Quantitative Data Analysis
otherwise known as
2009
Tracy Culkin
Learning Outcomes
• Aim
to be introduced to the concept
of Statistics and their
application to the Research
Process
• Objectives – by the end of this
session you should be able to
• Give a definition of Statistics
• Identify the Research
Approach to which it belongs
• Describe why a knowledge of
statistics is necessary
What do you know about Statistics?
Definition
• Statistics
the body of knowledge which
is concerned with the
collection, organisation,
analysis, interpretation and
presentation of numerical data
(Wickham 2006, p 146)
Definition
• Statistics
the whole subject involved with
collecting and analysing real-
life data, to summarise it and
to display it, also to draw
inferences from it to assist
decision making
(Watson et al, 2006, p 197)
Types of Statistics
• Descriptive Statistics – organisation of
data by tables and graphs and can also
summarise the data collected
• Most often used with Surveys
• Highlights the most interesting findings
• Inferential Statistics – Infer conclusions
about a population from sample data
• Most often used with Randomised Control
Trials
• Allows the researcher to move from what
they know in their study to predict what will
happen in similar situations
Descriptive Statistics
• Can provide a quick way of seeing the details and results of a
sample or study
• These statistics allow a lot of figures to be clearly presented,
compared and tracked e.g. Maternity units present birth statistics
Birth Statistics for St Bloggs Hospital (RCM 2004, P 39)
January February March
Total Births 200 (100%) 220 (100%) 180 (100%)
Normal Births 150 (75%) 150 (68%) 140 (77%)
Instrumental Births 20 (10%) 20 (9%) 10 (5.5%)
Caesarean Section
(Emergency)
20 (10%) 20 (9%) 10 (5.5%)
Caesarean Section
(Elective)
10 (5%) 30 (14% 20 (11%)
Inferential Statistics
• Are used to test Hypotheses (a statement
of facts that has yet to be proved,
disproved, supported or rejected)
• This involves selecting a small sample of
people for study and from the results of this,
allowing the researcher to make inferences
(assumptions) about the population
• They are the ‘bread & butter’ of randomised
control trials
Randomised Control Trials
(RCTs)
• Definition
‘An experimental design characterised by the independent variable.
Participants are randomly assigned to the experimental group or
control group’
(Lanoe 2005, p 96)
 Independent variable – is the experimental factor that is
deliberately manipulated
 Dependent variable – is the aspect being studied to see if the
experimental factor has any effect
Examples of independent and
dependent variables
• Does one-to-one midwifery care reduce the length
of labour?
• One-to-one midwifery care is the independent variable
• Length of labour is the dependent variable
• The independent variable is what is being tested
(Cluett 2000, p32)
Randomised Control Trial
Identify potential participant
Ensure eligibility for the study
Obtain informed consent
No Yes
Exclude Randomize
Treatment group (1 to 1 care) Control Group
The language of Statistics
The language of Statistics
• Number are usually classified into groupings –
known as the level of measurement -:
 Categorical -:
Nominal – data described with the use of
numbers, allocates subjects to their named
categories e.g. gender (1. Male 2.Female)
Ordinal – this puts the number into rank order e.g.
hotel star ratings
 Numerical -:
Interval –a way of measuring numbers, data is put
into measurable portions with a known and set
distance between them e.g. 2-4, 6-8
Ratio –largest set of numbers, these are the
numbers that can be analysed with statistical tools.
Different from ‘interval’ in that an absolute zero is
present e.g. weight or height are ratio variables
The language of Statistics
• Parametric and Non-Parametric data
• Parametric – describes data that meet certain
parameters in relation to the way that data are
distributed in the population as a whole.
• Interval & Ratio data are said to be parametric
(e.g. they can measure height, weight)
• Non-parametric –do not meet the same
criteria around things like distribution – cannot
be added, subtracted, multiplied or divided
• Ordinal & Nominal data (e.g. like the numbers
on the footballers’ shirts)
The language of Statistics
• Mean – the arithmetic average
• Mode – the score that occurs
most frequently in a set of
scores
• Median – the number that
occurs in the middle of an
ordered sequence of scores
The language of Statistics
• Probability = P value
• This determines whether the results from an
experimental study have occurred by chance or
that there is a true difference.
• In experimental research participants are divided
into groups and sometimes by chance alone
differences can occur
• The probability table determines the level of
statistical significance of these differences and
informs the researcher if the study’s findings have
occurred by chance or not.
