questionss
Health Data Analytics
Written Assignment
Read the scenario below and answer the questions related to research studies after the scenario.
Scenario: Researchers decided to investigate factors that contribute to patients not being given a
definitive diagnosis for several months after seeing multiple physicians. The researchers investigated by
interviewing patients who have dealt with a delayed diagnosis. Sixty-five patients were interviewed. The
interviews were audio-recorded, transcribed, and a group of researchers analyzed the transcripts to
identify recu
ing themes regarding factors contributing to delayed diagnosis. Some of the themes
identified were lack of thorough investigation, physicians not listening to patient views, not reviewing
test results, etc. The researchers wanted to fully understand factors impacting delayed diagnosis, so they
also searched the literature for articles published by other researchers on factors contributing to delayed
diagnosis to see what they found, and if other researchers found similar factors. Once they identified the
themes, they computed frequencies for each theme by counting how many patients mentioned the
same thing (theme), and determined which themes were more frequent.
1. Which type of research study is this? Quantitative, Qualitative, etc.? Why?
2. If you feel the study is a Qualitative research study, what type of qualitative research is applicable?
Why?
3. Is the research considered Descriptive Research, Analytical Study, or an Experimental Research
Study? Why?
4. Each of these three types of studies listed in question 3 have more specific research categories
(sub-types). That is, Descriptive research studies can be Observational or Cross-Sectional. For the
type you selected as your answer for question 3, what sub-type is applicable for the study described
in the scenario? Why?
Title
Health Data Analytics
Study Design
and
Sample Selection
Research Studies
• Study – Data driven investigation to understand …
Understand a disease, business process, behavior, etc.
• Quantitative Research – Quantifies a problem by generating
and analyzing numerical data
Generalize results from a sample to a population
Uses measureable data to formulate facts and uncover
trends, patterns, etc.
• Qualitative Research – Exploratory research used to gain an
understanding of underlying reasons, opinions, motivations, etc.
Provides insights, opinions, etc. into a problem
Helps develop ideas & hypotheses, etc. for potential
development of quantitative research
• Mixed Methods Research – Study included both qualitative
and quantitative components
https:
www.snapsurveys.com
log/qualitative-vs-quantitative-research
https:
www.snapsurveys.com
log/qualitative-vs-quantitative-research
Qualitative Research Studies
• Five Types of Qualitative Research
Ethnography – researcher immersed in participants’
environment
Na
ative – weave together views, events, etc. from participants
to form a cohesive story
Phenomenological – combination of methods (interviews,
literature review, observation, etc.) to understanding
phenomena under study (small sample size)
Grounded Theory – build a theory grounded in data – interview
transcripts, etc. (larger sample size)
Case Study – deep understanding by analyzing multiple types of
data sources
https:
measuringu.com/qual-methods
https:
measuringu.com/qual-methods
Other Names for Research Studies
• Descriptive Research Studies
• Analytical Research Studies
• Experimental Research Studies
Other Names for Research Studies
• Descriptive Research Studies – Describe characteristics of a
population or a phenomenon being studied (qualitative)
Information collected to learn about a population, area, etc.
Investigate variables to determine running analytic study
No manipulation of variables - Addresses the question What?
Does not answer how, when, why, etc.
Cannot be used to determine relationships (causal, etc.)
• Analytical Studies – Investigates relationship between variables
Effect of independent variable on dependent variable
Limited control over conditions
• Experimental Research Studies – Investigates relationship
etween variables in very controlled environment
Effect of independent variable on dependent variable
“Scientific Research” - Observation under controlled conditions
Descriptive Research Studies & Methodology
• Observational Study – Draws inference from sample where
variables not under control of researche
Very little is known about the variables being studied
Data is collected in order to generate hypotheses for Analytic and
Experimental studies
• Cross-Sectional Study – Describes the characteristics of the
population under study at a specific point in time
Example – Disease prevalence
Unit of Analysis – patients, employees, billed claims, etc.
• Methodology for Descriptive Research Studies
Survey – Assessment of opinions based on survey results
Focus Groups – Groups of individuals discussing topic / idea
Interviews – Commonly one-on-one conversation - In-person, Phone
Etc.
