Critical Analysis Flashcards
What should you look for when trying to critically analyse an introduction?
Sufficient background, Appropriate citations, Clear and justified aims, a clear hypothesis
What should you look for when trying to critically analyse a methods section?
A method which appropriately addresses the aim, appropriate statistical testing, clear descriptions, ethical considerations, a justified sample
What should you look for when trying to critically analyse a results section?
Clear presentation of findings, appropriate figure and table legends, appropriate analysis.
What should you look for when trying to critically analyse a discussion section?
Is there a summary of their main conclusions, are they hiding anything?, do the results suggest something different, is their a comparison to other research?
What should you look for when trying to critically analyse the abstract and title?
Good or bad at describing the work? Is it an overstatement etc…
Name the 2 different types of experimental factor (qualitative) and describe them (use examples).
- Nominal
Named categories such as male/ female - Ordinal
Ranked in an order or scale i.e. agree, strongly agree etc…
Name the 2 different types of experimental covariate (quantitative) and describe them using examples.
- Discrete
count- number of children in family - Continuous
Measure such as length
Describe different types of graphs and their uses.
- Scatterplot – continuous data
- Histogram – continuous data – Frequency chart; shows number of times get each value. Helpful to show distribution of data
- Bar chart – categorical, discrete and nominal • Pie chart – categorical, discrete and nominal
- Boxplot – categorical, discrete and nominal
Discuss the pros and cons of expressing data as a mean.
The average of a dataset is useful when the data is normally distributed.
However the mean can be a poor representative of data when there are a large number of outliers (abnormal distribution).
Discuss the pros and cons of expressing data as a median
The number is less skewed by outliers but more likely to not represent full dataset.
Discuss values which can show the variability of data.
- standard deviation (SD) = variation in the values of a variable – As you collect more data, SD of the population will be more precise – this could increase or decrease
- standard error of the mean (SEM) = spread that the mean of a sample of the values would have if you kept taking samples. – How precisely do you know the true mean of the population – Takes into account the sample size – Tends to be less than SD, especially as sample size increases
Describe different techniques for normality testing.
• Anderson-Darling – good for detecting non-normality in tails
• Ryan-Joiner or Shapiro-Wilk – not for large data sets (n>50) – If RJ near 1, the population is likely to be normal
• Kolmogorov-Smirnov – requires high n (>1000)
High p-value (>0.05) = normal distribution
• Transform non-normal data to achieve normality
Describe parametric statistical testing.
Parametric stats – Normal Data
• Assume a particular data distribution
• Have more statistical power
• Not robust
Describe non- parametric statistical testing.
Non-parametric stats – Non-normal data
• Do not assume a particular distribution
• Less statistical power
• More robust
What kind of probability tests should be used on parametric data comparing 2 groups of data?
2-sample t-test or Paired t-test
- tests whether difference occurred by chance
