RDS Exam

Question 1

What is a hypothesis?

Answer 1

A statement that predicts the finding of a research study

Question 2

What is a p-value

Answer 2

Probability that an observed difference between groups (e.g. fracture incidence) was observed by chance.

Question 3

What does the alpha a -value represent?

Answer 3

Probability of rejecting the hypothesis when it is true. It is the value at which you reject the hypothesis when the p-value is less than it.

Question 4

What is a research question?

Answer 4

The question the research is trying to answer. The research question provides the frame for the entire research project

Question 5

What are the different types of data?

Describe each

Answer 5

Quantitative – Can be split into counted (discrete e.g. number of children in a family) and measured (continuous e.g. temperature)

Qualitative – So like descriptive and free text

Categorical (kinda between qual and quant) - Nominal (individual terms like colours or gender) and ordinal (where there is a scale and can be converted into numbers)

Question 6

How do you test for normality?

Answer 6

Kolmogorov Smirnov test for sample size above 50

Shapiro-Wilks test for sample size below 50

Question 7

What is skewness? Describe it in terms of positive and negative skew and where the medians and modes and means are.

Answer 7

Asymmetric distribution with many data points in the high or low end. Left-skewed is also negatively skewed (long tail is on left or lower end). Right skew is also called positive skew. The mode is the highest point, mean is towards on the side of the long tail and the median is between mode and mean.

Question 8

What is kurtosis?

Answer 8

Kurtosis describes data that are heavy-tailed or light-tailed relative to a normal distribution. They either make a ‘wider’ normal distribution (heavy kurtosis) or a ‘narrow’ normal distribution (low kurtosis)

Question 9

What is the variance?

Answer 9

A measure of the spread of the numbers away from the mean value.

Question 10

What is standard deviation?

Answer 10

Square root of the variance. Measures the spread of a set of data

Question 11

What is the range?

Answer 11

difference between largest data value and smallest data value. Measures how far a set of number are spread out from their average value.

Question 12

What is the interquartile range?

Answer 12

Q3 - Q1.

Question 13

What is the standard error of mean?

Answer 13

Standard deviation divided by the square root of the sample size. Measures how well the sample mean approximates to the population mean.

Question 14

What are confidence intervals?

Answer 14

Standard error multiplied by 1.96

Question 15

When would you use a chi-square test?

Answer 15

When comparing two discrete sets of data. Either finding differences or similarities (you also know how to do a chi-square test: i.e. finding “expected” values and comparing those with the “observed” values and seeing if there is a significant difference)

Question 16

When would you use a Wilcoxon test?

Answer 16

To compare two non-parametric sets of paired data to find differences.

Question 17

When would you use a Mann Whitney test?

Answer 17

To compare two non-parametric sets of unpaired data to find differences.

Question 18

When would you use an independent t-test?

Answer 18

To compare two parametric sets of unpaired data to find differences.

Question 19

When would you use a paired t-test?

Answer 19

To compare two parametric sets of paired data to find differences.

Question 20

When would you use a Kruskal Wallis test?

Answer 20

To test for differences between 3 or more sets of non-parametric data, which are paired

Question 21

When would you use ANOVA? What are the different types?

Answer 21

Compare differences in means of 3 or more sets of parametric data
One-way ANOVA tests unpaired sets
Repeated measures one-way ANOVA compare paired sets

Question 22

When would you use a Friedman test?

Answer 22

To test for differences between 3 or more sets of non-parametric data, which are unpaired

Question 23

What is the difference between correlation and regression?

Answer 23

They both look at similarities between two sets and the relationship between two variables.
Correlation is looking at the strength of the association or relationship between two variables.
Regression looks at what happens to one variable when you change the other variable by a certain amount. - becomes linear regression y = mx + c

Question 24

How do you test correlation in a parametric data sets?

Answer 24

Pearson's Correlation test: returns an r-value (can be either negative or positive):
= 0-0.2 => very low correlation
0.2-0.4 => low correlation
0.4-0.6 => reasonable correlation
0.6-0.8 => high correlation
0.8-1.0 => very high correlation

Can also return a r^2 value which represents how closely your data is fitted to the correlation line. (same rule, higher r squared value means line is more reliable).

Question 25

How do you test correlation in a non-parametric data sets?

Answer 25

Spearman rank correlation gives two things:

• A correlation coefficient (Spearman’s rho, denoted by ρ) is the equivalent of the Pearson r-value.
• The p-value, once again, tells you how reliable the rho-value is. The smaller the p-value, the more reliable the rho-value

Question 26

How is regression defined?

Answer 26

By the equation y = mx + c (or y = a + bx)
c = y-intercept
m = gradient

Question 27

What do you use as a measure of centrality in a normally distributed set of data? and in a non-normally distributed set?

Answer 27

Normal: mean

Non-normal: median (as the skew could cause the mean to be unrepresentative of the centre)

Question 28

When to use a pie chart?

Answer 28

To show relative sizes of the whole or portions.

Question 29

When to use a bar chart?

Answer 29

Comparative data where one data set is categorical and the other is numerical

Question 30

When to use a scatter plot?

Answer 30

To identify similarities or a relationship between two continuous variables

Question 31

What are the values on a box and whisker plot?

Answer 31

Max value
Q3
Median
Q1
Min value

Question 32

When to use a histogram?

Answer 32

Comparing two data sets, one of which is quantitative counted (e.g. frequency) and one is quantitative continuous/measured

Question 33

When to use a Dot-plot?

Answer 33

Graphical display of data using dots. It is very similar to bar-chart but the bars have been replaced by dots. Provides better visualisation of spread/dispersion of data but used for smaller data sets.

Question 34

How do you calculate odds ratio?

Answer 34

Odds ratio = (number of cases/number of controls in exposed group) ÷ (number of cases / number of controsl in unexposed group)

Question 35

How do you calculate relative risk?

Answer 35

RR = (number of cases / total in exposed group) ÷ (number of cases / total in unexposed group)