This single copy is for your personal, non-commercial use only.

For permission to reprint multiple copies or to order presentation-ready copies for distribution, contact CJHP at [email protected]

RESEARCH PRIMER

Creating a Data Analysis Plan: What to

Consider When Choosing Statistics for a Study
Scot H Simpson

There are three kinds of lies: lies, damned lies, and statistics. or gender, ethnicity, exercise frequency, weight, treatment group,
– Mark Twain1 and blood glucose. Each variable will have a group of categories,
which are referred to as values, to help describe the characteristic
of an individual study participant. For example, the variable “sex”
INTRODUCTION
would have values of “male” and “female”.

S tatistics represent an essential part of a study because, regard-

less of the study design, investigators need to summarize the
collected information for interpretation and presentation to
Although variables can be defined or grouped in various
ways, I will focus on 2 methods at this introductory stage. First,
variables can be defined according to the level of measurement.
others. It is therefore important for us to heed Mr Twain’s The categories in a nominal variable are names, for example, male
concern when creating the data analysis plan. In fact, even before and female for the variable “sex”; white, Aboriginal, black, Latin
data collection begins, we need to have a clear analysis plan American, South Asian, and East Asian for the variable “ethni-
that will guide us from the initial stages of summarizing and city”; and intervention and control for the variable “treatment
describing the data through to testing our hypotheses. group”. Nominal variables with only 2 categories are also referred
The purpose of this article is to help you create a data analysis to as dichotomous variables because the study group can be
plan for a quantitative study. For those interested in conducting divided into 2 subgroups based on information in the variable.
qualitative research, previous articles in this Research Primer For example, a study sample can be split into 2 groups (patients
series have provided information on the design and analysis of receiving the intervention and controls) using the dichotomous
such studies.2,3 Information in the current article is divided into variable “treatment group”. An ordinal variable implies that the
3 main sections: an overview of terms and concepts used in data categories can be placed in a meaningful order, as would be the
analysis, a review of common methods used to summarize study case for exercise frequency (never, sometimes, often, or always).
data, and a process to help identify relevant statistical tests. My Nominal-level and ordinal-level variables are also referred to as
intention here is to introduce the main elements of data analysis categorical variables, because each category in the variable can be
and provide a place for you to start when planning this part of completely separated from the others. The categories for an
your study. Biostatistical experts, textbooks, statistical software interval variable can be placed in a meaningful order, with the
packages, and other resources can certainly add more breadth interval between consecutive categories also having meaning.
and depth to this topic when you need additional information Age, weight, and blood glucose can be considered as interval
and advice. variables, but also as ratio variables, because the ratio between
values has meaning (e.g., a 15-year-old is half the age of a
TERMS AND CONCEPTS USED IN DATA 30-year-old). Interval-level and ratio-level variables are also
referred to as continuous variables because of the underlying
ANALYSIS
continuity among categories.
When analyzing information from a quantitative study, we As we progress through the levels of measurement from
are often dealing with numbers; therefore, it is important to begin nominal to ratio variables, we gather more information about
with an understanding of the source of the numbers. Let us start the study participant. The amount of information that a variable
with the term variable, which defines a specific item of informa- provides will become important in the analysis stage, because we
tion collected in a study. Examples of variables include age, sex lose information when variables are reduced or aggregated—a

C J H P – Vol. 68, No. 4 – July–August 2015 J C P H – Vol. 68, no 4 – juillet–août 2015 311
 This single copy is for your personal, non-commercial use only.
For permission to reprint multiple copies or to order presentation-ready copies for distribution, contact CJHP at [email protected]

