Research and Statistical Reasoning Flashcards
Non-reponse bias
Error caused when certain types of people are less likely to respond to a survey
Self-selection bias
Error caused when people are randomly allowed to choose to participate in a particular study over another
Publication bias
Error caused when undesirable data is excluded from a publication
Type I error
The null hypothesis is incorrectly rejected
Type II error
The alternative hypothesis is accepted when it is actually false
Generalizability
The ability of a research study to extrapolate their data (sample group) to a larger group or population. The data of the sample group should represent the group as a whole
Reliability
When the results are consistently obtained when retesting under the same conditions
Reporting bias
Error caused by the tendency to ignore unexpected results or explain them away as statistical error
External validity
Ability to apply results found in one population to another. It checks whether the casual relationship in the study can be generalized or not
Internal validity
The extent to which the experiment is free from errors and any difference in measurement is due to independent variable and nothing else. The focus is on the research methods
Recall bias
Error caused when participants are required to recall information (sometimes incorrectly)
Observer bias
Error caused by the categorical misclassification of information due to observer perception of participant status
Interview bias
Error caused by improper standardization of patient interviews
Exposure identification bias
Error caused by an incorrect classification of a percentage of patients. Ex/ Saying 30 patients are obese when not all patients are obese
FINER method
Is the study feasible, interesting (does it have utility), novel (has it been done before), ethical, and relevant?
Positive skew
The tail is on the right. The mean is more than the median. The mode is the top of the bell
Negative skew
The tail is on the left. The mean is less than the median. The mode is the top of the bell
Response bias
tendency of subjects to systemically respond to a stimulus in a particular way (one that makes them seem more desirable) due to non-sensory factors (memory, motive, emotion, experience)
Confirmation bias
tendency to focus on info that fits an individual’s beliefs
Bimodal distribution
If there is sufficient separation between the two peaks, the data can be analyzed as 2 separate distributions
68-95-99 rule
95% of the data within a normally distributed data set must fall within 2 SD of the mean. This means that only 5% of the data falls outside of this range
Outlier
When a data point is more than 3 SD from the mean
Mutually exclusive events
Outcomes that can not occur at the same time
P value
The probability that we report a difference between 2 populations when one does not actually exist. For data to statistically significant, the p value must be < 0.05
Significance level (Ɑ)
If the Ɑ > the p value, then we reject the null (**there is statistical difference between 2 groups)
If the Ɑ < the p value we fail to reject the null (there is no statistical difference)
Ɑ = 0.05
Power (1-β)
The probability of correctly rejecting a false null hypothesis (reporting a difference between 2 populations when one actually exist)
Confidence
The probability of correctly failing to reject a true null hypothesis (reporting no difference between 2 populations when one does not exist)
Confidence intervals
A range of values from the sample mean and standard deviation used to find how confident we are in the accuracy of the true value of the population
Family pedigree chart symbols
Square = male
Circle = female
Filled in shape = effected
Independent variable
what is manipulated. What varies between control and experimental. Ex/ genetics and environment (twin studies)
Dependent variable
what is measured. What is expected to change
Positive control
A control group that ensures a change in the dependent variable when a change is expected
Negative control
A control group that ensures no change in the dependent variable when no change is expected
Systemic error
failing to provide clear and detailed instructions to the participants
Single blind experiments
Only the subject or the assessor (person who takes measurements) is blinded
Double blind experiments
The investigator, subject, and assessor all do not know the subjects group
Cohort studies
Observational approach where subjects are sorted into groups based on differences in risk factors and then assessed at various intervals to determine the outcome of each group
Cross sectional studies
Observational approach where subjects are categorized into different groups at a single point in time. This examines an entire population
Ex/ prevalence of lung cancer in smokers vs. non-smokers at a given time
Case control studies
Observational approach where a number of subjects with or without a particular outcome are identified and their history for exposure to risk factors is assessed
Selection bias
Error that occurs when subjects used for a study are not representative of a target population
Detection bias
Error that results from educated professionals using their knowledge in an inconsistent way
Observation bias
Also known as the Hawthorne effect. This is error caused when the behavior of subjects changes because they know they are being studied
Confounding bias
Data analysis error. The data may or may not be flawed, but an incorrect relationship is characterized. Meaning that a “third party” variable or a confounding variable could potentially be the cause of a relationship
Sufficiency
To be able to say that certain conclusions are able to be drawn from a study (or that the study is sufficient), all variables need to be accounted (or controlled) for
Actor-observer bias
Actors attribute their own behavior to situational attributions, whereas observers attribute the actors behavior to dispositional attributes
Null hypothesis vs. alternative hypothesis
Two populations are equal vs. two populations are not equal
Confounding variables
Experimental variables that have an unforeseen effect on the dependent or independent variables which complicates a relationship that can be seen between the two
