• No results found

2 Research Context

5.6 Wave 2: In Depth Interviews

5.6.3 Interview Design

5.6.3.1 Settings and Protocol

Participants were contacted by email which they had provided during the online survey to set up a convenient time for an interview. Each interview lasted anywhere between an hour and a

half to two hours. The participants were advised to bring along their student cards for identification. All interviews were conducted at DIT Aungier Street campus in pre booked rooms and audio recorded in Audacity® (Mazzoni and Dannenberg, 2011), a software application that allows digital recording of sound. The basic format and sequence for the interviews is presented below.

1. The overarching aims of the research and related ethical concerns (section 5.7) were addressed in a 10 minute discussion preceding the actual interview.

2. The name generators were administered as will be explained in section 5.6.3.2.

3. Permission was sought to audio record the interview and an opening cue prompted the participants to talk about the people that they named in response to the name generator questions as will be explained in section 5.6.5.

4. Towards the end of the interview, the participants were required to fill up the adjacency matrix (section 5.6.5.1) and an alter attribute chart (section 5.6.3.3). Finally, a short activity comprising a concentric circles diagram as will be described in section 5.6.5.2 concluded each interview.

5. The participants were offered a cash incentive of € 15 each for their participation in the interview.

5.6.3.2 Name Generators

The technique for gathering data on personal networks generally consists of using a name generator, which is a tool that uses a question(s) to produce names of people who share a particular relation with the participant (Carpentier and Ducharme, 2007). Name generators can

emphasize different aspects of a network (e.g. friendships, resource flows, information flows, exchanges of social support), but the social-support perspective is the one most often used. Researchers describe social support as an interactive process through which emotional, instrumental, or financial assistance is obtained from one’s social network (Wasserman and Faust, 2007). Depending on the name generator and the selection criteria utilized (e.g., people one takes advice from or people one plays sports with), the network will assume a different size, form, and content. Since it is not possible for a researcher to study all contacts in an individual’s social surroundings, the name generator technique offers a way to extract “a fraction of respondents’ social contacts” who are relevant to the research question (Marsden, 2005).

The key considerations then are to decide on the number of alters to be elicited, the use of one versus multiple name generators and the choice of a criteria in generating alter names.

McCarty and colleagues (2007) suggest that free recall of 25 alters will capture the same structural pattern as a network of 45 alters. For most ego centric studies where the focus in on eliciting strong ties, free recall questions that place no definite upper limit on the overall network size have been found to be generally useful (Ferligoj and Hlebec, 1999; Carrasco et al., 2008).

While administrating a single name generator has the benefits of saving time and reducing respondent burden, researchers examining the quality of network data argue that the use of multiple name generators where possible is the preferred method of choice in most cases (Marin and Hampton, 2007; Bidart and Charbonneau, 2011). This is because the use of a single name generator induces people to remember the most obvious people whereas it may be

relevant to prod memory into searching for significant persons not easily remembered (Bidart and Charbonneau, 2011). For example, an individual may not be able to recall relevant ties because they live far way or because the individual hasn’t seen them for a long time.

The choice of criteria in a name generator is generally based on the nature of study. For example two classic studies on ego networks use name generators identifying people with whom the focal actor ‘‘discussed important matters with in the past 6 months’’ (Burt, 1984) or with whom he or she ‘‘frequently socialized with” (Fischer, 1982). Many others follow suit. On the whole, research indicates that people are generally better at recalling typical or routine relationships and interactions than they are on transactions that occur within highly specific time frames (Marsden, 1990).

In light of the above discussion, the following name generators were administered in this study. 1. From time to time people discuss important matters with other people. Who are the people with whom you discuss matters important to you? Give me their first names and last initial.

2. Who are the people you really enjoy partying/socializing with? Give me their first names and last initial.

3. Please list anyone who is especially close to you, who you have not listed in one of the previous questions. Give me their first names and last initial.

Naturally the outcome of such questions is dependent upon the habits of sociability in a particular culture (Ruan, 1998; Bidart and Charbonneau, 2011). Given the magnitude and

significance of drinking in the social life of Ireland, it was expected that these name generators will elicit significant alters relevant to the phenomenon of interest i.e. drinking behaviour. 5.6.3.3 Name Interpreters

Name generators are generally followed by a series of ‘name interpreter’ questions designed to elicit information about the network members such as their attributes, properties of the relationship they share with the ego and their ties with other named alters (Wasserman and Faust, 2007). One of the key considerations in designing and administrating name interpreter questions is to balance respondent burden and the information sought (Marsden, 2005). An important decision that needed to be made in this regards relates to whether the number of alters examined in the name interpreter questions be limited or not? The General Social Survey (Burt, 1984) and McCallister and Fischer (1978) elicited alter data for only the first 5 and 8 people named respectively as a concession to time constraints. Burt’s reasoning for this is quoted:

“Somewhere between the time saving choice of asking too few names and the unacceptably time consuming choice of asking too many names, is that number of alter names which reaches the border of the respondent’s interpersonal environment to reveal social heterogeneity. We do not know where this point is …With no guide other than common sense, I expect the corresponding upper limit to be a single digit number. There is evidence to suggest an upper limit of five to seven (e.g. Miller (1956) and Simon (1974)). In the interest of increasing measurement precision and decreasing measurement bias - under a severe time constraint - the five alter limit proposed for seems judicious.” (p 315)

In context of the current research, it was deemed adequate to collect information on all alters for two reasons. First, the studies mentioned above, were questionnaire based. Guided by methods adopted in the past research (Bellotti, 2008), it was anticipated that the conversational nature of the in depth interviews and the opportunity to ask for clarifications will make the process less burdensome for the participants. Second, the Irish love of drinking and conversation is, of course, proverbial and has earned them a reputation as one of the most talkative and entertaining folk (Haining, 2003). The pilot interviews revealed that having an hour or two of conversation related to one’s social circle and drinking was not by any means boring for the participants. Further, it was the beginning of the semester; students were generally relaxed with no impending projects or class assignments that required immediate attention.

In order to maintain the natural flow of conversation and avoid unnecessary disruption, the name interpreter items related to basic attributes of the alters (age, gender, occupation, city of residence, relation and time known) were presented to participants in the form of a chart at the end of the interview. Most participants were able to complete the chart fairly quickly.