Probability
• P values are represented on a scale from 0
to 1
• 0 represents no chance of the results
having come about by chance
• 1 would mean they definitely came about
by chance
• If p value is 0.1 – this means there is a 1 in
10 chance that the results came about by
chance
• If p value is 0.5, there is a 1 in 2 chance
• What to look for – if p value is 0.05 or less,
then the results of the study are considered
statistically significant
• p value 0.005 considered to be highly
significant
Class Exercise
An Example of Statistical Tests
• ANOVA (Analysis of Variance)
• Chi-squared test (X2)
• Kruskal –Wallis
• Mann-Whitney U-test
• McNemar test
• Pearson’s coefficient
• Spearman’s coefficient
• t-test
• Wilcoxon test
Statistical Software Packages
• SPSS – Statistical Package for the Social
Sciences
• Allows for much easier handling of data
• However, researcher can become lost in
the maze of analyses available
• Therefore, need to ensure
 You use the correct analysis tool
 You understand what the requested
analysis does
 You understand all the output
• If no to any of the above – you will not
know if the computer is giving you valid
results
• Or ‘garbage’ (Jordan et al, 1998, p 61)
Why a knowledge of Statistics is
important to the researcher
• To ensure that the way in which the data is collected,
will make the data easy to manage
• That enough of the right kind of data is collected in the
most accurate way
• Organisation of the data makes it easier to handle
• Once the data is collected and organised, it needs to
be analysed using specific statistical tests
• Statistical tests determine the results of the study and
how likely they are to be significant
• The researcher then needs to look at the results in
relation to the wider picture
References
• Devane, D et al, (2004) Methodological Issues in Nursing Research. Journal of Advanced
Nursing. 47 (3), 297-301
• Hicks, C M (1996) Undertaking Midwifery Research. Churchill Livingstone: Edinburgh
• Royal College of Midwives (2004) Using Research in Practice A Resource for Midwives. RCM:
London
• Watson, R et al, (2006) Successful Statistics for Nursing & Healthcare. Palgrave Macmillan:
Basingstoke
• Wickham S (2006) Appraising Research into Childbirth. Elsevier: Edinburgh
Analysis

Analysis

  • 1.
    Quantitative Data Analysis otherwiseknown as 2009 Tracy Culkin
  • 2.
    Learning Outcomes • Aim tobe introduced to the concept of Statistics and their application to the Research Process • Objectives – by the end of this session you should be able to • Give a definition of Statistics • Identify the Research Approach to which it belongs • Describe why a knowledge of statistics is necessary
  • 3.
    What do youknow about Statistics?
  • 4.
    Definition • Statistics the bodyof knowledge which is concerned with the collection, organisation, analysis, interpretation and presentation of numerical data (Wickham 2006, p 146)
  • 5.
    Definition • Statistics the wholesubject involved with collecting and analysing real- life data, to summarise it and to display it, also to draw inferences from it to assist decision making (Watson et al, 2006, p 197)
  • 6.
    Types of Statistics •Descriptive Statistics – organisation of data by tables and graphs and can also summarise the data collected • Most often used with Surveys • Highlights the most interesting findings • Inferential Statistics – Infer conclusions about a population from sample data • Most often used with Randomised Control Trials • Allows the researcher to move from what they know in their study to predict what will happen in similar situations
  • 8.
    Descriptive Statistics • Canprovide a quick way of seeing the details and results of a sample or study • These statistics allow a lot of figures to be clearly presented, compared and tracked e.g. Maternity units present birth statistics Birth Statistics for St Bloggs Hospital (RCM 2004, P 39) January February March Total Births 200 (100%) 220 (100%) 180 (100%) Normal Births 150 (75%) 150 (68%) 140 (77%) Instrumental Births 20 (10%) 20 (9%) 10 (5.5%) Caesarean Section (Emergency) 20 (10%) 20 (9%) 10 (5.5%) Caesarean Section (Elective) 10 (5%) 30 (14% 20 (11%)
  • 9.
    Inferential Statistics • Areused to test Hypotheses (a statement of facts that has yet to be proved, disproved, supported or rejected) • This involves selecting a small sample of people for study and from the results of this, allowing the researcher to make inferences (assumptions) about the population • They are the ‘bread & butter’ of randomised control trials
  • 10.
    Randomised Control Trials (RCTs) •Definition ‘An experimental design characterised by the independent variable. Participants are randomly assigned to the experimental group or control group’ (Lanoe 2005, p 96)  Independent variable – is the experimental factor that is deliberately manipulated  Dependent variable – is the aspect being studied to see if the experimental factor has any effect
  • 11.