common practice that is not recommended.4 For example, if age examine the relative number of participants within each category
is reduced from a ratio-level variable (measured in years) to an (by reporting the percentages within each category), whereas a
ordinal variable (categories of < 65 and ≥ 65 years) we lose the bar graph can also be used to examine absolute numbers. For
ability to make comparisons across the entire age range and example, we could create a pie graph to illustrate the proportions
introduce error into the data analysis.4 of men and women in a study sample and a bar graph to illustrate
A second method of defining variables is to consider them the number of people who report exercising at each level of
as either dependent or independent. As the terms imply, the value frequency (never, sometimes, often, or always).
of a dependent variable depends on the value of other variables, Interval-level and ratio-level variables may also be inter-
whereas the value of an independent variable does not rely on preted using a pie graph or bar graph; however, these types of
other variables. In addition, an investigator can influence the variables often have too many categories for such graphs to
value of an independent variable, such as treatment-group provide meaningful information. Instead, these variables may be
assignment. Independent variables are also referred to as better interpreted using a histogram. Unlike a bar graph, which
predictors because we can use information from these variables to displays the frequency for each distinct category, a histogram
predict the value of a dependent variable. Building on the group displays the frequency within a range of continuous categories.
of variables listed in the first paragraph of this section, blood glu- Information from this type of figure allows us to determine
cose could be considered a dependent variable, because its value whether the data are normally distributed. In addition to pie
may depend on values of the independent variables age, sex, eth- graphs, bar graphs, and histograms, many other types of figures
nicity, exercise frequency, weight, and treatment group. are available for the visual representation of data. Interested read-
Statistics are mathematical formulae that are used to organize ers can find additional types of figures in the books recom-
and interpret the information that is collected through mended in the “Further Readings” section.
variables. There are 2 general categories of statistics, descriptive Figures are also useful for visualizing comparisons between
and inferential. Descriptive statistics are used to describe the variables or between subgroups within a variable (for example,
collected information, such as the range of values, their average, the distribution of blood glucose according to sex). Box plots are
and the most common category. Knowledge gained from useful for summarizing information for a variable that does not
descriptive statistics helps investigators learn more about the follow a normal distribution. The lower and upper limits of
study sample. Inferential statistics are used to make comparisons the box identify the interquartile range (or 25th and 75th
and draw conclusions from the study data. Knowledge gained percentiles), while the midline indicates the median value (or
from inferential statistics allows investigators to make inferences 50th percentile). Scatter plots provide information on how the
categories for one continuous variable relate to categories in a sec-
and generalize beyond their study sample to other groups.
ond variable; they are often helpful in the analysis of correlations.
Before we move on to specific descriptive and inferential
In addition to using figures to present a visual description
statistics, there are 2 more definitions to review. Parametric
of the data, investigators can use statistics to provide a numeric
statistics are generally used when values in an interval-level or
description. Regardless of the measurement level, we can find the
ratio-level variable are normally distributed (i.e., the entire group
mode by identifying the most frequent category within a variable.
of values has a bell-shaped curve when plotted by frequency).
When summarizing nominal-level and ordinal-level variables, the
These statistics are used because we can define parameters of the
simplest method is to report the proportion of participants
data, such as the centre and width of the normally distributed
within each category.
curve. In contrast, interval-level and ratio-level variables with
The choice of the most appropriate descriptive statistic for
values that are not normally distributed, as well as nominal-level
interval-level and ratio-level variables will depend on how the
and ordinal-level variables, are generally analyzed using nonpara-
values are distributed. If the values are normally distributed, we
metric statistics.
can summarize the information using the parametric statistics of
mean and standard deviation. The mean is the arithmetic average
METHODS FOR SUMMARIZING STUDY
of all values within the variable, and the standard deviation tells
DATA: DESCRIPTIVE STATISTICS
us how widely the values are dispersed around the mean. When
The first step in a data analysis plan is to describe the data values of interval-level and ratio-level variables are not normally
collected in the study. This can be done using figures to give a distributed, or we are summarizing information from an
visual presentation of the data and statistics to generate numeric ordinal-level variable, it may be more appropriate to use the
descriptions of the data. nonparametric statistics of median and range. The first step in
Selection of an appropriate figure to represent a particular identifying these descriptive statistics is to arrange study partici-
set of data depends on the measurement level of the variable. pants according to the variable categories from lowest value to
Data for nominal-level and ordinal-level variables may be inter- highest value. The range is used to report the lowest and highest
preted using a pie graph or bar graph. Both options allow us to values. The median or 50th percentile is located by dividing the

312 C J H P – Vol. 68, No. 4 – July–August 2015 J C P H – Vol. 68, no 4 – juillet–août 2015
 This single copy is for your personal, non-commercial use only.
For permission to reprint multiple copies or to order presentation-ready copies for distribution, contact CJHP at [email protected]

