This exercise simply invites you to compile your own egonet dataset to familiarize yourself with the mechanics of handling network data. It uses the basic egonet data collection of a name generator and follow-up questions to create a dataset we label MyNames. Using your own egonet data allows you to connect social network research methods to a real world setting.
This exercise is not a full template for an egonet survey, it is just a single case study. The Northern California Community Study (NCCS) (Fischer 1982) is the classic, large-scale egonet survey. Elsewhere we examine the NCCS as a template for egonet survey design and data analysis .
2.3.1: Task One: Egonet data collection: Name generator and follow-up questions
Egonet data collection begins by eliciting a list of network contacts (‘names’ or alters = others in latin) from the respondent (ego = self in latin). These are known as name generator questions. Classically the names are first names, nicknames or IDs that do not identify the alter in any public way. Each ego-alter tie is a specific real world social relation but with unidentified alters we cannot have information about reciprocated (alter-ego) di-ties nor longer paths of social connection. (Egonet research that asks for publicly identifiable names moves into sociometric type network methods.)
Each ‘name’ given by a respondent (ego) specifies a real world social tie (di-tie) from them to that person (alter) of the type specified in the prompt of the name generator question. Follow-up questions can then gather information about the alter, such as age, gender, and so on, factual information about the nature of the ego-alter relationship, family, friend, colleague, how and how often they meet, origins of the relation and so on and the respondent’s evaluation of that relationship, do they see this alter as a confidant, a friend or just an acquaintance.
This data collection yields first order network data. It can be analysed in its original format without using UCINET or any other network software (see later section…). In this exercise we will convert and save it into the tie data/ node data (VNA) format introduced in the previous section…. The conversion will allow us to explore the potential of UCINET/NetDraw network diagrams but then allows for the collection of alter-alter relations to create a 1.5 egnoet dataset.
The following document is the paper version of the questionnaire devised for this exercise. You can use the ???
- Name generator question: I use a single question: People important to you now? (from Spencer and Pahl, 2006; Rethinking Friendship).
- Ego included here to facilitate handling of node data common to ego and alters.
- Household ID. This keeps track of people in the same households. Each household has an ID (initial and number). This creates membership (or group) data, data with each person assigned to a collective group. (SNA designates this as ‘2-mode data’.)
- Relation to self: This left as open-ended (you can make up the description that best suits. Double (or more) options are possible e.g. colleague and close friend, neighbor and friend….
- Three questions eliciting respondent’s evaluation of the relationship including a simple assessment of tie strength.
Fill in your information in one of available spreadsheets –
Google doc [access]
- Egonet name generator question prompts free recall of a respondent’s personal network contacts
- Follow-up questions can elicit information about personal attributes of the alter, the type of relationship and factual information about it, and the respondent’s assessment of the tie strength.
- It is possible to extract useful social network information from first order egonet data in its original format. We can see how many contacts a respondent has, their demographic variety, the range and type of relations and the respondent’s evaluations of each relation. (See later section…)
- Converting the data to node data/ tie data lists allows us to draw network diagrams and collect alter-alter ties as additional network data .
Discussion: How many names on name generator lists.
Discussion: Why is number of names an important research question?
Discussion: What is the number of names given by NCCS respondents?
Next task: Save the data in node data/ tie data format.
2.3.1 Task Two: Construct tie data and node data lists (VNA edgelist1 format)
Compile your original data into a spreadsheet. Label that worksheet as ‘Raw data’ then save the full spreadsheet as MyNames.
Create two extra worksheets; ‘Node data’ and ‘Tie data’.
- Original names and character data
- Next worksheet labelled ‘Node data’
- Third worksheet labelled ‘Tie data’
Now create the node IDs.
- Remember that node IDs are the key fields that link node and tie data lists. I recommend you create them in the raw data worksheet and always cut and paste from there.
- Eliminate any white spaces and special characters from the names (e.g. Malcolm A should be MalcolmA or Malcolm_A). Do this also for the column headers (e.g. “Relation to you” becomes Reln_to_ego.
- I recommend that the IDs retain the original order of names.
- Procedure: Create a list of fore-tags (p01, p02… ) and attach them to each name. Any sorting will then retain your ordering of names.
- List of names checked for white spaces. (Same also for column headers).
- Column inserted and populated with alphanumeric fore-tags. Note that leading zeros are crucial to retain the sorting order. You can use fill down (as series) to populate this column.
- Insert a new column, ID, and use formula to join (concatenate – &) the fore-tag (col A), a delimiter (underscore character “_”), and the alter name (col B).
- Once you hit enter use Fill Down to populate the ID column (not shown).
- To change the IDs to embedded characters (they are created as formulas) you need to copy the column of IDs then use Paste Special, Paste as values (not shown).
Paste the data into the node data worksheet and clean it.
Software programs all have special characters that you cannot include in your input data. White spaces are the most problematic for NetDraw node data. You should clean any data where these may occur.
- Select a column of data which may contain white spaces or other special characters. (Include the column header if you have not checked it already.)
- Go to Find and Replace.
- In ‘Find’, enter a white space by hitting the space bar. (You can also use the Microsoft identifier ^w).
- In ‘Replace’ enter a underscore _, another neutral character or nothing.
- Click “Replace All” (and close the dialog box).
- You will need to eyeball your data for other special characters. I have an apostrophe (Daughter’s Partner) in my original list that also has to be deleted.
Create a tie data list.
- Copy and paste the node data into the tie data worksheet at the second column (col B). Change the column header of ID to TO. Then type in FROM as the first column header. (Case is not critical but I use upper case as these are required words.)
- Delete all the columns of the node data except the final one – TieStrength. Note: The columns of Fun and Serious (Y/N) data are really tie data. (They tell us, the outsider observer, about the tie between the ego and alter, as reported by the ego.) The VNA format only accepts numeric tie data (‘strength’) however.
- Copy and paste the ID of the respondent into the first cell of FROM data.
- Use Fill Down (Editing menu, bottom right cell corner or Ctrl-D) to populate the FROM column.
- Give the reflexive tie (first data row) TieStrength 0.
Your tie data list is now complete.
Discussion: Why would you want to use use the extra tie data information?
- The node IDs are the key field linking node and tie data. Once you have created the node IDs allows copy and paste.
- The TO column of the (edgelist1) tie data are the node IDs of the ego and all alters. The FROM column is the ego ID repeated for each row.
- VNA format accepts only numeric tie data attributes (TieStrength). However node data can include character data (without white spaces). Note that I have left TieStrength as a node data attribute as well as atie data attribute.
Next task: Read the data into UCINET/NetDraw