Flashcards
Quantitative Research Terms
Cards in this deck (31)
- 1What is Bias?Deviation of results or inferences from truth
- 2What is Selection Bias?Systematic differences in groups being compared caused by incomplete randomisation (allocation stage)
- 3What is Performance Bias?Systematic differences in care provided apart from intervention being evaluated (intervention stage)
- 4What is Exclusion Bias?Systematic differnces in withdrawals/exclusions of people from the trial (follow-up stage)
- 5What is Detection Bias?Systematic differences in ways outcomes are assessed (outcomes stage)
- 6What is Blinding?Observers/subjects are ignorant to which group subjects are assinged to elimiante bias (may be single, double or triple)
- 7What is Concealment of Allocation?Process used to ensure person deciding to enter participant into RCT does not know which group individual will be allocated (to avoid selection bias)
- 8What is a Confidence Interval?Can show size of likely variation from true figure - 95% CI means 95% chance true result lies within specified range (wide CI indicates less precise estimates of effect)
- 9What is a Confounding Variable?Characteristics that may be distributed differently between study + control groups that can affect outcome (can be due to chance or bias)
- 10What are Controls?Group that acts as comparator for 1 or more experimental observations in controlled trial (e.g. do not have disease of interest)
- 11What is the Effective Size?Standardised effect observed - usually defined as difference in means between intervention + control groups divided by SD of control/both groups
- 12What is Estimate of Effect?Observed relationship between intervention + outcome expressed as e.g. odds ratio, relative risk reduction, mean difference
- 13What is Heterogeneity?In systematic reviews refers to variability/difference between studies:
- Statistical - differences in effect estimates
- Methodological - differences in study design
- Clinical - differences in participants, interventions or outcome measures - 14What is Homogeneity?Degree to which results of studies in systematic review are similar - clinical homogeneity means partcipants, interventions + outcome measures in review are similar/comparable
- 15What is Intention to treat analysis?Individual outcomes in clinical trial analysed according to randomisation group regardless of whether they dropped out, fully complied or crossed over - better measure of effectiveness
- 16What is the Inter-quartile range?Extent of spread of middle 50% of ranked data (reported with median for skewed data)
- 17What is the Median?Value of observation that comes halfway when ranked in order (often used when data are skewed + distribution uneven)
- 18What is the Mean Difference?Standard statistic that measures absolute difference between mean of 2 groups - estimates amount by which on average intervention changes outcome compared to control
- 19What is Normal Distribution?Key assumptions are that data are symmetrically distributed around a mean value - shape of distribution can be described using mean + SD (if assumptions do not hold, non-parametric analyses should be used)
- 20What is the Null Hypothesis?Statistical hypothesis that 1 variable has no association with another variable - or 2 or more population distributions do not differ from one another
- 21What is an Odds Ratio?Measure of treatment's clinical effectiveness - odds of an event in one group to odds of event in another (if >/<1 effects of treatment are more/less than control treatment
- 22What is the P value?Possibility that outcome could have occurred by chance - range from 0 (impossible to happen by chance) to 1 (certain to happen) - normally <0.05
- 23What is Power?Ability of study to demonstrate association between 2 variables - 80% is typical level (80% chance of showing statistically significant treatment effect)
- 24What is Randomisation?Process of allocating participants to one of the groups using random sequence + means of concealing sequence (quasi-randomisation is when not truly random e.g. birth date)
- 25What is a Sample size calculation?Designed to highlight likelihood of detecting true difference between outcomes in control + intervention groups (to work out how large a sample required to have moderate, high or very high chance of detecting true difference between groups)
- 26What is Standard Deviation?Statistical measure that describes extent of data set calculated as average difference from mean value in the sample
- 27What is Statistical Significance?A result that is unlikely to happen by chance - usual threshold is that results would occur by chance with probability of <0.05 if null hypothesis is true
- 28What is Stratification?Groups are separated into mutually-exclusive sub-groups of population that share characteristics (stratified randomisation)
- 29What is a Two-sided (or two-tailed) test?Hypothesis test in which values for which null hypothesis can be rejected are located in both tails of probability distribution (tests whether 1 treatment is either better or worse than another rather than testing whether 1 treatment is only better than another)
- 30What is a Type I error?Conclusion that a treatment works when it actually does not - incorrect rejection of null hypothesis
- 31What is a Type II error?Conclusion that there is no evidence a treatment works when it actually does - incorrect acceptance of null hypothesis
Information
- Description
- Research Methods 29/05/19 Flashcards on Quantitative Research Terms, created by Bryony Williams on 17/05/2019.
- Created by
- Bryony Williams
- Views
- 0
- Rating