The methodology of this research will follow a positivism philosophy in an effort to collect data from an observable reality and find similarities in the data collected to make them generalizable to other researchers theories (Remenyi et al. 1998, cited in Saunders, 2009, p.144). Gill and Johnson (2002, pp.39-40) add that the use of positivism emphasizes on a “cause and effect” relationship between variables, and the possible identification of these variables. The rationale of this philosophy is the intention to create a prospect decree (Fisher, 2010, p.32) of Les Roches students likelihood to display violent behavior when exposed to certain variables/factors.
This is equally consistent with the research philosophy performed by Homel and Clark (1994, p.1) where they aimed to explain which situational and managerial factors were most projecting of violent behavior. Furthermore Saunders, Lewis and Thornhill (2012, p.135) state that through the use of a positivism philosophy, empirical research will be employed in order to systematically identify which factors predominantly affect Les Roches students tendency to behave violently by providing quantitative measurements of “statistical analysis”. Nevertheless, much critique exists about positivism philosophy concerning its lack of “cultural sensitivity and historical specificity” (Calhoun,1992, cited in Mouton, 1996, pp. 200-201), as well as its lack of development in knowlege considering the basis of the research is on previously established theories (Popper,1972, cited in Henn, Weinstein, & Foard, 2006, p.11). Most importantly, positivism excludes proper interpreation and connotation of quantitative data findings, leading to an unrealistic generalisability (Swanson & Holton III, 2005, p. 18). Even so, Interpretivsm was not choose for this topic because the Research Question does not aim to describe nonrepresentational significance through observations, but rather “uncover the truth and facts of an issue as quantitatively specified relations among variables” (Swanson & Holton III, 2005,p.20).
Approach, Design and Time Frame
In accordance with the “theory-then-research method” (Henn, Weinstein, & Foard, 2006, p.13), this research opted with a deducive approach in order to confirm and make predictions based on previously exemplified theories (Tharenou, Donohue, & Cooper, 2007, p.14), pertaning to factors influencing violent behaviour. Swanson and Holton III (2005, p.164) explain that deductivism bases itself on a scale development that requires it to use a classification scheme before data is collected. This approach is consistent in many researches exploring this issue, for instance, in Hughes et al.’s (2007,p.1) research, background information is already known, for this reason they decided upon a deductive approach in order to “explore variation in alcohol use and negative nightlife experiences between young individuals who drank prior to going out, versus those who do not drink until reaching bars and nightclubs”. An advantage of choosing a deductive approach, is that, if used adequately, it will aid in the guaranteeing of content validity in the final result (Swanson & Holton III, 2005, p. 165). Interestingly tough, Liang (1967, cited in Jill & Johnson, 2002,p.41) draws attention to the concern that because it relies on scientific analysis and testing (Saunders, Lewis, & Thornhill, 2012, pp. 124-125), this will not consider the natural science of the human behaviour in the final analysis of data, which can result in incorrect generalizations and/or oversimplifying the final answer, also caused by one’s inclination to suppose the initial findings as the only plausible answer based on prior research rather seeking to falsify it (Popper, 2002, cited in Fisher, Buglear, Lowry, Mutch, & Tansley, 2010,p.65). This study attempts to ascertain the relationship that independent variables (consumption, environment, management practise and cultural heritage) have on the production of a dependent variable (violent behaviour), for this reason an explanaory research design was choosen (Saunders, Lewis, & Thornhill, 2012, p.172). As a result of the analysis of primary data, the independed variables were already known, for this reason the exploratory and descriptive research design methods were not employed (Swanson & Holton III, 2005, p.117). Through the use of cross-sectional time framing, the research employed a data collection snapshot strategy, which implies it beeing at a specific time (Swanson & Holton III, 2005, p.80), due to the time sensitive nature of the study, findings (Saunders, Lewis, & Thornhill , 2012, p.190). Moreover, according to Jupp (2006, p.31), the use of cross-sectional time frame is both “cheaper and easier to organize”; thus, considering the study was time constrained for its completition, longitudinal time framing was not selected as it can take from a few years or months to decades to complete (Jackson, 2009, pg.327).
