I2, Sampling and Experiments Flashcards

Descriptive statistics

Statistics used to summarise or present data

Examples of descriptive statistics

the arithmetic mean, median, and mode, plots that summarise the observed data such as histograms and scatterplots

Inferential statistics

As a subject, refers to generalising from samples to populations

Examples of Inferential Statistics

Hypothesis testing, checking for relationships between variables, making predictions

Target population

The entire group of subjects (e.g. people, animals, objects) for which it is desired to estimate some characteristic of interest

Study population

The subset of the targe population from which the sample is drawn, and any example could have been chose for inclusion

Possible issue with study population

If the study population is too narrow, it may not be possible to infer the characteristic of interest in the intended target population

Sample

The subset of the study population that is observed

Population parameter

A single value describing the characteristic of interest in the whole population, it is assumed to be fixed but unknown

Sample statistic

A single value describing the characteristic of interest in the sample, it is known but subject to sampling variability

Biased sample

A sample that does not represent the population well

Precision

The variability in the results

Precise + biased sample (confidence interval)

The confidence interval will be narrow but unlikely to contain the true population parameter value

Unbiased + imprecise sample (confidence interval)

The confidence interval will likely include the true value of the population parameter, but it will likely be much wider than that of a precise sample

Biased + imprecise sample (confidence interval)

The centre of the confidence interval will not be near the true value of the population parameter, however it might still be contained in the wider confidence interval

Improvement through greater precision

Greater precision improves your inference only when your sample in unbiased, for a biased sample greater precision makes it more likely that the population parameter is not contained in the resulting confidence interval

Simple random sampling

Units are selected at random (e.g., using a random number generator)

Stratified sampling

Units are selected by systematically diving the population into strata (groups) and units are sampled from within each stratum (e.g. when the subjects included in the sample are humans, the strata could be age groups)

Systematic sampling

Every k-th unit from the population is included in the sample

Cluster sampling

The population is broken into many groups that are representative of the population and (randomly) a fixed number of groups are selected, from which all units are included in the sample

Convenience sampling

where a sample is selected by taking the members of the population that are easiest to access (e.g. surveying university students), this likely results in a biased sample

Voluntary response sampling

when people choose to be included a sample by responding to a broad appeal (e.g. online polls or course evaluations), this likely results in a biased sample

Subjective sampling

When the researcher uses their subjective judgement to choose a sample they think is representative of the population (e.g. choosing a diverse focus group from a class register), this likely results in a biased sample

Observational study

A study in which the researchers do not take any action that affects the collected data

Confounding variable

A variable which impacts both the response and explanatory variables

Longitudinal study

A group of subjects are studied over time, and measurements are recorded at set time points, a type of observational study (e.g. an environmental scientist might take water samples from lakes that are extracted every two weeks to identify possible changes in water quality)

Cross-sectional study

A group of subjects are studied at a specific time, a type of observational study (e.g. a group of athletes might participate in a sporting event and their physical characteristics are studied to see if any particular features appear to be related to better performance)

Case-control study

a group of subjects with a certain characteristic are compared to a control group without that characteristic to identify possible associations between the characteristic and an outcome, these are retrospective and a type of observational study (e.g. individuals with a rare disease can be matched with other similar individuals without the disease to assess previous exposures that might have contributed to the condition. )

Cohort Study

A specific group of individuals are followed though time, the group is not selected based on a spefific characteristic of interest, these are prospective and a type of observational study

Hawthorne effect

A phenomenon where individuals modify their behavior in response to being observed or studied, which can impact the validity of research findings.

Explanatory variable

A variable that is manipulated or categorized to examine its effects on a response variable in a study.

Response variable

A variable that measures the outcome or effect in a study, influenced by the explanatory variable.