number of participants into 2 groups, such that half (50%) of the list of inferential statistics that would be relevant to a partic-
the participants have values above the median and the other half ular study. Appendix 1 provides brief definitions of the inferential
(50%) have values below the median. Similarly, the 25th statistics named in these figures. Additional information, such as
percentile is the value with 25% of the participants having values the formulae for various inferential statistics, can be obtained
below and 75% of the participants having values above, and the from textbooks, statistical software packages, and biostatisticians.
75th percentile is the value with 75% of participants having
values below and 25% of participants having values above. What Is the Research Question?
Together, the 25th and 75th percentiles define the interquartile range. The first step in identifying relevant inferential statistics for
a study is to consider the type of research question being asked.
PROCESS TO IDENTIFY RELEVANT You can find more details about the different types of research
STATISTICAL TESTS: INFERENTIAL STATISTICS questions in a previous article in this Research Primer series that
One caveat about the information provided in this section: covered questions and hypotheses.5 A relational question seeks
selecting the most appropriate inferential statistic for a specific information about the relationship among variables; in this
study should be a combination of following these suggestions, situation, investigators will be interested in determining whether
seeking advice from experts, and discussing with your co-inves- there is an association (Figure 1). A causal question seeks infor-
tigators. My intention here is to give you a place to start a mation about the effect of an intervention on an outcome; in
conversation with your colleagues about the options available as this situation, the investigator will be interested in determining
you develop your data analysis plan. whether there is a difference (Figure 2).
There are 3 key questions to consider when selecting an
What Is the Study Design?
appropriate inferential statistic for a study: What is the research
question? What is the study design? and What is the level of When considering a question of association, investigators
measurement? It is important for investigators to carefully will be interested in measuring the relationship between variables
consider these questions when developing the study protocol and (Figure 1). A study designed to determine whether there is
creating the analysis plan. The figures that accompany these consensus among different raters will be measuring agreement.
questions show decision trees that will help you to narrow down For example, an investigator may be interested in determining

Figure 1. Decision tree to identify inferential statistics for an association.

C J H P – Vol. 68, No. 4 – July–August 2015 J C P H – Vol. 68, no 4 – juillet–août 2015 313
 This single copy is for your personal, non-commercial use only.
For permission to reprint multiple copies or to order presentation-ready copies for distribution, contact CJHP at [email protected]

Figure 2. Decision tree to identify inferential statistics for measuring a difference.

whether 2 raters, using the same assessment tool, arrive at the independent or unpaired when the information is taken from dif-
same score. Correlation analyses examine the strength of a ferent groups. For example, we could use an unpaired t test to
relationship or connection between 2 variables, like age and compare the mean age between 2 independent samples, such
blood glucose. Regression analyses also examine the strength of a as the intervention and control groups in a study. Samples are
relationship or connection; however, in this type of analysis, one considered related or paired if the information is taken from the
variable is considered an outcome (or dependent variable) and same group of people, for example, measurement of blood
the other variable is considered a predictor (or independent vari- glucose at the beginning and end of a study. Because blood
able). Regression analyses often consider the influence of glucose is measured in the same people at both time points, we
multiple predictors on an outcome at the same time. For could use a paired t test to determine whether there has been a
example, an investigator may be interested in examining the significant change in blood glucose.
association between a treatment and blood glucose, while also
considering other factors, like age, sex, ethnicity, exercise What Is the Level of Measurement?
frequency, and weight.
When considering a question of difference, investigators As described in the first section of this article, variables can
must first determine how many groups they will be comparing. be grouped according to the level of measurement (nominal,
In some cases, investigators may be interested in comparing the ordinal, or interval). In most cases, the independent variable in
characteristic of one group with that of an external reference an inferential statistic will be nominal; therefore, investigators
group. For example, is the mean age of study participants similar need to know the level of measurement for the dependent
to the mean age of all people in the target group? If more than variable before they can select the relevant inferential statistic.
one group is involved, then investigators must also determine Two exceptions to this consideration are correlation analyses and
whether there is an underlying connection between the sets of regression analyses (Figure 1). Because a correlation analysis
values (or samples) to be compared. Samples are considered measures the strength of association between 2 variables, we need

314 C J H P – Vol. 68, No. 4 – July–August 2015 J C P H – Vol. 68, no 4 – juillet–août 2015
 This single copy is for your personal, non-commercial use only.
For permission to reprint multiple copies or to order presentation-ready copies for distribution, contact CJHP at [email protected]