Unfortunately cross-sectional studies are considered to be feeble on “internal validity” (Tharenou, Donohue, & Cooper, 2007, p.20).
The preferred strategy chosen to collect the data needed in order to answer the research question was through the use of surveys. Fisher (2010, p.69) reminds that surveys allow for easier analysis of relationships between subjects thus establishing of possible strong or weak associations or correlations between these variables, which is the case in this study.
Experiment’s and survey’s are predominantly, or completely, considered the main two strategies utilized when dealing with quantitative data research design (Saunders, Lewis, & Thornhill, 2012, p. 173), for this reason archivial research, case studies, ethnogrpahy, action research, grounded theory and narrative inquiry were not selected for this study. The reason behind choosing survey strategy, over the others, is due to its effortlessness in the studying large groups of individuals, although the concern in using this strategy lies with its reliability on whether the individuals in the sample group are representative of the entire population related to the study, resulting on a potential incorrect generalization (Jackson, 2009, p.16). Even so, Fisher (2010, p.69) states that when inquiring on finding a broad and representative overview of a situation the suggested strategy are surveys. In addition, the use of surveys is broadly used in many studies on this topic, such as the research conducted by Donath et al. (2011, p.1), where the use of survey’s was employed throughout European High Schools in order to compare how “different migration backgrounds and rural vs. urban residence affect alcohol consumption and binge drinking in adolescents”.
Data Collection and Sampling
Secondary data was collected through online scholar databases, the selection of data was categorized by the relevance to the present study. Much research focused on similar topics, such as consumption (Hughes et al.,2007;Homel & Clark, 1994), environment (Livingston,2008; Macintyre & Homel,1997), management practices (Romain-Glassey et al., 2012; Roberts, 2007) and cultural heritage (McNulty and Bellair,2003; Donath, et al., 2011). The use of this secondary data enabled the creation and formatting of reliable and concise surveys that aided in the production of theoretical framework for this research (Saunders, Lewis, & Thornhill, 2012, pp.304-307), in addition to supplementing (Jill & Johnson, 2002, p.151) and supporting in the final comparing of the final findings (Bryman & Bell, 2003, p.212). Ghauri and GrÃ¸nhaug (2005, cited in Saunders, Lewis, & Thornhill, 2009, p.268) highlight the benefits of collecting secondary data by emphasizing on its relative aid in saving both time and money. Even so, one of the main concerns of making use of secondary data is whether the definitions used in previous research are accurate or somewhat interpreted according to the researchers own judgment thus not providing for accurate differencing of the analysis of data (Saunders, Lewis, & Thornhill, 2009, p.271), especially in the case of qualitative data (Jupp, 2006, p.275).
In this study, the primary data collection method employed was a mixture between delivery and collection questionnaire and web-based questionnaire, in order to facilitate the examinatining and storing of data (Swanson & Holton III, 2005, p.108). Most of the questionnaires were filled out on printed paper copies, randomly distributed to Les Roches Students that fell into the sample size chosen for this study. Meanwhile the local authorities and venue owner were requested by phone if they would agree to participate in the questionnaire and after their approval, the data was collected through structured interviews in accordance to the original questionnaire distributed to the Les Roches students. Waters (2011, p.85) underlies the importance of conducting primary data collection, considering its appropriateness to the study as it is both current and reliable. Nevertheless, it must be acknowledged that the collecting of primary can be costly and mostly time consuming (Saunders, Lewis, & Thornhill, 2012, p. 306).
All the same, questionnaires were implemented because of their competence (Bryman, 2004, cited in Tharenou, Donohue, & Cooper, 2007, p. 102) and commonness in many studies involving “attitude, values, preferences, intentions and personalities” (Moorman & Podsakoff,1992 cited in Tharenou, Donohue, & Cooper, 2007, p. 102). The questionnaires were generated using an online web-based software in order to create clear and concise questions that ensured reliability and internal validity (Saunders, Lewis, & Thornhill, 2012, p. 428).
