CHAPTER 2. TWO MECHANISMS 59
It gives us the tools to address the questions that Hacker identified as necessary for explaining institutional change: “what kinds of institu-tional changes [are] propelled by what kinds of social processes [and]
are most likely under what kinds of political configurations” (Hacker et al. 2015: 180). The mechanism of deinstitutionalisation and its three sub-mechanisms of disruption, conversion and displacement are driven by specific party behaviours which this dissertation believes are cap-tured in the three concepts of: issue-ownership, strategic adjustment and policy-reversal.
CHAPTER 2. TWO MECHANISMS 60
documents, elite interviews, diaries, e-mail correspondence, comments and articles in media by policymakers and senior oﬃcers, formal reports by the government, parliament, governmental and parliamentarian com-mittees, reports and memos by the military Headquarters as well as earlier research. The oﬃcial reports and memos from government and the military have been used to trace the chronology of the decision-making process and the major political events and developments. Com-ments and appearances in the media by key actors have been used to get an idea of the social and political climate as the decision-making process proceeded. Archival research has been limited to the military’s own analyses of key decisions, as well as to the parliamentarian com-mittees that have examined the development of Sweden’s recruitment models.
An important part of the sources in this dissertation is semi-structured elite interviews. The interviews have almost exclusively been with senior decision makers including the defence ministers (with one exception), the supreme commanders, director generals of the National Recruitment Agency, members of parliament and committees. The interviews have been semistructured, recorded and transcribed. In every interview, the respondent has later read and approved the transcript. The point with the interviews has mainly been to cover gaps in the policy process by ob-taining information that could not be gathered in any other way. They have played an important part for retelling the story, and have added a sense of social context, but have not been instrumental in accounting for the outcome. If anything, they have added confidence to conclu-sions that could, theoretically but with less confidence, otherwise also be drawn from the other sources of information.
Two challenges, validity and reliability, should be mentioned when it comes to interviewing. Validity points to the possible problem on whether the interview, with the information coming from the respond-ent, actually gauges the properties it is supposed to measure. Validity can be oﬀset by asking the wrong questions or if the respondent an-swers in deceiving ways. This dissertation has tried to overcome this obstacle by gaining as much knowledge as possible about the topic for the interview beforehand (reading research by others, looking into me-dia appearances from the time, and using the information from other interviews). This has enabled a careful and informed use of the inter-view material. Another and maybe more important approach is that the interviews have focused on “what-” questions rather than “why-”
questions. That is, asking what happened, (focusing on behaviours) at a specific juncture, instead of asking why something happened (probing for explanations) (Mosley 2013: 21). The why-questions are ultimately up for the combined analysis to come down to, even though it must
CHAPTER 2. TWO MECHANISMS 61
be mentioned that the respondent’s “why” view is a valuable piece of information for getting an idea of his view at things. Moreover, the respondents are asked to remember things that happened between one and two decades ago and may choose either to revise their accounts to their favour or, less sinfully, for remembering things in the wrong way. That is, can we be confident that the information gathered from the interviews are accurate? First oﬀ, most of the respondents had few problems of retelling their accounts even in detailed ways. Maybe be-cause most of the respondents were at the top of their careers and the subject was an important one in their portfolio. This speaks in favour of their ability to remember things correctly. Moreover, in using the inter-view material the dissertation has made use of triangulation with other sources of information to guard against potential reliability problems.
Cross-checking and triangulating with other sources of information (e.g.
journalistic accounts and earlier research) is widely believed to be an important and useful way of guarding against reliability problems (King et al. 1994).
Triangulating has been especially important in certain parts of the policy process, to make sure that the account given is balanced and correct. Such parts especially include the years between 1998 and 2001 and the years between 2007 and 2009. In these parts I have as far as possible tried to cross check the data with several sources and kinds of information, including interview accounts from diﬀerent persons, me-dia, oﬃcial documents and, if possible, secondary literature. To take one example, the conflict between the Headquarters and the Defence Ministry on the organisational purpose of the Armed Forces has been corroborated by interview accounts of the three main involved actors, media, oﬃcial documents and secondary literature. The work of the 2007 Committee on Military Service was similarly cross checked with diﬀerent sources of information. When sensitive information has been provided about on a certain situation, for instance in an interview, but that information could later not be corroborated in other sources I have opted for rejecting the information. While this sometimes can mean that an important piece of information is excluded, it however ensures that the argument given in this dissertation is as valid as empirically possible given the available data.
That being said about validity, there is also a reliability problem that points in the other direction. Namely whether the researcher used interview material selectively to support the argument. This is espe-cially important to highlight since thus dissertation makes extensive use of quotations from interviews conducted recently, but the period under study occurred several years ago. On a general note I have as far as pos-sible extracted the main points of information given by each interview
CHAPTER 2. TWO MECHANISMS 62
subject. This means that I have extracted quotations that capture the general picture provided to me by the person being interviewed. What is more, the arguments presented in this dissertation have only been formed after the interviews have been conducted. That is to say, the theoretical arguments have not been developed before the interviews.
I have first gathered the empirics, and then developed my theoretical account. This I believe minimises the risk of using the data selectively, even unintentionally. In addition, when there is a quote the date for that is explicitly stated — which indicates to the reader that it is tem-porally separated from the other text. The tense also changes in these quotes, from present to past tense. All transcriptions from the inter-views have been read and approved by the interview subjects. Finally, in general, the interviews have oﬀered few surprises when it concerns the big picture of the account given in this dissertation. The accounts oﬀered by the respondents have most of the time not departed from the ones that transpire in comments to media and other sources of publicly available information.10