Chapter 2 Literature Review
3.5 Data collection methods available
3.5.1 Questionnaire survey
The questionnaire survey is the most commonly used method o f data collection in the social sciences in general and accounting research in particular. It is the most popular method used in previous research on IS security (Abu-Musa 2004a and b; Chang and Ho 2006; Davis 1997; Henry 1997; Hitchings 1995; Hong et al. 2006; Huang et al.
2006; Kankanhalli et al. 2003; Kotulic and Clark 2004; Loch et al. 1992; Ryan and Bordoloi 1997; W hitman 2004; Yeh and Chang 2007).
' Hussey and Hussey (1997, p. 161) stated that “a questionnaire is a list o f carefully structured questions; chosen after considerable testing, with a view to eliciting
reliable responses from a chosen sample”. Questionnaires are associated with both positivistic and phenomenological methodologies. Under a positivistic approach, closed questions should be used, whereas a phenomenological approach suggests open-ended questions.
There are a number o f different ways in which questionnaires can be administered.
They can be either self-administered questionnaires or interviewer-administered questionnaires as shown in Figure 3.2 below.
Figure 3.2 Types o f questionnaires
Questionnaire
Self-Administered
Postal Questionnaire
Delivery
& Collection Questionnaire
i
Internet-Mediated Questionnaire
Interviewer-Administered
Telephone Questionnaire
Structured Interview
(Based on Saunders et al. 2007, p.357)
The self-administered questionnaires are completed by respondents. They can be sent by post to the intended respondents, who are then expected to complete and return them by post (postal or mail questionnaire). They can be delivered by hand to each respondent and collected later (delivery and collection questionnaire). They also can be administered electronically using the internet (internet-mediated questionnaire).
On the other hand, responses to interviewer-administered questionnaires are recorded by the interviewer based on each respondent’s answer. They can be administered using telephone (telephone questionnaire) or face-to-face (structured interview) (Blaxter et al. 2001; Saunders' et a l 2007).
The choice between questionnaire methods depends on the nature o f the survey, the sample, time and cost constraints, the importance o f response rates and the types o f
‘ questions (De Vaus 2002).
Questionnaires, as with any other method, have both strengths and weaknesses, and hence advantages and limitations that the researcher must be aware of. Questionnaires are less expensive and less time-consuming than other methods, particularly when responses from a large, dispersed population are required (Bums 2000; Hussey and Hussey 1997; Sarantakos 2005). They can be completed at the respondents’
convenience. Thus, there is a better chance that respondents will take time to think about their answers, to check personal records and to consult other sources if necessary (Frankfort-Nachmias and Nachmias 1992; Neuman 2000; Zikmund 2000).
They offer less opportunity for interviewer bias because respondents are not influenced by interviewer characteristics or techniques (Frankfort-Nachmias and Nachmias 1992; Oppenheim 1992). They offer greater assurance o f anonymity o f respondents, which is important, particularly when sensitive issues are involved (Frankfort-Nachmias and Nachmias 1992; Neuman 2000). Moreover, Bums (2000) argued that a questionnaire that can guarantee confidentiality might elicit more truthful responses than would be obtained from a personal interview. In addition, errors resulting from the recording o f responses by interviewers are reduced. They are a stable, consistent, uniform measure and free from variation (Sarantakos 2005).
However, there are a num ber o f problems associated with the use o f questionnaires.
Questionnaires require simple, easily understood questions and instructions. They do not allow the opportunity to correct misunderstandings, to probe for additional information, or to offer explanations or help (Frankfort-Nachmias and Nachmias 1992; Oppenheim 1992). They do not provide opportunities for motivating the respondents to participate in the survey or to answer the questions (Sarantakos 2005).
The researcher has no control over the conditions in which the questionnaire is completed. Researchers are not sure whether the right person has answered the questions and whether the order o f the questions - where required - was followed (Neuman 2000; Wilson 1996). Incomplete questionnaires cannot be followed up (Bums 2000; Oppenheim 1992).