to consider the level of measurement for both variables. Regres- analysis plans and review the plans of others. Biostatisticians have
sion analyses can consider multiple independent variables, often a wealth of knowledge in the field of statistical analysis and can
with a variety of measurement levels. However, for these analyses, provide advice on the correct selection, application, and inter-
investigators still need to consider the level of measurement for pretation of these methods. Colleagues who have “been there and
the dependent variable. done that” with their own data analysis plans are also valuable
Selection of inferential statistics to test interval-level variables sources of information. Identify these individuals and consult
must include consideration of how the data are distributed. An with them early and often as you develop your analysis plan.
underlying assumption for parametric tests is that the data Another important resource to consider when creating your
approximate a normal distribution. When the data are not analysis plan is textbooks. Numerous statistical textbooks are
normally distributed, information derived from a parametric test available, differing in levels of complexity and scope. The titles
may be wrong.6 When the assumption of normality is violated listed in the “Further Reading” section are just a few suggestions.
(for example, when the data are skewed), then investigators I encourage interested readers to look through these and other
should use a nonparametric test. If the data are normally distrib- books to find resources that best fit their needs. However, one
uted, then investigators can use a parametric test. crucial book that I highly recommend to anyone wanting to be
an investigator or peer reviewer is Lang and Secic’s How to
ADDITIONAL CONSIDERATIONS Report Statistics in Medicine (see “Further Reading”). As the title
What Is the Level of Significance? implies, this book covers a wide range of statistics used in medical
research and provides numerous examples of how to correctly
An inferential statistic is used to calculate a p value, the prob-
report the results.
ability of obtaining the observed data by chance. Investigators
can then compare this p value against a prespecified level of sig-
CONCLUSIONS
nificance, which is often chosen to be 0.05. This level of
significance represents a 1 in 20 chance that the observation is When it comes to creating an analysis plan for your project,
wrong, which is considered an acceptable level of error. I recommend following the sage advice of Douglas Adams in The
Hitchhiker’s Guide to the Galaxy: Don’t panic!14 Begin with
What Are the Most Commonly Used Statistics? simple methods to summarize and visualize your data, then use
In 1983, Emerson and Colditz7 reported the first review of the key questions and decision trees provided in this article to
statistics used in original research articles published in the New identify relevant statistical tests. Information in this article will
England Journal of Medicine. This review of statistics used in the give you and your co-investigators a place to start discussing the
journal was updated in 1989 and 2005,8 and this type of analysis elements necessary for developing an analysis plan. But do not
has been replicated in many other journals.9-13 Collectively, these stop there! Use advice from biostatisticians and more experienced
reviews have identified 2 important observations. First, the overall colleagues, as well as information in textbooks, to help create
sophistication of statistical methodology used and reported in your analysis plan and choose the most appropriate statistics
studies has grown over time, with survival analyses and multi- for your study. Making careful, informed decisions about the
variable regression analyses becoming much more common. The statistics to use in your study should reduce the risk of confirming
second observation is that, despite this trend, 1 in 4 articles Mr Twain’s concern.
describe no statistical methods or report only simple descriptive
References
statistics. When inferential statistics are used, the most common 1. Twain M (Kiskis MJ, editor). Mark Twain’s own autobiography: the chapters
are t tests, contingency table tests (for example, ␹2 test and Fisher from the North American review. 2nd ed. Madison (WI): University of
exact test), and simple correlation and regression analyses. This Wisconsin Press; 2010. 318 p.
2. Austin Z, Sutton J. Qualitative research: getting started. Can J Hosp Pharm.
information is important for educators, investigators, reviewers, 2014;67(6):436-40.
and readers because it suggests that a good foundational knowl- 3. Sutton J, Austin Z. Qualitative research: data collection, analysis, and
edge of descriptive statistics and common inferential statistics management. Can J Hosp Pharm. 2015;68(3):226-31.
will enable us to correctly evaluate the majority of research 4. Dawson NV, Weiss R. Dichotomizing continuous variables in statistical
analysis: a practice to avoid. Med Decis Making. 2012;32(2):225-6.
articles.11-13 However, to fully take advantage of all research 5. Tully MP. Research: articulating questions, generating hypotheses, and
published in high-impact journals, we need to become choosing study designs. Can J Hosp Pharm. 2014;67(1):31-4.
acquainted with some of the more complex methods, such as 6. Harwell MR. Choosing between parametric and nonparametric tests. J
multivariable regression analyses.8,13 Couns Dev. 1988;67(1):35-8.
7. Emerson JD, Colditz GA. Use of statistical analysis in the New England
Journal of Medicine. N Engl J Med. 1983;309(12):709-13.
What Are Some Additional Resources? 8. Horton NJ, Switzer SS. Statistical methods in the journal. N Engl J Med.
2005;353(18):1977-9.
As an investigator and Associate Editor with CJHP, I have 9. Guyatt G, Jaeschke R, Heddle N, Cook D, Shannon H, Walter S. Basic
often relied on the advice of colleagues to help create my own statistics for clinicians: 1. Hypothesis testing. CMAJ. 1995;152(1):27-32.

