Week 6

Question 1

What is the purpose of descriptive statistics?

Answer 1

Check for errors/outliers
Describe and summarise
Spread of the data
Ensure appropriate analysis

Question 2

What are the ways to summarise data? (Ratio or interval)

Answer 2

Measure of central tendency
- mean, median, mode

Measure of dispersion
- variation, range, standard deviation

Normal curve, Skegness, kurtosis

Question 3

What are inferential statistics?

Answer 3

All statistical tests are of common structure:

• set up a null and alternative hypothesis
• establish a level of statistical difference (also known as alpha)
• determine statistical significance (pvalue)
• if the pvalue is greater than the alpha you can accept the null hypothesis

Question 4

What are the steps to undertaking a hypothesis test?

Answer 4

Define study question

Set null an alternative hypothesis

Calculate a test statistic

Calculate a p value

Make a decision and interpret

Question 5

What are parametric tests?

Answer 5

Interval/ratio data
Assumes that data is drawn from a normally distributed population and have the same variance in the variable being measured

Question 6

What are non parametric tests?

Answer 6

Nominal/ordinal

Do not make any assumptions about the characteristics of the sample in terms of its distribution

Question 7

What is a t test used to compare?

Answer 7

Means

Most common statistical procedure used by researchers

Question 8

What are independent sample designs?

Answer 8

Comes from different groups - compare differences between groups

More random error because control group may be very different to treatment group

Question 9

What is a paired design?

Answer 9

Data comes from one group at different times - compare differences between times

Each person is their own control so variation is limited

Question 10

What are independent samples t tests?

Answer 10

Measurements in condition 1 are independent of condition 2

If the H0 is true we expect the difference between 1 and 2 to be 0

Using a sample will produce error

If probability less than 5% reject H0 and accept Ha

Question 11

What are the assumptions for independent t tests?

Answer 11

Dependant variable is ratio/interval

If either group is small (30 or less) the distribution of the dependent variable should not be badly skewed

The variance of the DV should be be very different

A problematic difference in variances is indicated by a significant Levene’s test

If significant, interpret p value associated with ‘equal variances not assumed’ and vice versa

Question 12

What is a paired samples t test?

Answer 12

DV is ratio/interval and IV has two categories
Each measurement in condition 1 has a match in condition 2
One measurement is ducted from the the other so each case has a different score

Question 13

Describe non parametric equivalent to t tests

Answer 13

If we have an ordinal scale DV or a ratio/interval DV that does not meet parametric assumptions, we use non parametric equivalents

These compose medians rather than means

Usually less powerful

Question 14

When to use non parametric tests

Answer 14

Used when assumptions of parametric tests are not met

When it’s not possible to correct for problems with the distribution of a data set

Many of the tests use ‘ranked’ data

Question 15

What is the Mann Whitney U test?

Answer 15

Used when data is independent but not parametric

Used to test the null hypothesis that two samples come from the same population

Question 16

What are the assumptions of the Mann Whitney U test?

Answer 16

The two samples are independent
Uses ranks instead of raw data values
Data values assigned ranks to both samples combined

Question 17

When to use the Mann Whitney U test?

Answer 17

Sample sizes are small and normality is questionable

Data contains outliers or extreme values that distort the mean values

Data are ordinal

Assumes distribution of the two groups are the same shape

Question 18

What is the sign test and Wilcoxon signed-rank test?

Answer 18

Sign test can be used to measure the differences between each variable as non parametric alternatives to the samples t test

Wilcoxon signed-rank test can be used to compare paired data as non parametric alternatives to the paired t test

Used when you cannot justify a normality assumption for the differences

Sign test counts the number of differences that are +ve and -ve and makes a decision based on these counts

Question 19

What is the p value?

Answer 19

The exact probability that H0 is true
When p is no more than .05 you reject H0 which leaves up to a 5% chance that you are wrong in concluding that there is a difference.

Question 20

What is a type 1 error?

Answer 20

Rejection of the null hypothesis when it is true

Question 21

What is a type 11 error?

Answer 21

Failure to reject the null hypothesis even though it is not true