Structural analyses of egonets are only interesting if we have 1.5 (alter-alter) data. We can apply the same procedures to these 1.5 egonets that we use with whole networks tomorrow.
Ron Burt (Structural Holes 1992) developed a significant array of structural measures specific to 1.5 egonets. These are available in UCINET. We need to collect your alter-alter data and map your personal community in order to explore these measures.
Collecting alter-alter data: The standard half matrix (Matrix editor)
The standard way of collecting alter-alter relations is to present the respondent with half of the sociomatrix and ask them how well do each pair know one another.
In classic SNA surveys I.5 data is collected as direct, inter-personal ties. The interview constructs a matrix, or half matrix from the name generator names and then asks the respondent for each pair of names: Does A know B? Does A know C? Does A know D? and so on. This becomes very time consuming and tedious with more than about 15-20 names so most studies take just a selection of names rather than the full list.
Here is Ron Burt’s instrument for collecting Alter-alter relations.
We can create a matrix of alter-alter ties directly in UCINET. Open Data -> Data editors ->Matrix editor and load the MyNames dataset into it. Maximise the screen for data entry comfort. (Note: although you can also open the file in the ‘Excel matrix editor’ it does not allow have the flexibility of the Matrix editor.)
You can now enter an estimate of the tie strength for the alter-alter relations. The conventional question is ‘How well do A and B know one another?’ This question prompts the respondent to think about undirected (bi-directional) ties rather than di-ties. It is asking about relationships rather than (directed) relations. You can automatic the entry process for this situation by checking the option for ‘Symmetrise as you type’. (Remember a symmetrical matrix has the same value for the From-To and To-From cells.)
- UCINET matrix editor has opened you MyNames dataset. It is displayed in matrix format.
- The node ID labels are given as the row and column headers. The ego-alter tie strengths appear. (Note: the symmetrical, alter-ego tie strengths are not in the dataset.)
- Symmetrise as you type is checked.
- The two entries I have made are entered as bi-directional data.
Read along the row for you first alter, find the diagonal (reflexive tie) and go to the next cell. Enter an estimated tie strength. Use 3 (Very close), 2 (Somewhat Close) 1 (Acquaintance) and 0 (Strangers) – the same ordinal scaling we used for the ego-alter ties.
When you are done save this as a UCINET dataset MyPersComm. (If you did not create a new default folder (MyPersComm) you can do it at this point.)
Note: The matrix format works quite well for up to 20 or so names but may become tedious for the respondent if there are more. However, by filling out the rows, rather than columns the task gets easier as they progress.
Read the new tie dataset into NetDraw. Now read in the attribute data you created for MyNames and use it to work with your diagram. (I have set the node colour by Gender and Size by Age).