C J H P – Vol. 68, No. 4 – July–August 2015 J C P H – Vol. 68, no 4 – juillet–août 2015 315
 This single copy is for your personal, non-commercial use only.
For permission to reprint multiple copies or to order presentation-ready copies for distribution, contact CJHP at [email protected]

10. Goldin J, Zhu W, Sayre JW. A review of the statistical analysis used in papers Norman GR, Streiner DL. PDQ statistics. 3rd ed. Hamilton (ON): B.C. Decker;
published in Clinical Radiology and British Journal of Radiology. Clin Radiol. 2003.
1996;51(1):47-50. Plichta SB, Kelvin E. Munro’s statistical methods for health care research. 6th ed.
11. Reed JF 3rd, Salen P, Bagher P. Methodological and statistical techniques: Philadelphia (PA): Wolters Kluwer Health/Lippincott, Williams & Wilkins; 2013.
what do residents really need to know about statistics? J Med Syst.
2003;27(3):233-8.
12. Hellems MA, Gurka MJ, Hayden GF. Statistical literacy for readers of
Pediatrics: a moving target. Pediatrics. 2007;119(6):1083-8.
13. Taback N, Krzyzanowska MK. A survey of abstracts of high-impact clinical
journals indicated most statistical methods presented are summary statistics. Scot H Simpson, BSP, PharmD, MSc, is Professor and Associate Dean,
J Clin Epidemiol. 2008;61(3):277-81. Research and Graduate Studies, Faculty of Pharmacy and Pharmaceutical
14. Adams D. The hitchhiker's guide to the galaxy. London (UK): Pan Books; Sciences, University of Alberta, Edmonton, Alberta. He is also an Associate
1979. Editor with the CJHP.

Further Reading Competing interests: None declared.

Devor J, Peck R. Statistics: the exploration and analysis of data. 7th ed. Boston Address correspondence to:
(MA): Brooks/Cole Cengage Learning; 2012. Scot H Simpson
Lang TA, Secic M. How to report statistics in medicine: annotated guidelines for Faculty of Pharmacy and Pharmaceutical Sciences
authors, editors, and reviewers. 2nd ed. Philadelphia (PA): American College of 3126 Dentistry/Pharmacy
Physicians; 2006. University of Alberta
Mendenhall W, Beaver RJ, Beaver BM. Introduction to probability and statistics. Edmonton AB T6G 2N8
13th ed. Belmont (CA): Brooks/Cole Cengage Learning; 2009. e-mail: [email protected]

Appendix 1. Glossary of statistical terms* (part 1 of 2)

ANOVA (analysis of variance): Parametric statistic used to Descriptive statistics: Numeric or graphic summaries (or
compare the means of 3 or more groups that are defined by descriptions) of a variable.
1 or more variables. Inferential statistics: Measures the difference between 2
• 1-way ANOVA: Uses 1 variable to define the groups for variables or subgroups of a variable. Allows the investigator
comparing means. This is similar to the Student t test when to make inferences about another group on the basis of
comparing the means of 2 groups. information generated from the study data.
• Kruskall–Wallis 1-way ANOVA: Nonparametric alternative
for the 1-way ANOVA. Used to determine the difference in Kappa (␬): Measures the degree of nonrandom agreement
medians between 3 or more groups. between observers or measurements for the same nominal-
• n-way ANOVA: Uses 2 or more variables to define groups level variable.
when comparing means. Also called a “between-subjects Kendall tau (␶): Nonparametric alternative for the Spearman
factorial ANOVA”. correlation. Used when measuring the relationship between
• Repeated-measures ANOVA: A method for analyzing 2 ranked (or ordinal-level data) variables.
whether the means of 3 or more measures from the same
Mann–Whitney U test: Nonparametric alternative for the
group of participants are different.
independent t test. One variable is dichotomous (e.g., group
• Freidman ANOVA: Nonparametric alternative for the
A versus group B) and the other variable is either ordinal or
repeated-measures ANOVA. It is often used to compare
interval.
rankings and preferences that are measured 3 or more times.
Pearson correlation: Parametric test used to determine
Binomial test: Used to determine whether the observed pro-
whether an association exists between 2 variables measured
portion is significantly different from a known or hypothesized
at the interval or ratio level.
proportion. The variable is dichotomous (nominal-level data
with 2 options). Phi (␾): Used when both variables in a correlation analysis are
dichotomous.
Biserial correlation (rank or point): Correlation technique
when one of the variables is dichotomous (or measured at Runs test: Used to determine whether a series of data occurs
the nominal level). from a random process.
Chi-square (␹2) test: Nonparametric test used to determine Spearman rank correlation: Nonparametric alternative for
whether a statistically significant association exists between the Pearson correlation coefficient. Used when the assump-
rows and columns in a contingency table. tions for Pearson correlation are violated (e.g., data are not
• Fisher exact: Variation of chi-square that accounts for cell normally distributed) or one of the variables is measured at the
counts < 5. ordinal level.
• McNemar: Variation of chi-square that tests statistical t test: Parametric statistical test for comparing the means of
significance of changes in 2 paired measurements of 2 independent groups.
dichotomous variables. • 1-sample: Used to determine whether the mean of a sample
• Cochran Q: An extension of the McNemar test that provides is significantly different from a known or hypothesized
a method for testing for differences between 3 or more value.
matched sets of frequencies or proportions. Often used as
a measure of heterogeneity in meta-analyses. continued on page 317