Analytical Research Studies
• Longitudinal Study – Repeated observations of the same
variables over period of time
Example – Study life events throughout lifetime / generation
• Cohort Study – Prospective analytic study
Study participants are ‘cohorts’ (with or without attribute) –
assigned prior to start of study
Participants followed over time (similar to longitudinal)
Calculate Relative Risk of an attribute – Likelihood of event
occu
ing under condition vs. not under condition
Likelihood of ER visit for those exposed to flu vs. not exposed
Analytical Research Studies
• Case-Control Study – Two groups differing in outcome are
compared on the basis of a ‘supposed’ causal attribute
Case – participants with the attribute under study
Control – participants without the attribute under study
Retrospective study – Looking back in time
Example – Patients with disease (case) vs. no disease (control)
Cannot determine causal relationship --- Only determine association
Association depicted by Odds Ratio – odds of an event occu
ing if
another event occurs or under a condition
~ Odds of contracting disease when a risk factor is present
Experimental Research Studies
• Randomized Control Trial (RCT)
Test effectiveness of treatment protocols, drug discoveries, etc.
Determination of Cause & Effect relationship
Monitored by the Food and Drug Administration (FDA)
Must follow strict study protocols outlining how subjects are selected
and treated during study
Intervention & Control Groups (arm of research)
~ Intervention – receive treatment / drug
~ Control – does not receive treatment / drug (placebo)
• Blinded Study Types
Single Blind – Subject (patient) does not know group assignment
Double Blind – Subject and researcher not know assignment
Triple Blind – Subject, researcher, analyst not know assignment
Sampling Terminology
• Sampling – Determining research subject participation
• Sample – subset of population
• Population (also called Universe) – Set of all units that are
eligible to be sampled
• Sampling Plan – Details inclusion & exclusion criteria and
sampling methodology
• Statistically Valid Sample Criteria
Sample that is large enough to provide information with
sufficient precision to meet the goals of the analysis
Probability sample where each item in population has equal
chance of being selected
Reproducible – Recreate sample by reviewing the process
Sampling Methods
• Probability Sampling – Each member of a population has a
known probability of being selected
Used when the goal of the study is to generalize results from
a sample to a specific population
• Non-Probability Sampling – Members of a sample are
deliberately selected for a specific purpose
Used when the goal of the study is to gain an understanding
of a process or exploratory data analysis
Probability Sampling Methods
• Simple Random Sampling
• Stratified Random Sampling
• Systematic Sampling
• Cluster Sampling
Simple Random Sampling
• Every member of the population has an equal chance of
eing selected for the sample
• Statistical equivalent of drawing sampling units from a hat
• Methods
Random Drawing
Numbering population & selecting sample using
Random Number Generator designating Random Seed
Systematic Random Sampling
• Random sample generated by selecting every 5th or 10th member
of the sampling population
Stratified Random Sampling
• Population divided into similar groups --- Strata --- based on set of
criteria
• Each subject is assigned to one and only one stratum (no overlap)
• Simple random sampling technique used to select subject from
each strata
Number selected is based on the size of the strata relative to
the size of the population (larger population – more selected)
Cluster Random Sampling
• Population divided into clusters (groups) before sample is selected
Clusters must be mutually exclusive & exhaustive (no overlap)
• Sampling performed as single-stage or two-stage
• Single-Stage Sampling
Clusters selected at random
All subjects (units) in a cluster are included in the sample
Randomly select a day of the month – Review all records for that day
• Two-Stage Sampling
Clusters are selected at random
Subjects (units) within the cluster are selected at random
All subjects (units) within the cluster are not included
Randomly select 3 shifts per month – Randomly select records from shift
Non-Probability Sampling Methods
• Convenience Sampling
• Judgment Sampling
• Quota Sampling
Convenience Sampling Methods
• Sample is chosen by convenience
Sample that are easiest to obtain
Samples that are at the lowest cost
• Not representative of entire population
Judgment Sampling Methods
• Judgment Sampling is completed by researcher with expert
knowledge of the subject being studied
• Sampling choices are based on researcher’s knowledge base
• Example – Data analysts selects specific medical claims that
equire further investigation
Quota Sampling Methods
• Segment population into mutually exclusive groups
• Judgment sampling used to select a number of
individuals for sample
Judgment is used in selection of individual units
• Example – Determine satisfaction of hospital food
Researcher approaches 10 males and 10 females to ask
interview them about their satisfaction
• Bias could impact sample selection
Researcher may be biased against particular groups