Definitions Flashcards
Falsification
Hypothesis testing
Hypothesis
Statement of relationship between variables
Null Hypothesis
No difference. Presumed that groups have the same results regardless of the treatment.
Standardised
Can be repeated and verified
Validity
Must measure what it’s intended to measure. Credibility.
Reliability
Must be repeatable with consistent results. Dependability
BHC temporal relationship
Exposure always precedes the outcome, essential presence
BHS strength
Stronger the association the more likely the relation of ‘A’ to ‘B’ is causal
BHS dose-response
Increasing exposure increases the risk
BHS consistency
Find same results consistantly
BHS Plausibility
Agrees with current understandings of pathological processes (has theoretical base)
BHS Consideration of alternative explanations
And effectively ruled them out
BHS experiment
Condition can be altered and prevented by appropriate experiment
BHS specificity
When single cause produces specific effect (weakest criteria)
BHS coherence
Association should be compatible with existing theory and knowledge.
Paradigms
Patterns of belief and general assumptions.
Attention arm
Similar to intervention but without the active ingredient
Population
Target group we are interested in
Simple random (probability sampling)
Random selection of everyone on population list - rare because hard to get population list.
Stratified random (probability sampling)
Put in groups according to characteristics (like gender) and then randomly selected.
Cluster (probability sampling)
Useful when not everyone in population is known. Random selection of larger units (like hospitals) which participants are then randomly selected from
Systematic (probability sampling)
Random selection at predetermined intervals (E.g. every 20th person)
Probability sampling
Unbiased sample, everyone who meets criteria has chance of selection. Can work out probability.
Non-probability sampling
Non-random, chance of being selected cannot be estimated.
Single blind
One person knows which arm they’re in. (might be obvious which treatment they’re receiving). Person assessing the outcome doesn’t know which.
Double blind
Neither the participant nor the person assessing the outcome knows the arm (E.g. placebo)
Internal validity:
Results legitimate because of the way the study was conducted, did independent really change dependent?
Another word for external validity
Generalizability
Hawthorne effect
Response to being in a study
Descriptive statistics
Ways of displaying and summarising data
Nominal measurement
Labelling variables, no quantitative. (M/F, hair colour, area you live).
Ordinal measurement
Order is important but differences between each is unknown
Interval measurement
Know the difference between values, intervals. 50-60 degrees
Ratio measurements
Same as interval but has clear 0 (height and weight)
Mean
Arithmetic average
Interquartile range
Difference between 20% and 75%
Standard deviation
Subtract mean from each number, mean of remaining values.
Variance
Standard deviation multiplied by self
P value
Probability of obtaining your study results if the null hypothesis is true
What can the P value be?
Between 0 and 1. (Closer to 0 =more likely null hypothesis should be rejected -> so there’s a difference, you were right).
Statistical significance
- Is often set at 5%.
- If P≤0.05 it’s closer to 0, evidence to reject null hypothesis.
- If P≥0.05 there is insufficient evidence to reject null hypothesis
Power study
Probability of being able to detect difference between the study groups. Usually %. E.g. 80% power -> 80% chance of detecting difference.
Confidence interval
Precision of the quantity of interest that is estimated
Emic perspective
Insider’s point of view
Constructivism
Construct social words. Individuals create meaning through interactions.
Ethnography
Study of culture. Origins in Anthropology
Phenomenology
Study of phenomena/ lived experience of individuals
Grounded theory
- Idea is to generate theory more than description.
- Specific set of methods
- Hypotheses generated
- Developed by Glaser and Strauss
Homogenous sampling
Opposite of maximum variation sampling. Example: first placement experience of male students.
Theoretical sampling
Uses grounded theory. Used to find participants who will help the research build the theory.
Data satuation
guides the sample size. Sampling stops when enough data has been collected, size not pre-determined.
Concurrent data analysis
Reducing the data, analysing it during data collection and in grounded theory.
Constant comparative analysis
To develop hypotheses, test them out at subsequent interviews.
Thematic content analysis
Themes are given a code (word or phrase) Codes then collapsed into categories. After data already collected.
Framework analysis.
Put data into categories..
Objectivity
Confirmability.