The questionnaire itself was presented on four pages as to avoid the respondents to become uninterested, considering research shows that questionnaires that are too long are often discarded and have a reduced response rate (Edwards et al. 2002 cited in Saunders, Lewis, & Thornhill, 2012, pp. 445-446). The Les Roches students questionnaire was divided into three sections: The first section inquired on the respondents demographics, the second section measured, through a five-point likert scale, the percieved influence allocated to certain violent-causing factors and effectiveness of prevention methods. Finally the third section offered multiple choice answers that inquired on the students current “party” habits, .
According to Fisher et al., (2010, p.227) likert scales are frequently used when respondents opinions and attitudes are required. Also according to Cook et al. (1981, cited in Swanson & Holton III, 2005, p.166), likert type scales are the most regularly used scaling techniques throughout this type of research. Moreover, multiple choice questions usually provide respondents with two to five options to answer a question, and it is required for them to choose one (Fisher, Buglear, Lowry, Mutch, & Tansley, 2010, p. 212). Henn, Weinstein, & Foard (2006, p.139) explain that in questionnaires, there should be a variation with the type of questions inquired in order to avoid the results beeing either skeewed or biased.
The reason for inquiring in these topics is because each information inquired in the questionnaires has a direct importance on each independent variable (consumption, environmental, management concerns and cultural background) which ultimately influences the dependant variable, being violent behavior (Powell, 1998, pp.2-3).
Pertaining to section one of the questionnaire, demographics were inquired considering Hughes (2006, p.1), showed a positive relationship with the younger the age of individuals to a high number of violent episodes caused by binge drinking. In order for the author to obtain primary data to compare and contrast to previous literature, such as Hughes studies, demographic questions were inquired (see Error: Reference source not found: Error: Reference source not found, Error: Reference source not found, Error: Reference source not found). In section two of the questionnaire, respondents were required to range their opinion in regards to factors influencing violent behavior, this is in reference to previously critically reviewed studies conducted by various authors where consumption, environmental factors, management practices and cultural heritages where established leading influential factors (Homel and Clark, 1997; Macintyre and Homel, 1997; Shaw and McKay,1969). Additionally, judgment of the effectiveness of prevention methods were inquired, as different authors demonstrated the efficiency of these methods in reducing violent behavior; although emphasized that the results would vary according to the location and type of clientele (Miller et al. , 2011; Mazerolle et al., 2012; Rossow & Nordstrom, 2011). Finally, in the third section, respondents were asked for current habits when “going out”, in regards to the multitude of research demonstrating positive correlations between individuals that frequently go out and consume alcohol to high number of altercations (Homel & Clark, 1994; Romain-Glassey et al., 2012). Additionally the venue owner and authorities were inquired on whether the altercations witnessed in their venues were majorly caused by Les Roches students or by locals, in order to determine the key promoters of violent behavior (see Error: Reference source not found: Error: Reference source not found, Error: Reference source not found, Error: Reference source not found).
In due course the aim of this study is to identify which factors influence violent behavior in the case of Les Roches students, whether this involves an individual’s age, nationality or perhaps drinking habits, or something out of the ordinary.
In regards to the sampling technique utilized in this study, probability sampling with simple random system was the preferred method as a result of “stratifying and clustering the sample in order to increase the accuracy” (Henn, Weinstein, & Foard, 2006, p.129). Stratification of the selected sample was made upon students that had been “engaged” for at least a semester in the Les Roches campus, so first year students (BBA1) were not part of the study research as they were not experienced in “The Les Roches Way of Life”. The questionnaires were distributed to at least one student from each year, and were approached directly or indirectly through teachers that were individually asked to assist in the questionnaire distribution, as such the concept of stratification and clustering was enhanced.
Probability sampling offers “statistical foundation that represents the chosen sample drawn is representative of the entire population” (Swanson & Holton III, 2005, p.101). Additionally, Fink (1995, cited in Swanson & Holton III, 2005, pp.101-102) mentions that each individual involved in the survey “will have a know, nonzero probability of beeing included in the sample”. Furthermore, Saunders, Lewis, & Thornhill (2012, p. 262) point out, this technique is often connected with survey strategies. In most cases, this technique offers advantages over the non-probability technique: Tharenou, Donohue, & Cooper (2007, p.53) confirm that probability sampling is better because it strenghtens “the external validity of a study”. The disadvantage and difficulty in using this technique occours at the beginning, due the possibility of making systematic discrepancies about the research population and the sampling frame, possibly resulting in an inaccurate representation of the target population (Jill & Johnson, 2002, p.101). Moreover, Jupp (2006, p.197) mentions that it can take more time, thus making itself costlier to use. The reason for not selecting a non-probability technique, is because as Henn, Weinstein, & Foard (2006, p.129) mention, this technique is only applicable if all characteristics of the population are known.
