Networking Flashcards
(22 cards)
How do you test your hypothesis in a network analysis?
- Calculate a coefficient to describe the relationship between your network measure and biological measure
- Randomise the dataset and take the coefficient - do this many times
- This allows us to know is what we observe different to what we would expect by chance?
What is a frequency distribution and what would be on the axes?
The frequency distribution is essentially the null distribution.
X-axis = our randomised correlation coefficients
Y-axis = frequency of the correlation coefficients
What happens if our observed coefficient is within the random distribution?
Its means that what we observe is no different to what we would observe by chance (e.g., non-significant)
Why are randomisation tests used for network analysis?
They don’t rely on the assumption that the data is independent
What is dyadic data?
Data measured between pairs of individuals or entities
What is a vertices?
Individuals or units
What is an edge?
Pairwise associations/interactions between individuals or entities
What does it mean for the edges to be weighted?
It shows the strength of the pairwise interaction
What does it mean for the edges to be directed?
It shows the direction of the interactions (e.g., who initiated it)
What are networks based on?
Graph theory
What is the mathematical representation of a network?
Matrix representation that compares the interactions with individuals in a network.
The higher the number, the stronger the interaction
How do you quantify the structure of networks?
- Whole network structure
- Dyadic relationships
- Individual based networks
What is the distance/path length in whole network data?
The average between two pairs of individuals/entities in the network. It looks at how many individuals you have to go through between the two individuals you are interested in via the shortest path.
What is clustering in whole network structure?
The degree to which the individuals you interact with are also interacting with each other.
What is small-world network?
Distance vs. clustering allowing us to quantify how connected each individual is in the data set.
What are dyadic relations in the network data?
Using the network to characterise the pairwise relationships.
Indirect: distances in network space (e.g., can show social space and geographical space)
Direct relationship: do not go through other individuals for this relationship (edges can be weighted to show the strength of the interaction)
What is individual based metrics?
Where an individual is in their network
What is the degree?
The number of direct connections an individual has in the network
What is the strength?
Summing the edges weight of the individuals to show how strong the interactions are (e.g., an individual that is more strongly connected may be more prone to parasite transmission)
What is betweenness?
A measure of how important an individual is in the flow through a network.
An individual with high betweenness is being passed through a lot.
What do you calculate to test the hypothesis for network analysis?
The degree is calculated for each individual
Why are networks not independent?
Measures are dependent on each other as it is the interactions they have with other individuals. If you remove one individual from the network, it affects the measurements of degree for all other individuals.
