Example 2
STATISTICAL ANALYSIS TO INVESTIGATE XXX
INTERVENTION
INTRODUCTION
Statistics is a form of analysis that uses quantified representations and synopses for a given set
of real-life studies (Scott & Mazhindu, 2014). It is especially important in public health as it
enables translation of numerical data into information about cause and effect, health risks and
the effectiveness of interventions (Friedman, Hunter & Parrish, 2005). However, because
statistical data is often secondary, it is open to misinterpretation and must therefore be analysed
critically to come to valid conclusions (Little & Rubin, 2019). There are a variety of statistical
software commonly used in public health such as SAS, Stata, Epi Info and SPSS (Sullivan,
Dean & Soe, 2009). However, for this report, SPSS will be used as that is the form in which
the data entered was provided. Additionally, it has pre-defined statistical analysis tools and
offers the capability of various descriptive and inferential techniques (Taylor et al, 2016) that
will efficiently address the questions.
Rationale
Using the xxx data presented in SPSS, the purpose of this report is to summarize, interpret,
critically discuss and employ the appropriate statistical techniques to address the given
questions. First, under methods, a discussion on preliminary investigations will be held and
analyses will be presented in the form of descriptive techniques. Subsequently, a critical
discussion on inferential statistical techniques and finally, each of the questions will be
critically addressed and presented accordingly.
1
METHODS
Preliminary analyses and investigations
Prior to conducting data analysis, it is crucial to screen and clean the dataset for errors involving
missing data, data validity and out-of-range values (Tien, 2008). This prevents errors in the
interpretations and assumptions derived from the statistical analyses to be conducted (Barton
& Peat, 2014). This can be done through descriptive statistics (DS), which organizes,
summarizes and describes measures of a sample to provide an initial impression of the data,
that then informs the appropriate analytical techniques (Weiss & Weiss, 2012). It differs from
inferential statistics, which is used to draw conclusions and sometimes make predictions about
the properties of a population based on the representative sample (Sahu, Pal & Das, 2015). The
choice of which descriptive or inferential method is suitable to use depends on the type of
variables (Lewis-Beck, Bryman & Liao, 2004). Categorical variables
…………………………………………………………………………………………………
…………………………………………………………………………………………………
…………………………………………………. (Agresti, 2018). On the other hand, numerical
variables represent quantitative data a… … … … … …
…………………………………………………………………………………………………
…………………………………………………………………………………………………
…………………………………………………………………………………………………
………………………………………………… (Ali & Bhaskar, 2016).
Categorical data is ideally reported in form of frequencies, percentages, bar graphs or pie charts
(Cox, 2018). On the other hand, numerical data is presented through measures of central
tendency such as mean, median and mode or measures of dispersion such as the standard
deviation, variance, minimum and maximum values (Bickel & Lehmann, 2012). Although both
methods summarize distribution, the former indicates the central point of the distribution while
2
the latter indicates how observations vary around that middle (Altman, 2000). As such, DS was
conducted accordingly to identify errors, give a general picture of the data and form the basis
of the preliminary investigations as part of the more extensive statistical analysis to be
conducted.
Descriptive analysis
Table 1.2 continuous variables
N Range Minimum Maximum Mean Std.
Deviation
Variable 1
Variable 2
Variable 3
Variable 4
Valid N
Pie chart here
Figure 2. pie chart showing intervention status (figure is removed to minimise potential for inadvertent
plagiarism)
Important to note that the analysis revealed an error in form of a missing value in the xxx data.
Having summarized the data appropriately, the section that follows will be a discussion on the
broad inferential statistical techniques to be conducted for questions x-y.
3
Inferential Statistical Techniques
Parametric and Non-Parametric tests
There are 2 main types of inferential statistical techniques used to investigate hypotheses:
parametric and non-parametric tests (Pallant, 2016). Whereas parametric tests require several
assumptions of validity to be met in order to come to reliable conclusions, non-parametric tests
are used specifically when these assumptions are violated (Adams & Lawrence, 2018). These
assumptions include: a large sample size, continuous measurement scale (for the dependent
variable), independence of observations, normal distribution and homogeneity of variance
(Field, 2018). Conversely, non-parametric tests are applicable to a small data set, a wider range
of data including nominal, ordinal, interval or data with outliers and non-normally distributed
data (Ghasemi & Zahediasl, 2012).
