to tell us about experimental error

13
Q

What are treatment effects?

A

the systematic source of variability that comes from the different conditions of the IV used in each treatment grouo

14
Q

What does partitinoning the deviations mean?

A

Basically, finding out the standard deviations

15
Q

What does A1, A2, A3 etc mean?

A

the different treatment conditions

16
Q

Why are the mean and standard deviation sometimes referred to as models?

A

Because they are values not actually present in the data, but they are used to describe and represent the data that is present

17
Q

What does standard deviation look at?

A

How far away individual values are from the mean
. In this way it assesses how good the mean is as a model of the data

18
Q

How do we calculate the size of a deviation? (NOT STD DEV)

How and why do we use this to calculate the sum of squares?

A

We subtract the mean (called x with a line on top) from the observed values.

Deviances may be positive or negative so it can give silly total deviances when some of the individual deviances are positive and some are negative.

That’s why we create a squared error (we square the deviations) - adding all of these together forms the Sum of Squares

19
Q

What is variance?

How is it calculated?

Why do we use it?

A

Variance is the average error between the mean and the actual data value

It is the sum of squares divided by n (the number of samples)

We use it because lots of data gives us a higher sum of squares so it can be hard to compare a larger dataset with a smaller one.

20
Q

How do we calculate a standard deviation?

Why do we use it instead of variance?

A

It is the square root of the variance

We use this frequently because it is a more meaningful way to describe the data - variance is squared and you can’t have ‘3.2 words squared’!

21
Q

What is the grand mean?

A

the mean of several treatment conditions together

22
Q

What is a sample mean?

A

the mean of each treatment group alone

23
Q

What does the between groups deviation represent?

A

both error and treatment effects

24
Q

What does the within-subjects deviation represent?

A

error alone

25
Q

What are the 4 assumptions of a between-groups ANOVA?

A

interval or ratio level data

normal distribution

equal variances between groups

scores are independent