week 1: design/statistical power & fCM Flashcards
section 1 (65 cards)
What is statistical power?
Statistical power is the probability of correctly rejecting the null hypothesis when it is false. It reflects the ability of a study to detect an effect if one exists.
What are the four key components affecting statistical power?
- Effect size
- Sample size (n)
- Significance level (α)
- Variance (σ²)
How does increasing sample size affect power?
It increases power by reducing standard error, making it easier to detect a true effect.
What is the relationship between power and Type II error (β)?
Power = 1 - β. A higher power means a lower probability of a Type II error (failing to reject a false null).
Why is power analysis important in experimental design?
To ensure enough participants/data are collected to detect a meaningful effect, avoiding underpowered (wasted) or overpowered (unethical) studies.
What is flow cytometry used for?
To measure physical and chemical characteristics of cells or particles, typically cell size, complexity, and fluorescence.
What are the main components of a flow cytometer?
- Fluidics system (guides cells through the laser)
- Optics system (lasers and detectors)
- Electronics system (signal conversion and data analysis)
What is forward scatter (FSC) vs side scatter (SSC)?
- FSC correlates with cell size
- SSC correlates with granularity/internal complexity
What is gating in flow cytometry?
Selecting specific cell populations from data based on scatter or fluorescence to analyse subsets.
How is fluorescence used in flow cytometry?
Fluorescent-labelled antibodies bind specific cell markers; lasers excite the fluorophores which emit light detected as a signal.
What is the null hypothesis (H₀) when comparing male and female heights?
That there is no difference in height between men and women.
What is the alternative hypothesis (H₁) in the height comparison example?
That men’s height is not equal to women’s height.
What key considerations are needed when designing a study comparing two groups?
Sample size (N), data collection method (e.g., measured vs self-reported), inclusion/exclusion criteria (age, ethnicity, disability), and randomisation.
What type of data analysis is typically used in group comparisons like male vs female height?
Descriptive stats (mean/median), distribution plots (histograms), and statistical tests (e.g., t-test).
What defines a manipulative experiment?
It involves deliberately altering one or more factors to explore cause and effect.
What are examples of manipulative experiments?
Testing DNA recovery in wet vs dry conditions
Measuring cancer cell counts under varying drug concentrations
Comparing DNA yield from different extraction methods
What defines an observational experiment?
Observes variables in natural conditions without manipulation.
Give examples of observational research questions.
Effects of smoking on lung function
Identifying cancer survival biomarkers
Why is statistical power important in clinical research?
To ensure that the study can reliably detect a real difference, avoiding wasted resources or ethical risks.
What happens if the sample size is too small?
You may fail to detect a true effect (Type II error).
What happens if the sample size is too large?
Subjects may be unnecessarily exposed to harm or burden.
What is the purpose of power analysis in study design?
To calculate the minimum number of samples needed to detect an effect with confidence, balancing ethical and statistical needs.
What are the four types of power analysis?
A priori – before the study (to plan sample size)
Post hoc – after the study (not recommended)
Compromise – adjusts power, N, and alpha simultaneously
Sensitivity – determines minimum effect size detectable with given N
Why is post-hoc power analysis discouraged?
Because a non-significant result doesn’t confirm the null hypothesis; post-hoc assumes an effect exists without evidence.