    Examples of independentand dependent variables • Does one-to-one midwifery care reduce the length of labour? • One-to-one midwifery care is the independent variable • Length of labour is the dependent variable • The independent variable is what is being tested (Cluett 2000, p32)
  • 12.
    Randomised Control Trial Identifypotential participant Ensure eligibility for the study Obtain informed consent No Yes Exclude Randomize Treatment group (1 to 1 care) Control Group
  • 13.
    The language ofStatistics
  • 14.
    The language ofStatistics • Number are usually classified into groupings – known as the level of measurement -:  Categorical -: Nominal – data described with the use of numbers, allocates subjects to their named categories e.g. gender (1. Male 2.Female) Ordinal – this puts the number into rank order e.g. hotel star ratings  Numerical -: Interval –a way of measuring numbers, data is put into measurable portions with a known and set distance between them e.g. 2-4, 6-8 Ratio –largest set of numbers, these are the numbers that can be analysed with statistical tools. Different from ‘interval’ in that an absolute zero is present e.g. weight or height are ratio variables
  • 15.
    The language ofStatistics • Parametric and Non-Parametric data • Parametric – describes data that meet certain parameters in relation to the way that data are distributed in the population as a whole. • Interval & Ratio data are said to be parametric (e.g. they can measure height, weight) • Non-parametric –do not meet the same criteria around things like distribution – cannot be added, subtracted, multiplied or divided • Ordinal & Nominal data (e.g. like the numbers on the footballers’ shirts)
  • 16.
    The language ofStatistics • Mean – the arithmetic average • Mode – the score that occurs most frequently in a set of scores • Median – the number that occurs in the middle of an ordered sequence of scores
  • 17.
    The language ofStatistics • Probability = P value • This determines whether the results from an experimental study have occurred by chance or that there is a true difference. • In experimental research participants are divided into groups and sometimes by chance alone differences can occur • The probability table determines the level of statistical significance of these differences and informs the researcher if the study’s findings have occurred by chance or not.
  • 18.
    Probability • P valuesare represented on a scale from 0 to 1 • 0 represents no chance of the results having come about by chance • 1 would mean they definitely came about by chance • If p value is 0.1 – this means there is a 1 in 10 chance that the results came about by chance • If p value is 0.5, there is a 1 in 2 chance • What to look for – if p value is 0.05 or less, then the results of the study are considered statistically significant • p value 0.005 considered to be highly significant
  • 19.
  • 20.
    An Example ofStatistical Tests • ANOVA (Analysis of Variance) • Chi-squared test (X2) • Kruskal –Wallis • Mann-Whitney U-test • McNemar test • Pearson’s coefficient • Spearman’s coefficient • t-test • Wilcoxon test
  • 22.
    Statistical Software Packages •SPSS – Statistical Package for the Social Sciences • Allows for much easier handling of data • However, researcher can become lost in the maze of analyses available • Therefore, need to ensure  You use the correct analysis tool  You understand what the requested analysis does  You understand all the output • If no to any of the above – you will not know if the computer is giving you valid results • Or ‘garbage’ (Jordan et al, 1998, p 61)
  • 23.
    Why a knowledgeof Statistics is important to the researcher • To ensure that the way in which the data is collected, will make the data easy to manage • That enough of the right kind of data is collected in the most accurate way • Organisation of the data makes it easier to handle • Once the data is collected and organised, it needs to be analysed using specific statistical tests • Statistical tests determine the results of the study and how likely they are to be significant • The researcher then needs to look at the results in relation to the wider picture
  • 24.
    References • Devane, Det al, (2004) Methodological Issues in Nursing Research. Journal of Advanced Nursing. 47 (3), 297-301 • Hicks, C M (1996) Undertaking Midwifery Research. Churchill Livingstone: Edinburgh • Royal College of Midwives (2004) Using Research in Practice A Resource for Midwives. RCM: London • Watson, R et al, (2006) Successful Statistics for Nursing & Healthcare. Palgrave Macmillan: Basingstoke • Wickham S (2006) Appraising Research into Childbirth. Elsevier: Edinburgh

Editor's Notes

  • #10 Descriptive stats allow the researcher to make statements only about the results obtained, whereas inferential statistics allow the researcher to make assumptions beyond the set of data
  • #19 0.05 = 1 in 20 chance but this is the accepted. 0.005 1 in 200 chance