Exam 1: Research Questions Flashcards
Translational research
“…the transformation of knowledge through successive fields of research from a basic science discovery to public health impact—a complex process that requires both research (e.g., bench-work and clinical trials) and nonresearch activities (e.g., implementation)”
How could you generate new research ideas?
• Identify gaps or weaknesses in the literature
• Extend study new a new population, set of
materials, setting
• Apply different outcome measures
• Assess social validity of the research
• Use state-of-the art testing and analytical approaches
Topic/Broad Problem
• Introduces the reader to the importance and context of the research study
• Examples • How do children learn to identify speech sounds in their native language? • How do adults identify speech sounds? • Do noise-reduction algorithms in hearing aids improve speech understanding?
Theory
• Motivates/provides a framework for the current experiment (though you may ultimately disprove it)
- Organizes, explains, and account for data
- Describe relations among concepts to explain a phenomenon
• Example: Auditory Theory of Speech Perception (e.g., Fant, 1960; Stevens & Blumstein, 1978)
Listeners identify acoustic patterns or features by matching them to stored acoustic representations
Theory and Data
good research should go both ways - cycle to drive scientific method and constant improvements
Data driven research- reasoning from the data to the general theory
Theory driven - reasoning from a general theory to the data
Evaluating Theories
list the 4 ways
Some criteria for good theories:
- Comprehensiveness
- Precision and testability
- Parsimony
- Heuristic value
Comprehensiveness
- Broad enough to account for as much data as possible
- If there are data relevant to a theory that it cannot for: Either adapt the theory to account for the new data or Develop a new theory that incorporates the new data
- Example: Can the auditory theory account for perceptual phenomena, such as categorical perception, phonemic restoration?
Precision and testability
- A good theory should have concepts that are clearly and explicitly defined: Contains rational, logically related statements
- Empirically testable hypotheses
- Some ESP believers argue that the presence of a disbeliever can prevent someone with ESP from being able to perform.
- We aim to disprove theories (not prove them) never “prove” a hypothesis
Beware of Omnipotent Theories
- So powerful, general, or flexible that they can account for everything
- Not testable/falsifiable
- Freudian theory isn’t falsifiable (Sir Karl Popper)
- Behavior with a clear sexual or aggressive motivation is proof of the presence of the Id.
- If this behavior isn’t displayed, you have a reaction formation against it.
•Must tread carefully between explaining a lot of data and too much!
need a plausible null that could be true
Parsimony
Occam’s razor
The explanation of any phenomena should make as few assumptions as possible.
All things being equal, the simplest explanation tends to be the best
the fewer assumptions made the better
Heuristic Value
•Makes (basic or applied) predictions, generates new knowledge, stimulates future research
theory is only useful if it can make predictions
important to generate new knowledge and stimulate future research
line of best fit - relates the points and makes predictions
Empirical research
Gather new information through observation/
measurement maybe through sample surveys, test scores, OAE responses, etc.
non-empirical research
• Uses existing information rather than collecting new information
ex: literature reviews
quantitative research
- Numerical data or data to which statistics can be applied
- Quantify attitudes, opinions, behaviors, etc.
- Generalize results from a sample to a population
- Methods include surveys, structured interviews/observations, systematic experiments
qualitative research
• Primarily exploratory: Aim to reveal underlying reasons, opinions, motives, trends
• Often used to generate hypotheses to be tested in
subsequent quantitative research
• Methods include unstructured/semi-structured
techniques, verbal measures
•Focus groups, individual interviews, observations, field-work
• Sample size is typically small
Types of Empirical Designs
• Experimental/Quasi-Experimental
Manipulate conditions
e.g., vocoded vs. non-vocoded speech
• Non-Experimental/Observational
Examine existing conditions
e.g., cochlear implant patients vs. normal-hearing individuals
can’t randomly decide who might be in a CI group and who would listen to vocoded speech
Independent Variable (IV)
- The input
- Characteristic or manipulation of study
- Maybe experimental or non-experimental
Dependent Variable (DV)
The output measure that depends on the IV
PICO acronym
Patient Population
Intervention
Comparison
Outcome
PICO:
Patient populations
age, sex, culture, ethnicity, health status, condition, impairment, disorder, handicap
PICO:
Intervention
Therapeutic strategy/approach, risk factor/behavior, assessment tool/technique, service delivery, referral, case management
PICO:
Comparison/Alternative
alternative therapeutic strategy, no intervention, alternative assessment/tool, no risk factor/behavior
PICO:
Outcome
short-term goal, long-term goal, function, ability, accuracy, employability, quality of life, diagnosis, rate of recurrence/relapse
PICO Questions:
Etiology-
Are P who I compared with those who C at greater risk for developing O?
Prediction-
For P, how does I compared with C predict future O?
Diagnosis-
When assessing P, is I more accurate than C for diagnosing O?
Prevention-
In P, is I better than C in preventing O?
Intervention-
when working with P, does I or C result in better O?
Management-
Do P who receive I compared with C report greater or fewer O?
Null Hypothesis
Statement of no difference H0
Well-Formed Hypotheses
- Constructs are operationalized
- Precise, specific IV and DV
- General statement –> Specific
- Children with language disorders –>
- Adults with bilateral hearing loss –>
Operationalized - made into something you can measure
• Measures are valid
* the mean of the points on a bullseye is the target
• Measures are reliable
* the points on a bullseye are clustered
Validity
what are the types?
• How accurately a measure represents the knowledge/ skill/trait you aim to assess
Empirical forms:
1) Construct validity
2) Criterion validity
Non-empirical forms:
1) Face validity
2) Content validity
Construct validity
A construct is a set of behaviors that share some meaningful association. Test developers examine participant response behaviors. Look at patterns/relationships among test items.
convergence: when a tester performs in a similar way among test items
divergent/discriminant relationships- emerge when a researcher compares test items that represent different behavioral constructs - ex: analysis of test items for cognition and motor skills
underlying theory
Criterion validity
*extent to which a measure is related to an outcome
predictive - SAT scores related to how well you do in college
concurrent - if you got this score on the SAT you would get this score on the ACT (same time, same person)
Reliability
* list the types
- Consistency of measurement (especially critical when drawing conclusions for an individual)
- Inter-observer- when two different examiners use the procedure to test the same person
- Test-retest- test each participant at two different points in time
- Parallel-forms- administer one form to a group and then sometime after administer the second form to the group
- Split-half
split half reliability
using 1 half of the items and the other - even odd (focus and attention)
QALMRI acronym
Question Alternatives Logic Methods • How to implement the design logic Results • What was the outcome Inferences • What do the results means
First paragraph of a paper
Start with broad, global ideas
Subjectivity
• Try not to insert yourself into the article
(e.g., “I thought that…”)
• Nor the reader into the article (e.g., “You
may be surprised to learn that….”)
• Minimize phrases like “interestingly” or
“surprisingly”
• Show that the results are interesting (by setting up a convincing argument), don’t tell.
Last Paragraphs
- End with a summary of what your study is about and what you’re actually doing.
- Provide a lead-in to what you’re doing in this study • Summarize the specific research question
- State hypotheses and predictions
Intro Critique Considerations
- In general, is the Intro well written?
- Are all of the structural elements of an Intro present?
Research Problem
• Is there sufficient justification from the literature that this
problem is significant and warrants additional research?
Hypotheses
• Is the question testable?
• Can you easily identify the IV and DV from the questions/hypothesis?
• Does the question include constructs that are operationally defined?
