Monday - Reliability

13 important questions on Monday - Reliability

What to do against random errors?

- standardisation (instruction and procedures)
- repeat measurements and take mean
- in case of questionnaires/exams: ask more questions
- more subjects in experiment

What are random and systematic errors?

Random: occur randomly
Systematic: defect measurement equipment, misreading etc.

What to do against systematic errors?

- standardisation
- use gold standard to assess bias
  • Higher grades + faster learning
  • Never study anything twice
  • 100% sure, 100% understanding
Discover Study Smart

What is included in noise?

Natural variation + measurement error

What is the repeatability coefficient?

2Sd (standard deviation) of the difference between the two measurements
- when two measurements are taken under identical conditions (close in time)

What is the reproducibility coefficient?

Also 2Sd of the difference between the two measurements
- when two measurements are measured under possibly different conditions (different weather, observers, instruments etc).
- no systematic differences

How are the residuals calculated?

Natural variation (σp) + instrument error (σ e)

--> Variance =   σ^2 tot = σ^2 p + σ^2 e
σ tot = sqrt(variance)

What is a different name for the reliability coefficient?

Intra-class correlation (ICC). Is the p in the sample size formula for the paired t-test

Precision can be influeced by taking multiple measurements, the reliability when taking multiple measurements can be calculated with which equation?

See equation (also on formula paper)
h is the amount of measurements done

What is a bland-Altman plot?

Measure with 2 devices and calculate the mean (xi + yi)/2
Also calculate the differences di = yi - xi
Plot with x-axis is mean and y-axis is difference (di vs mean)

What is the con of a Bland-Altman plot?

We cannot decide which instrument is better, only if the two instruments measure equally

What is Cohen's kappa?

Agreement between two graders, score must be given on qualitative (ordinal) scale

How do you calculate Cohen's kappa?

Make a cross-table where the rows contain observer 1 and the columns observer 2.
Calculate the complete agreement, this are all counts on diagonal and divide this by the total observations. 
Calculate the expected agreement = P11 + P22 + P33 etc. = (32/50)*(29/50 + (8/50)*9/50) + (12/50)/(10/50)
Then, k = (Po - Pe)/(1 - Pe)

The question on the page originate from the summary of the following study material:

  • A unique study and practice tool
  • Never study anything twice again
  • Get the grades you hope for
  • 100% sure, 100% understanding
Remember faster, study better. Scientifically proven.
Trustpilot Logo