This lecture introduces quantitative data analysis methods for nursing research. It covers different types of data, ways to summarize and present descriptive data, including measures of central tendency, dispersion, and common charts. The lecture also discusses inferential statistics, the normal distribution, sampling error, and confidence intervals.
MSc Nursing Research Methods Quantitative Data Analysis
1. MSc/Dip/Cert Advancing Nursing Practice
MSc by Research (Nursing)
RESEARCH METHODS IN NURSING AND
NURSING STUDIES
NURSING STUDIES
HEALTHCARE (B)
QUANTITATIVE DATA ANALYSIS 1
Dr. Sheila Rodgers
Nursing Studies
University of Edinburgh
2. MSc/Dip NURSING Research Methods
This lecture aims to:
NURSING STUDIES
NURSING STUDIES
•Introduce the different types of data
•Look at ways of summarising and presenting
descriptive data
•Explain measures of dispersion
•Introduce the principles of inferential
statistics
3. NURSING STUDIES MSc/Dip NURSING Research Methods
NURSING STUDIES
Types of data
NOMINAL discrete, no true value (yes/no)
ORDINAL categories in an order (staff grade)
INTERVAL count or category of equal
spaces (age group, no. children)
RATIO equal interval, fixed zero, continuous
(age VAS)
4. MSc/Dip NURSING Research Methods
PRESENTING DATA
NURSING STUDIES
NURSING STUDIES
Tables – frequency, contingency
Charts –
Pie or bar charts for nominal data
Histograms for numerical or ratio data
Line graphs and box plots for numerical data
but not frequencies
Scatter plots for numerical and ordinal data
9. • Box plots
Clinical peripherality scores
by NHS Board area (Swan et
al., 2004)
The box plot shows median
values and interquartile range of
scores for each NHS board
area. Higher values represent
greater clinical peripherality. A
summary plot of clinical
peripherality scores for non-
urban practices in each NHS
Board area is shown in this
figure. NHS Boards serving the
more remote and rural areas of
Scotland show greater median
values and a wider scatter of
clinical peripherality values for
their practices.
A detailed analysis can be
found in the Rural Action Team
Report.
http://www.show.scot.nhs.uk/sehd/nationalframework
11. NURSING STUDIES MSc/Dip NURSING Research Methods
NURSING STUDIES
MEASURES OF CENTRAL TENDENCY
Mean
Median
Mode
12. NURSING STUDIES MSc/Dip NURSING Research Methods
NURSING STUDIES
MEASURES OF VARIATION OR DISPERSION
Range, inter/semi quartile
Standard deviation – average distance of
each measurement from the mean
Variance – SD before the square root is taken
13. NURSING STUDIES MSc/Dip NURSING Research Methods
NURSING STUDIES
NORMAL DISTRIBUTION
Bell shape curve reaching to infinity
Uni-modal / Bi-modal
Positive or negative skews
14. MSc/Dip NURSING Research Methods
NORMAL DISTRIBUTION
NURSING STUDIES
NURSING STUDIES
15. NURSING STUDIES MSc/Dip NURSING Research Methods
NURSING STUDIES
16. NURSING STUDIES MSc/Dip NURSING Research Methods
NURSING STUDIES
17. NURSING STUDIES MSc/Dip NURSING Research Methods
NURSING STUDIES
18. NURSING STUDIES MSc/Dip NURSING Research Methods
NURSING STUDIES
20. NURSING STUDIES MSc/Dip NURSING Research Methods
NURSING STUDIES
SAMPLING ERROR
SE = SD divided by the square root of n
CONFIDENCE INTERVALS
95% chance of having the true population mean
within a certain range:
Mean + (1.96 x SE) AND Mean - (1.96 x SE)
21. NURSING STUDIES MSc/Dip NURSING Research Methods
NURSING STUDIES
22. NURSING STUDIES MSc/Dip NURSING Research Methods
NURSING STUDIES
Notas del editor
Data has been described in different categories according to it properties. Nominal – discrete categories by name, no true value, male female Ordinal – categories in some sort of order, no information about how big the spaces are between each category, maybe a likert scale Interval counts or categories with equal spaces between them, no kids, age groups, no dressing packs used Ratio equal intervals may be continuous has a fixed zero, temp, age vas cost of dressing packs It is important to know what type of data you are dealing with in order to work out if it has been presented and analysed correctly. There are rules as to how different types of data should be presented and analysed. When reading and reviewing papers you shuodl look to see if these rules have been followed.
How we present that data is really important as we want readers to clearly understand what comes out of the data. Whilst it is possible to give tables with large amounts of data, it can often be difficult to make sense of lots of rows and columns of data. Looking at a picture of how that data can be represented or summarised allows us to take in much more information, to make sense of it and to spot trends in the data. Certain types of data are best presented in certain ways.
Nice that is 3D and coloured But there is no key MCI = Mild Cognitive Impairment, no actual numbers are given nor the full total number of people in the study. There could be 36 people in this study or 3,000.
This chart summarises a lot of information in one place about different types of day care provision in Scotland. Colours are used well to separate out different types of data and yet show the relation between categories. There are spaces between sets of bars from different client groups to show that the data is not continuous. A common error when presenting bar charts is to have the bars touching each other implying that the data follows sequentially from on e bar to the next. This is only true in histograms when plotting out data that is at least at interval level.
In this histogram, two different populations are presented, international and UK nurses. They are quite rightly shown as two different colours. They clearly show the trend for a marked increase in the number of international nurses registering in the UK with a slight decline in 2002/3. There is a space between the bars because the two different categories of nurses are being kept separate.
In this line graph, both colours and shape markings on the lines are used to try and differentiate between the different sets of data. However the colours are not that distinguishable and the markings very hard to read. It is easy however to spot the trend in this data and also see the rate of change over time.
This box plot looks at the different regions of Scotland in terms of how remote and rural their health services are. A scale was created to indicate this and ranged from 0 to 100. The higher the score, the more widely dispersed is their health care provision. This type of graph is a good way of presenting numerical data which might have quite a range of values within it. For example, in Lothian (Edinburgh is the heart of this region), the majority of the services are quite centralised with the average score for Lothian’s services being 20 – the middle of the box. However, the ends of the lines represent the top and bottom 25% of the whole data set which shows here that over 25% of Lothian’s services have a high score of rurallity and dispersion at over 60.
Scatter plots can be useful to look at the relationships between two sets of data that are both numerical and of at least ordinal value. It is much easy to interpret how scores on one variable might change when scores on the other variable change. In this example we can see how when arterial blood pH is high that capillary blood pH also tends to be high. This study was trying to assess if capillary blood sampling could be used for pH measurement rather than taking arterial blood. What do you think ?
Interquartile range to show middle 50% of values around the median SD is the square root of variance so that the average variation is shown in the correct units Variance – sum (X-mean) 2 n-1
Infection control paper Ethics do we have the right to use patient record data – ethics committees may give to people to use