Distributions and Probability

Question 1

What is normal distribution?

Answer 1

When data is symmetrical around central scores

Mean, median and mode are equal

Data should fit along a “Gaussian curve”

Question 2

How do you calculate Pearson’s coefficient of skew?

Answer 2

3(Mean - Median) / standard deviation

Question 3

What does it mean if Pearson’s coefficient of skew is <0?

Answer 3

Negatively skewed

Question 4

What does it mean if Pearson’s coefficient of skew is >0?

Answer 4

Positively skewed

Question 5

What does a Gaussian curve mean?

Answer 5

From mean and standard deviation of data alone, can predict y value for any x value

Question 6

What do parametric tests assume?

Answer 6

Normal distribution

Values such as mean and standard deviation accurately reflect population distribtuion

Question 7

What are examples of statistical tests?

Answer 7

t-test
- one-sample
- independent
- paired

ANOVA
- factorial
- one-way

Correlation

Regression

Question 8

What does a normal distribution say about percentages of population within certain standard deviations?

Answer 8

68% of population within (mean +/- 1 * SD)

95% of population within (mean +/- 2 * SD)

99.7% of population within (mean +/- 3 * SD)

Question 9

What does transforming data into z scores do?

Answer 9

Helps standardise data and reduce impact of skewness

Can tell us how many standard deviations someone was from the mean

Question 10

What are the pros of z scores?

Answer 10

Can transform data to standardised scale

Scale adheres to normal distribution

Can compare things relative to their own population

Use the entire dataset

Question 11

What is the standard error?

Answer 11

Tells how likely it is that our sample will vary from one sample to another

How confident are we that we know the true population mean?

Use SE of the mean to say how confident we are that our sample values represent the population

Smaller = better

Question 12

What are the largest influences on the standard error?

Answer 12

Variability of original data (standard deviation of population)

Total N used to create sample mean

Question 13

What is a confidence interval?

Answer 13

Range of values that, in a certain proportion of the samples, contain true value of a statistic (e.g. mean)

Can be used for visualisation - error bars

Question 14

What are error bars?

Answer 14

Can be SD, SEM or CI but must be explicit which one is being used

Non-overlapping SEM bars often imply significant differences between conditions