week 13 Flashcards
Why use logistic regression in a study?
You can study the association between exposure and outcome while controlling for other factors that might introduce confounding.
In a study about the role of body mass index and physical activity on mental health, the research assistant sometimes forgets to ask study participants to take their shoes off when they measure height and weight. This is an example of:
bias
In a hypothetical study of self-reported latex allergies, a survey is given to patients in an allergy physician’s office before they see the doctor. The survey questions include if the patient has any allergies, and if so to what kind. The survey results were compared with the physician’s diagnosis of allergies (considered the gold standard) after the patient’s visit. What kind of study is this?
validity
The positive predictive value comparing self-reported to measured weight is 82%. How do you interpret this?
82% probability that a person who self-reported as being overweight, is actually overweight
When is a test with high specificity particularly suitable?
When confirming the disease in the event of a positive test result. This is important in the case of rare disorders.
What is sensitivity used for in a test?
to detect disease
In a study of the association between head circumference and later developmental outcomes in infants, the research assistant sometimes used a different tape measure that isn’t standardized. This happened more when measuring male infants than female infants. Which is this an example of?
differential measurement bias
What is quality assurance?
Activities to ensure quality of data before data collection; used to prevent/minimize systematic or random errors in data collection
What is quality control?
Efforts to monitor and maintain quality of data during conduct of study; minimize bias and reliability problems
What are some quality assurance strategies to reduce error?
study protocols, calibrate equipment, training research personnel, pretesting and pilot studies
What are some quality control strategies to reduce error?
watching data collection, check trends, validity studies
What is validity?
the degree to which a tool measures what it claims to measure; accuracy and lack of bias
what is reliability?
the degree to which an assessment tool produces stable and consistent results; precision and reproducibility
What is the interpretation of the percent agreement?
The proportion of individuals who are classified the same
What is the interpretation of the percent positive agreement?
the proportion of individuals who are classified as positive by both tests