Moreover, researchers cannot visually observe the respondents’ reactions to
* questions, physical characteristics or the setting (Neuman 2000). Respondents may be limited from providing free expression o f opinions particularly in the case o f
closed-ended questions (Bums 2000). Finally, and perhaps the most serious problem, is the low response rate, particularly with mail or postal questionnaires (Bums 2000;
Frankfort-Nachmias and Nachmias 1992; Neuman 2000; Oppenheim 1992;
Sarantakos 2003). Hussey and Hussey (1997) indicated that response rates o f 10 percent or less are common and this introduces the problem o f sample biases because those who respond may have a particular interest in the topic and, therefore, may not at all represent the population from which they are drawn.
Despite the above disadvantages o f questionnaires, every effort was made to minimise these limitations by pre-testing them prior to distribution, which is presented in more detail in Section 3.6.1.2.
However, some effective techniques can be employed for improving the response rate. An attractive questionnaire design and question wording both help to assure a good response rate (Zikmund 2000). The cover letter must succeed in convincing the respondents to complete the questionnaire and mail it back. It should therefore identify the sponsor o f the study, explain its purpose, the method o f sampling used and how the respondents are chosen, and should tell the respondents the importance o f completing the questionnaire (Frankfort-Nachmias and Nachmias 1992;
Oppenheim 1992). Questionnaires could be accompanied by a self-addressed, freepost return envelop (Neuman 2000; Sarantakos 2005). In addition, the respondents’ motivation for returning a questionnaire may be increased - in certain cases - by offering m onetary incentives or by offering them the chance to win a major prize if the questionnaire is returned (Oppenheim 1992; Zikmund 2000).
Follow-up reminder letters could be sent to those who have not returned the questionnaire (Neuman 2000; Zikmund 2000). Finally, and most importantly, is the length o f the questionnaire. Hussey and Hussey (1997) argued that response rates could be increased by keeping the questionnaire as short as possible.
3.5.2 Interviews
Interviewing is one o f the most common methods o f data collection in social research.
Interviews have become increasingly utilised in recent years in IS research in general
“and AIS security in particular, either as the sole data collection method (Keller et al.
2005) or combined with other methods (Mitchell et al. 1999; Straub and Welke 1998;
Tryfonas et al. 2001). An interview is a conversation or purposeful discussion between two or more people (Robson 1993, p.228). It is a method o f collecting data in which selected participants are asked questions in order to find out what they do, think or feel (Hussey and Hussey 1997, p. 156).
Figure 3.3 Types o f interviews
Interviews
Standardised
Interviewer-Administered
1
Face-To-Face Interviews
(Based on Saunders et al. 2007, p .3 13)
One-To-One
Non-Standardised
One-To-Many
i i
Telephone Internet Group Interviews Mediated Interviews
Interviews (focus groups)
1
Internet Mediated Interviews
(focus groups)
Interviewing is employed as a data collection method in most research designs regardless o f the underlying methodology (Sarantakos 2005). Interviews are associated with both positivistic and phenomenological methodologies. Under a positivistic approach they are highly formalised and structured, whereas a phenomenological approach suggests informal and unstructured conversations. In between, there are intermediate positions (Hussey and Hussey 1997; Saunders et al.
2007). Interviews could be conducted either face-to-face or by telephone or online, and these forms o f interview are summarised in Figure 3.3.
Broadly speaking, there are three types o f interviews used in social research:
structured interviews, semi-structured interviews, and unstructured interviews.
Structured interviews: These are the least flexible type o f interviews. Structured interviews employ a structured questionnaire, which is verbally presented to
respondents, with the answers recorded in the questionnaire by the interviewer (Frankfort-Nachmias and Nachmias 1992; Sarantakos 2005). Every respondent is asked the same questions with the same wording and in the same order (Wilson 1996). There is very little flexibility in the way questions are asked or answered in the structured interview setting (Fontana and Frey 2000). The interviewer is expected to act in a neutral manner, keeping the same tone o f voice across the interviews, offering a consistent impression to the respondents, using the same style, appearance, prompts, probes, etc. (Sarantakos 2005). Thus, the highest degree o f objectivity and uniformity in procedures can be achieved and the opportunities for interviewer bias can be restricted (Smith 2003). All or nearly all o f the questions are closed-ended and the respondents are forced to select their answers from a limited set o f previously established responses (Bum s 2000). The use o f closed-ended questions makes coding o f answers easier and has advantages for the subsequent analysis. Closed-ended questions also eliminate the opportunities for error associated with open-ended questions. Flowever, they sacrifice the comparative advantage o f the interview method by failing to include the flexibility and richness o f responses offered by open- ended questions (Smith 2003). In addition, this detachment and impersonal approach can prevent trust and rapport building up between the interviewer and respondents (Bums 2000). This form o f interview can be employed in quantitative research.
