Measurement: Scales, Numbers, Rates, Ratios and Risk Flashcards
In a sample population, give three examples of how we want it to be shaped.
Representative - The sample group should be able to represent a general population
Unbiased - want it to be as true as possible
Precise - no uncertainty
Give two types of error that can occur in a study that may influence the results.
Chance (Random error)
Bias (Systematic error)
Explain random error. How can it be reduced?
It is due to sampling variation. If you pick out of the hat you might get unlucky and get a sample that is not representative for the general population. This can be reduced by having a larger sample size, i.e. a bigger hand picks out of the hat.
Explain systematic error. Will an increased sample size reduce the this error?
Systematic error is quantified by the difference between the true value, and the expected value. A systematic value is not by chance but by inaccuracy. So if we keep doing an experiment the same way over and over again using inaccurate equipment, the systematic error will persist. However the random error might correct itself if there is no systematic error.
How does precision relate to random error?
As you increase the sample size, the precision also increases. Thus reducing random error.
Give three examples of selection bias in systematic error.
External validity (study sample) Internal validity (group selection within a study) Healthy worker effect
Explain external validity.
A study sample that is not representative of the entire population in interest. For example if we want to measure obesity in the UK, having a sample that is only from university students is not a good representation of the entire population. It would be a good representation for obesity in university students though.
Explain internal validity.
When a group within a study may not be comparable to another group within the study. For example measuring mortality of smoking in a population where there is an age difference. Old people are more likely to die than young people, and therefore not giving an accurate representation.
Explain healthy worker effect.
If you use a sample of workers, workers usually exhibit a lower overall mortality than the general population as you don’t take severely ill, disabled, drug addicts, homeless etc, into account.
Give four additional examples of systematic error.
Recall error
Observer or interviewer error
Measurement error
Misclassification
Explain recall error.
A difference in recollection from study participants regarding events or experiences from the past. For example if two women were exposed to the same environment during pregnancy, if one woman has a miscarriage she is more likely to remember things that happened during the pregnancy, than the woman that had a healthy birth.
Explain observer error.
If the study observer has a preconceived expectation or knowledge this can influence the result. For example if you know a patient is alcoholic, you are more likely to think liver disease.
Explain measurement error.
If the observer measures faulty. If the tool used to measure is not correct. If you use two different tools to make the same measurement, calibration can come into play. If two different observers make the measurements, one observer might do it slightly differently, influencing the result.
Explain misclassification.
When a participant is put into the wrong group, for example diseased when they are no diseased. This usually arises from observer error or measurement error.
Define prevalence
The proportion of people who have a disease at a given point in time. It is not a rate!