Whereas both techniques are widely used to test hypotheses, it is possible to come to wrong
conclusions, known as type 1 and 2 errors (Aberson, 2019). … … … … … … …
…………………………………………………………………………………………………
…………………………………………………………………………………………………
…………………………………………………………… (Cohen, 2013).
According to Pallant (2016), parametric tests have greater statistical power as when an effect
exists, they are more likely to detect it and hence often considered more robust than non-
parametric tests. Field (2018) argues … … …. … … ……..
…………………………………………………………………………………………………
…………………………………………………………………………………………………
…………………………………………………………………………………………………
……………………….…..
4
(Wilcox, 2009; Warner, 2008). However, Salkind (2004) and Pallant (2016) note that most
researches indeed suggest n>30 because it is the minimum required before an analysis based
on normal distribution can be considered valid. Thus, this report will assume large=n>30.
From the discussion above, it is evident that the strength and reliability of statistical analysis
strongly relies on understanding and choosing the appropriate test. Choosing the right statistical
technique necessitates an understanding of the nature of variable (independent or otherwise),
measurement scale and underlying distribution. These assumptions will be checked prior to
conducting any statistical test. The specific tests and their additional assumptions (if any) will
be discussed while addressing each of the questions.
QUESTION x: Determine whether there are any … … … … … … … … … … … … …
…………
As was briefly discussed in the preceding section, understanding the nature of variable is key
to choosing the appropriate test. Both parametric and non-parametric tests are designed in such
a way that it is either for paired or independent data and this is crucial to producing accurate
results (Kim, 2015). Data is said to be independent when the sets of data are derived from
separate individuals or groups (Peacock, Kerry & Balise, 2017). Contrariwise, data is paired
when derived from the same individual but on 2 different occasions (Derrick, Toher & White,
2017). Using this definition, the groups in question, intervention and control, are unrelated,
therefore independent groups.
…………………………………………………………………………………………………
…………………………………………………………………………………………………
…………………………………………………………………………………………………
…………………………………………………………………………………………………
…………………………………………………………………………………………………
5
…………………………………………………………………………………………………
…………………………………………………………………………………………………
……………………………. However, the appropriate choice strongly relies on the
assumptions discussed earlier. As such, before proceeding to the formal statistical testing, the
data will be checked for normal distribution and all other relevant assumptions.
Assumption analysis
There are 2 commonly used graphical techniques through which normal distribution can be
investigated: the histogram and quantile-quantile (Q-Q) plot (D’Agostino, 2017). With a
histogram, normal distribution is typically demonstrated if the frequency distribution shows a
symmetrical bell-shaped curve (Dancey, Reidy & Rowe, 2012). However, while it is a good
technique, sometimes it can be interpreted differently by different people and in that case, a Q-
Q plot would provide more clarity (Rayat, 2018). A Q-Q plot demonstrates normal distribution
when the points approximately fall in a straight line (Tufte, 2001).
With this in mind, both a histogram and Q-Q plot were constructed, and normal distribution
was demonstrated as seen below.
(figures removed to minimise potential for inadvertent plagiarism)
Based on the above discussions, with a large sample size (n=xx), normal distribution and 2
independent groups, the data satisfied parametric assumptions, therefore an independent
samples t-test was deemed appropriate as seen below.
6
Test analysis
(table removed to minimise potential for inadvertent plagiarism)
Results
An independent t-test was conducted to determine whether a difference in xxx between the
intervention and control groups exists. With p=x.xxx, the analysis revealed a statistically
significant difference in xxx between the intervention group (M=xxx, SD=xxx) compared to
the control group (M=xxx, SD=xxx). The magnitude of mean difference xxx) between the 2
groups was (t(xx)= xxx; p=xxx; 95% CI, xxx - xxx).
QUESTION 2: …………………………………………………………….
CONCLUSION
This report set out to summarize, interpret and apply appropriate statistical techniques to
investigate the …………. through addressing the given questions. Following critical statistical
analysis using SPSS, the effect of the intervention on …… and …. was mostly found to be
statistically significant. However, it is important to note that p-value and confidence interval
do not indicate the size, precision and strength of the significance and the results do not
necessarily mean the effect is real. The appropriate inferential techniques facilitated
comparison of risk between the intervention and control groups; as well as prediction of
relationships amongst variables. All these statistical techniques are essential to evaluating the
impact of public health interventions.
7
REFERENCES
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXBROMBIN,
8
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
