1/121
Looks like no tags are added yet.
Where do ideas for research questions come from?
discussions
practicum supervisors
current events
clinic observations
publications
ongoing projects
collaboration
what is the purposes of organizing ideas for questions?
helps to think about the IMPACT the research could have
provides options
Criteria used to evaluate quantitative data
internal validity
external validity
reliability
objectivity
answers: “What” and “when”
criteria used to evaluate qualitative data
credibility (trustworthiness)
transferability
dependability
confirmability
answers “why” and “how” and “how come”
common types of quantitative research
descriptive
correlational
causal-comparative/ quasi-experimental
experimental
common types of qualitative research
historical
meta-analysis
narrative- focuses in on a specific event
grounded theory- where you build a new theory from real-world data instead of starting with a guess or a pre-existing idea
phenomenology- general experiences over-time
case study
bolded=used the most
qualitative research steps
identify- general research question
choose- main methods, sites, subjects
determine- methods of data documentation
decide- on various aspects for collecting data
clarify- researchers’ roles
evaluate- study’s ethical implications
collect- data saturation
interpret- data by identifying concepts and theories
revise
collect and verify
complete
present
steps of quantitative research
make- observations about an unclear or new phenomenon
investigate- current theory surrounding phenomenon
hypothesize- an explanation for observations
predict- outcomes based on the hypothesis
formulate- plan to test prediction
collect and process- data
verify
make
present
Thesis statement
succinct summary about the central points of your research
blue-print for readers
hypothesis
testable statement, prediction
not all research needs an hypothesis
research questions
exploratory questions
all research has research questions
how to form a good research question
fills a gap and is relevant
addresses gaps in existing knowledge
builds upon existing knowledge
relevant to GC
reasonable scope
addresses a piece of an important issue without taking on too much
feasibility
clear and specific
concise- clear and brief summary of what the project is about
precise- states the who and the what
focusing on the basic questions before creating the research question
what: what is the problem that the research will address?
who: 3 current, peer-reviewed references that support the presence of the problem
how, where, and when: describe the impact of the problem
who: 3 current, peer-reviewed references that support the impact of the problem that the research proposes
why: identify the conceptual basis for the problem. What does the literature outline as the cause of the problem?
who: 3 current, peer-reviewed references that support the conceptual basis of the problem and describe the nature of that support
practical considerations for a thesis
time
15 months, with 3-6 months tops to focus on data collection
cost
$1,000
reasonable costs
available data sources
bloom’s taxonomy
clear, measurable verbs that accurately reflect the purpose and scope of your research aim
remember is the bottom, create is the top

simple formula to coming up with a research questions
action verb+ what you are studying+ in whom/where+ relevant outcome or relationship
mixed methods
integrates qualitative and quantitative evidence
mult-method
uses multiple methods without necessarily integrating findings into one analysis
multimodal
Uses different forms of data or expression, including visual and interactional approaches
positionality
how our identities, experiences, values, and professional backgrounds shape our perspectives
reflexivity
ongoing practice of critically examining these influences and responding to them throughout research
mixed methods designs→ explanatory sequential
Quant+qual= explain results
mixed methods designs→ exploratory sequential
qual+quant= extend findings
mixed methods designs→ convergent
quant+qual= bring findings together
survey is the ___ of your study
method
questionnaire is the data collection _____
instrument
cross-sectional could be the _______
study design
_____ could be quant, qual, multi-method, or part of mixed-method study
methodology
pros of surveying
ease of distribution
low cost
familiar format
feasible to describe characteristics of larger populations
ability to focus on very specific topic in a targeted manner
opportunity to reach a larger population than other methods (if online)
cons of surveying
low response rate
recall bias
self-report or hypothetical scenarios or future actions can lead to inaccurate conclusions
data may depend on cognitive load to complete
pros & cons of using open ended questions in surveys
pros: enables broader responses that might be impossible to obtain with close-ended questions
cons: coding answers for analysis
pros & cons of using close ended questions in surveys
pros: uniformity of responses; more easily processed/analyzed
cons: may miss capturing important/critical opinions, or broader range of responses
pros & cons of using multi-method questions in surveys
contains both open and closed questions
pros: able to capture more robust opinions/answers, in addition to standardizing portions of the data
cons: additional analysis; need to word open-ended questions appropriately or you may not get additional data
how to collect and report demographic data?
