Planning The Measements: Precision And Accuracy Flashcards
Measurements
Describe phenomenon that can be analyzed statistically
Precise
Free of random error
Accurate
Free of systematic error
Continuous variables
Are quantified on an infinite scale
Discrete variable
Units who are limited to integers. Can resemble continuous variables if large enough values are used
Categorical variable
Not suitable to be quantified
Dichotomous variable
Categorical variable with two possible values
Ordinal categorical variable
Categories that have an order (e.g. Severe, moderate, mild pain)
Random error
The greater the error, the less precise the measurement
Three main sources of random error
- Observer variability refers to variability in measurement that is due to the observer, and includes such things as choice of words in an interview and skill in using a mechanical instrument
- Instrument variability refers to variability in the measurement due to environmental factors such as temperature, aging mechanical components, different reagent lots, and so on
- Subject variability refers to intrinsic biologic variability in the study subjects due to such things as fluctuations in mood and time since last medication
Within-subject standard deviation
Reproducibility of continuous variables depends on this standard deviation
What should be used if the standard deviation and the participant’s mean demonstrates a linear association?
The coefficient of variation
How does one assess precision in categorical variables?
Using percent agreement and the kappa statistic
What are the 5 strategies for enhancing precision?
- Standardizing the measurements
- Training certifying the observers
- Refining the instruments
- Automating the instruments
- Repetition
Accuracy
Is the degree to which it actually represents what it is intended to represent
What are the three main classes of measurement error?
- Observer bias is a distortion, conscious or unconscious, in the perception or reporting of the measurement by the observer.it may represent systematic errors in the way an instrument is operated, such as a tendency to round down blood pressure measurements, or in the way an interview is carried out as in the use of leading questions
- Instrument bias can result from faulty function of a mechanical instrument
- Subject bias is a distortion of the measurement by the study subject, for example, in reporting an event (respondent or recall bias).
Gold standard
Accuracy of a measurement can be best assessed by comparing it with a “gold standard”—a reference technique that is considered accurate
What are the three ways to view and assess validity?
- Content validity examines how well the assessment represents all aspects of the phenomena under study—for example, including questions on social, physical, emotional, and intellectual functioning to assess quality of life.
- Construct validity refers to how well a measurement conforms to theoretical constructs; for example, if an attribute is theoretically believed to differ between two groups a measure of this attribute that has construct validity would show this difference
- Criterion-related validity is the degree to which a new measurement correlates with well-accepted existing measures.
Face validity
Uses subjective judgments
Predictive validity
The ability of the measurement to predict an outcome: the validity of a measure of depression would be strengthened if it was found to predict suicide
What is a general approach to validating an abstract measure?
Search the literature and consult with experts to find a suitable instrument
Measurement sensitive
Measurements should be able to detect differences in a characteristic that are important to the investigator
Measurement specificity
A measurement should be specific so therefore should only represent the characteristic of interest (e.g. The carbon monoxide level in expired air is a measure of smoking habits that is only moderately specific because it can also be affected by other exposures such as automobile exhaust
Measurement appropriateness
Measurements should be appropriate to the objectives of the study. A study of stress as an antecedent to myocardial infarction would need to consider which kind of stress (psychological or physical, acute or chronic)
Distribution of responses
Measurements should provide adequate distribution of responses in the study population. Values should be dispersed
Efficiency and parsimony of a study is described as what?
As collecting data with measurements that are at an affordable cost in time and money
Operational manual
Describes the method for conducting and recording the results of all measurements made in the study. To standardize instructions from examiner to examiner and from subject to subject, the protocol includes a script of instructions to be read to the participants verbatim
Precision (reproducibility, reliability, and consistency)
Is the degree to which it is reproducible, with nearly the same value each time it is measured
What are the 7 strategies of enhancing accuracy
- Standardizing the measurement methods
- Training and certifying the observers
- Refining the instruments
- Automating the instruments
- Making unobtrusive measurements
- Calibrating the instrument
- Blinding