Error in Research Flashcards
What is the actual goal of research?
To conduct a study that gives us an estimate of the XY association that is as close as possible to the truth
Why is it important to examine study accuracy?
To assess how precise our study was and how valid the results were; tells us whether or not we have a reputable study
What are the 2 components of study accuracy?
Precision + Validity
What is the difference between systematic & random error?
Systematic error is concerned with bias (validity) whereas random error is concerned with precision
What are the 3 types of systematic error in studies?
- Selection Bias
- Measurement Bias
- Confounding
Name the 4 types of selection bias
- Response Rate
- Volunteer Bias
- Healthy Worker effect
- Loss to follow-up
Name the 4 types of measurement bias
- Recall bias
- Social desirability bias
- Interviewer bias
- Observer bias
What are the 2 types of random error in studies?
- Sampling error
- Measurement error
Why can we not determine the true effect of X on Y within research?
Because no study is completely error-free (so the true relationship between X and Y cannot be known)
What is validity in terms of research?
Validity is how much we should believe the findings in a study to be close to the truth
How do we determine whether a study has validity?
by assessing the amount of error/bias a study has
If a study has error should we automatically ignore the findings?
No, because all studies have some degree of error
What is internal validity?
the extent that study results are representative of the sample
What is external validity?
The generalizability of study results to the population
Can we have external validity without internal validity?
No. The study findings aren’t accurate to the sample so it cannot be generalized to other populations.
When darts are clustered together in the same place on a dart board but not on the bullseye, what does this tell us about study accuracy?
The study is precise but lacks validity
What is selection bias?
systematic error due to selection & maintenance of sample
Explain response rate
the requirement that 70% or more of a random sample agree to a study-otherwise creating response bias
What studies are Response bias applicable to?
One’s that use random sampling
What is volunteer bias?
Idea that those who are healthier are often more likely to volunteer for a study
What studies can volunteer bias be a concern?
studies that use a convenience sample
What is the healthy worker effect?
The idea that selecting a sample that are all individuals belonging to an organized group may be healthier on average than the rest of the population, creating selection bias
If we have a study with membership bias do we automatically disregard the findings?
No. We must keep in mind that the study results will have low external validity unless generalized to a similar sample population/membership.
How can the internal validity of a study with healthy worker effect be compromised/reduced?
When the comparison group is from the general population and not the same source population (organization) as the cases
How can we avoid volunteer bias?
By using a random sample
What is loss to follow-up concerned with?
The maintenance of a sample over time; we need to maintain at minimum 70% of the sample to avoid bias
What studies can loss to follow-up be a concern for?
Any studies that have more than 1 data collection point (follow-up on samples)
What is measurement bias?
systematic error due to the way data was gathered in a study
What is observer bias?
When the ascertainment of Y in participants is not made independent of the researcher’s knowledge of the X status.
When is observer bias relevant?
When Y is determined by the researcher using signs rather than a biological test
What do we use to identify whether random error is present?
The CI: large CI that crosses effect sizes suggests random error is present
How do we reduce sampling error?
By using a large sample
What is measurement error?
Random error in instruments used to measure X or Y because they were not calibrated correctly or researcher was not trained how to use the instrument properly
What can we say about a study that has high systematic error but low random error?
It is precise but lacks validity