use the demographic survey question only if you plan to do something meaningful with the data
tell users the why and how
stick with checkboxes for the answer options, and enable multi-select
make your survey more inclusive with ‘other’
make demographic questions optional
types of scales: interval
order, constant unit
ex: incomes: $20,000-29,999, $30,000-$39,999
types of scales: ordinal
answers with a natural order to categories; differences between categories vary and cannot be quantified
Ex: not likely, somewhat likely, very likely
types of scales: nominal
answers with categories that are unranked, unordered
ex: Blood type: A, B, AB, O
likert scales
ordinal
label all anchors
offer balanced options
beware of extremes (always, never)
be careful w/ neutral options (neither agree nor disagree)
be careful with “don’t know” and “not applicable”
rules when doing interview surveys
interviewers must be neutral in appearance and actions
personal opinions and interviewer presence (head nods, etc) should have no effect on respondents’ choice of answers
interviewers must be adequately trained
familiar with the questionnaire
follow question order and wording exactly
record responses exactly how they were given
interviewers can use probes
elaboration on an incomplete or ambiguous response
probes→ neutral
all interviewers should use same probes
avoid interview surveys that elicit sensitive info
tips when using surveys
make questions clear
provide reference frames
make sure all respondents are thinking about the same time and place
avoid double-barrel questions (ask one question at a time)
short questions are best
avoid negative & double-negative questions
use vocab for target population
avoid questions using leading, emotional, or evocative language
scale anchor
defining what a scale variable means
ex: 5= very likely, 1= never
survey design summary
define what you want to measure
write clear, specific, unbiased questions and response options
identify the target population and appropriate sampling strategy
pretest the questionnaire, including with intended respondents
consider validity, reliability, accessibility, and response bias
common types of quantitative research
descriptive
correlational
causal-comparative/Quasi-experimental
experimental
why do we sample?
not feasible to study entire population
use statistics to make inferences about population parameters
access can determine population/sample as well as scope of your project
sampling types: probability-based
random
study conclusions apply to entire population
better for quantitative studies
sampling types: non-probability
non-random
cannot be applied beyond the sample
encouraged for qualitative studies
sampling: population
all members of some group defined by specific criteria
report criteria to audience
sampling: parameters
quantitative characteristics of a population
ex: mean age of all patients in clinic
sampling: sample
subgroup of target population (study participants)
sampling: statistics
quantitative characteristics of a sample
probability-based: simple random sampling
create a list of possible participants, select using random process
each member and each sample must have an equal chance of being selected
produces unbiased sample
probability-based: stratified random sampling
statistical method where a researcher divides a larger population into smaller, non-overlapping groups—called strata—based on shared characteristics (such as age, income, department, or education level), and then randomly selects participants from each subgroup
take a few individuals from ALL of the groups.
probability-based: cluster sampling
a probability sampling method where you divide a large population into smaller, mutually exclusive groups called clusters (such as schools, neighborhoods, or workplaces), and then randomly select entire clusters to collect data from
take ALL of the individuals from a FEW of the groups.
nonprobability sampling: convenience samples
already have access to sample or quick access
nonprobability sampling: purposeful samples
selected based on ability to provide valid data
how to collect & report demographic data
use demographic survey questions only if you plan to do something meaningful with the data
tell users “why'“ and “how”
use checkboxes for the answer option & enable multi-select
make questions optional
limitations of demographic data on surveys
self-identification- may not capture complexities of individual identities
subjectivity- can be influenced by personal interpretation and societal pressures
limited options- may not cover full range of backgrounds and identities
ambiguity and overgeneralization- oversimplify diversity
ethical considerations- privacy, stereotyping, discrimination
intersectionality
why include comparison groups using publicly available data?