316 C J H P – Vol. 68, No. 4 – July–August 2015 J C P H – Vol. 68, no 4 – juillet–août 2015
 This single copy is for your personal, non-commercial use only.
For permission to reprint multiple copies or to order presentation-ready copies for distribution, contact CJHP at [email protected]

Appendix 1. Glossary of statistical terms* (part 2 of 2)

• Independent-samples t test (also referred to as the *Sources
Student t test): Used when the independent variable is a Lang TA, Secic M. How to report statistics in medicine:
nominal-level variable that identifies 2 groups and the annotated guidelines for authors, editors, and reviewers. 2nd
dependent variable is an interval-level variable. ed. Philadelphia (PA): American College of Physicians; 2006.
• Paired: Used to compare 2 pairs of scores between 2 groups
(e.g., baseline and follow-up blood pressure in the Norman GR, Streiner DL. PDQ statistics. 3rd ed. Hamilton (ON):
intervention and control groups). B.C. Decker; 2003.
Wilcoxon rank–sum test: Nonparametric alternative to the Plichta SB, Kelvin E. Munro’s statistical methods for health care
independent t test based solely on the order in which observa- research. 6th ed. Philadelphia (PA): Wolters Kluwer Health/
tions from the 2 samples fall. Similar to the Mann–Whitney Lippincott, Williams & Wilkins; 2013.
U test.
Wilcoxon signed-rank test: Nonparametric alternative to the
paired t test. The differences between matched pairs are com-
puted and ranked. This test compares the sum of the negative
differences and the sum of the positive differences.

This article is the 12th in the CJHP Research Primer Series, an ini- Tsuyuki RT. Designing pharmacy practice research trials. Can J
tiative of the CJHP Editorial Board and the CSHP Research Com- Hosp Pharm. 2014;67(3):226-9.
mittee. The planned 2-year series is intended to appeal to Bresee LC. An introduction to developing surveys for pharmacy
relatively inexperienced researchers, with the goal of building re- practice research. Can J Hosp Pharm. 2014;67(4):286-91.
search capacity among practising pharmacists. The articles, pre-
senting simple but rigorous guidance to encourage and support Gamble JM. An introduction to the fundamentals of cohort and
novice researchers, are being solicited from authors with appro- case–control studies. Can J Hosp Pharm. 2014;67(5):366-72.
priate expertise. Austin Z, Sutton J. Qualitative research: getting started. Can J
Hosp Pharm. 2014;67(6):436-40.
Houle S. An introduction to the fundamentals of randomized con-
Previous articles in this series: trolled trials in pharmacy research. Can J Hosp Pharm. 2014;
Bond CM. The research jigsaw: how to get started. Can J Hosp 68(1):28-32.
Pharm. 2014;67(1):28-30. Charrois TL. Systematic reviews: What do you need to know to
Tully MP. Research: articulating questions, generating get started? Can J Hosp Pharm. 2014;68(2):144-8.
hypotheses, and choosing study designs. Can J Hosp Pharm. Sutton J, Austin Z. Qualitative research: data collection, analysis,
2014;67(1):31-4. and management. Can J Hosp Pharm. 2014;68(3):226-31.
Loewen P. Ethical issues in pharmacy practice research: an Cadarette SM, Wong L. An introduction to health care adminis-
introductory guide. Can J Hosp Pharm. 2014;67(2):133-7. trative data. Can J Hosp Pharm. 2014;68(3):232-7.

C J H P – Vol. 68, No. 4 – July–August 2015 J C P H – Vol. 68, no 4 – juillet–août 2015 317