Semi-structured interviews: In semi-structured interviews, the interviewer has a list o f questions on specific topics to be covered, often referred to as an interview guide, but he has greater freedom in the sequencing o f questions, in the exact wording and in the amount o f time and attention given to different topics (Bryman and Bell 2003;
Robson 1993). Some questions can be omitted in particular interviews and additional questions can be asked, as the interviewer sees fit, to examine associated issues that arise in the interview (Saunders et al. 2007; Smith 2003). This form o f interview allows respondents to answer more on their own terms than the standardised interview permits. The interviewer can seek both clarification and elaboration on the answers given, can have more latitude to probe beyond the answers and thus can enter into a dialogue with the interviewee (May 1997). This permits greater flexibility than the structured interview; however, the comparability o f the information between respondents is difficult to assess and response-coding difficulties may arise (Bums
2000). This form o f interview can be employed in quantitative or qualitative research (Sarantakos 2005).
Unstructured interviews: These are the most flexible type o f interviews. They take the form o f a conversation between the interviewer and the interviewee (Bums 2000).
The interviewer does not enter the interview setting with a planned sequence o f questions to be asked, but with a series o f topics for discussion (Sekaran 2003; Smith 2003). The interviewee is then allowed to talk freely about events, behaviour and beliefs in relation to the topic area, so that this type o f interaction is sometimes called non-directive (Saunders et al. 2007). The interviewer has a great deal o f freedom to probe various areas and to raise specific queries during the interview (Frankfort- Nachmias and Nachm ias 1992). The actual words and phrases used may therefore vary significantly between interviews (Smith 2003). This type o f interview is a powerful research tool and can provide a greater breadth o f data than other methods (Fontana and Frey 2000; Punch 1998). However, it demands a sensitive, skilled and cautious interviewer in order to understand the other person’s views and, at times, to assist individuals to explore their own beliefs (Easterby-Smith et al. 1991). The unstmctured interviews can take a long time, they can be difficult to interpret and analyse and the coding is a difficult task in spite o f improved coding techniques and systems (Ghauri and Gronhaug 2002). This form o f interview is mostly used in qualitative research.
Based on the above, it appears that these three types o f interviews have different strengths and weaknesses and different purposes in research. The type o f interview selected should be aligned with the strategy, purposes, and research questions.
Accordingly, given the research objectives, as well as the resources and time constraints, semi-structured interviews were adopted in the present study and full details o f the interview findings are presented in Chapter 5.
However, the researcher must be aware o f the strengths and limitations o f the interviews - despite their type - as a data collection method, in order to benefit from the strengths, minimise the limitations and maintain the quality o f research.
Interviews allow greater flexibility in the questioning process and can be adjusted to meet many diverse situations (Frankfort-Nachmias and Nachmias 1992). They result in a high response rate that makes the data more representative than data solicited through a postal questionnaire (Bums 2000). They give the interviewer the opportunity to observe non-verbal cues that help in understanding the verbal response, that can change or even, in extreme cases, reverse its meaning (Robson 1993). The identity o f the interviewee is known, and the interviewer has an opportunity to control the conditions under which the questions are answered (Sarantakos 2005). In addition, greater length is possible in interviewing than when other methods are used.
On the other hand, interviews are more expensive and time-consuming than other methods such as questionnaires (Zikmund 2000). They offer less anonymity since the interviewer knows the identity o f the respondent (Sarantakos 2005). They are less convenient than other methods. The flexibility o f interviews leaves room for the interviewer’s personal influence and bias (Frankfort-Nachmias and Nachmias 1992), and may generate difficulties in categorising and evaluating responses (Bums 2000).