determines whether sample differs from relevant population
helps to identify potential selection bias of differences in sample composition
strengthens interpretation of demographic or other descriptive findings
may eliminate the time, cost, and participant burden of recruiting an internal comparison group
considerations when using publicly available data as a comparison group
population comparability
variable alignment
time & geography
sampling methods
available variables
statistical weight
concerns with pre-existing data sets
ethical considerations
historical inaccuracies and categorization limits
Small N, identifiability risk
lack of consent for secondary use
equity considerations
deficit-based reporting and stigmatization
power dynamics and data ownership
the “othering” of minoritized groups
cleaning & transforming data is not…
changing or making up data
options for missing data
omit entire case
can be seen as ignoring data
biased estimates, inaccurate conclusions
omitting single variable
replace single variable with highly correlated variable for analysis
substitution methods (mean, median, mode)
produces biased means, reduce variability in responses
what do to when there is missing data
check for problematic participants
check for problematic patterns
descriptive statistics
describes chunk of raw data
helps with organizing, understanding data in a meaningful way
ex: summary stats, graphs, tables
inferential statistics
small sample of data to draw inferences about the larger population, outcome, or probability
representative sample/random
large enough sample size
ex: hypothesis tests. CI, regression
summary statistics
explains data which is already known, to summarize sample
3 key components of qualitative research
context
researcher is instrument
depth over breadth
limitations of qualitative research
time, cost
access
lack of generalizability
lack of reproducibility→ every group is different based on context
small sample sizes
3 methods of data collection for qualitative research
participant observation
interviewing/focus group
document analysis→ field notes→ constant comparison, TRIANGULATE DATA
qualitative data collection: field notes
on-site observations, notes from phone calls
interpretivism
no “true reality”, facts and meanings are socially constructed
uses of qualitative research in GC
tool in understanding health behavior, program utilization, program and product development
improving engagement
health services
programming
evaluation
health communication
better understanding of rare or stigmatized conditions
patient experience- VUS
qualitative research is…
participant led
triangulation
comparing interviews, observation, and documents to build a more complete and credible understanding
form of quality control→ trustworthiness
Reflexivity in qualitative research (validity testing)
ensures trustworthiness
aware of biases & how much they influence data
easier with teams
biases can change over time
Statistical analysis considerations
sample size
descriptive statistics used to summarize characteristics of the sample
methods used to compare certain groups of patients
graphs
multivariate models (establish association)
statistical significance level
software
point estimates are used to quantify
averages
variability
proportions
risk
confidence intervals…
must be presented with every estimate
95% CI means that we are 95% confident that the calculated confidence interval will cover the true target value for which it was constructed
question to ask when comparing two or more groups
do the observed proportion estimates reflect a difference in the population?
null hypothesis
H0: no difference in population values
alternative hypothesis
HA, population values are different
type I error
false positive
type II error
false negative
p-value
probability of obtaining that observed difference by chance alone
less than 0.05= reject null and conclude that there is a significant difference
represents <5% chance of obtaining those results by chance assuming no treatment difference
things to keep in mind regarding p-values
small p-value only indicates that treatment effect is nonzero
p-value is sensitive to sample size
statistical significance does not equal clinically significant
non-significant p-value DOES NOT mean the treatment effect is 0
two-sided tests
HA states that the population values simply differ
one sided tests
HA specifies a direction of effect
tests for comparing the means of continuous measurements between groups
independent sample t-test (2 independent groups)
paired t-test (2 dependent groups… measures before and after intervention)
ANOVA (more than 2 independent groups)
relies on distributions having bell-shaped curve
based on differences between means (SD)
nonparametric tests for comparing the medians of continuous measurements between groups
require milder assumptions, deal w/ rankings of measurements instead of measurements themselves
wilcoxon rank sum test (2 independent groups)
signed rank test (2 dependent groups)
kruskal-wallis ( more than 2 independent groups)
methods for comparing categorical variables
chi-square test
Fisher’s exact test for association
quantifying risk differences: relative risk
RR= chances of event in control group/chances of event in experimental group
quantifying risk differences: odds ratio
OR= odds (in control group)/ odds (in experimental group)
where
odds= probability (event)/ 1-probability (event)
RR=1
probabilities of the event in control and experimental are identical
RR>1
increased risk of the even among those in control group compared to patients in experimental group
RR<1
decreased risk of event among control patients compared to experimental group
true or false: interpretation for RR is similar to that of OR
true
Relative risk reduction notation: CER
chances of even in control group
Relative risk reduction notation: EER
chances of event in experimental group
relative risk reduction equation
RRR= (CER-EER)/CER
absolute risk reduction equation
ARR= CER-EER