In order to determine the accurate number required of students in order for it to qualify as representing the entire population, an arithmetic table was employed that took into consideration the total number of the population, acceptable margin of error and the confidence level (Jill & Johnson, 2002, p.102), which resulted that a total of 250 surveys were required.
Meanwhile, for the local authorities and venue owner, a non-probability technique of purposive sampling was employed due to direct need to obtain “a biased group for screening purposes” (Cooper & Schindler, 2003, cited in Tharenou, Donohue, & Cooper, 2007, p.56) to enforce the triangulation method of reliability. In addition, purposive sampling is characterized by being used in cases where the individuals in the sample group are directly related to the topic of issue (Swanson & Holton III, 2005, p.52). Neuman (2005 cited in Saunders, Lewis, & Thornhill, 2012, p. 287) highlites the use of this technique when the sample is relatevely small, which is evident in this study, and as well when the individuals in the sample are known to provide the right kind of information needed. But because this technique relies on individuals prejudice, there is an obvious downfall due to the desire of some members of the sample to “make a particular point” thus generalizing the findings to their standpoint (Swanson & Holton III, 2005, p.52)
The research procedure selected so far followed a deductive approach of quantitative data collection which emphasized on finding casual relationships. The research will subsequently endeavoured in the numerical and statistical analysis of data (Saunders, Lewis, & Thornhill, 2009, p.175), using the Statistical Package for the Social Sciences Version 21 (SPSS) system of statistical analysis. Moreover, it will employ simple linear regression analysis in order to investigate the relationship between of the dependent variable and several independent variables (Jupp, 2006, p. 65), to conclude with a testing of the model (see Error: Reference source not found), Although SPSS is highly efficient in statistical analysis (Ghauri & GrÃ¸nhaug, 2010, p. 164), it’s major drawback is that it takes time to find appropriate matching of variables and also loss of information that is not able to satisfy the matching criteria thus possibly losing statistical efficiency (Kleinbaum & Klein, 2010, p. 394).Even so, Lee & Lings (2008, p.321) reveal that it is the most popular quantitative analysis software used in research.
When analyzing the data, two types of statistics were used: descriptive statistics, which organize the data collected into tables and graphs representing percentages and values, which will help in the analyzing of the data (see Error: Reference source not found chapter), although will not present as much information as inferential statistics (Jackson, 2009, p. 109). Meanwhile, inferential statistics, that will deal with producing generalized predictions of an entire population (Jupp, 2006, p. 147), based on the analysis of a sample of the population from observations and analyses (Fisher, Buglear, Lowry, Mutch, & Tansley, 2010, p.239). Moreover, Marczyk, DeMatteo, & Festinger (2005, p.224) reveal that this method of data analysis allows researchers to “go beyond the parameters of the their study sample” and portray possible new conclusions for the entire population based on the sample observed. Even so, Mouton (1996, p.139) highlights an important notion to be aware of, it is imperative that the sample be representative of the population to which it is beeing generalized to. Accordingly, sufficient amount of respondents (N=250) participated in the questionnaire, in line with the sample calculation that considered a confidence of 95%, and margin of error of also 95%, the examined sample can be adequately generalized to the entire population.
When selecting inferential statistics there are different statistical tests that can be based on either parametic tests or nonparametic tests (Marczyk, DeMatteo, & Festinger, 2005, p.227). The parametic tests (z test, Pearson r and t test) require the researcher to make assumptions about the population based on the mean, standard deviation, and normal distribution, using interval (0-100) or ratio data (Jackson, 2009, pp.171-172). Meanwhile, the nonparametic test (chi-square, Spearman’s Rho and goodness-of-fit tests), do not involve parameters from the population, such as mean or standard deviation or normal distribution, and involve ordinal or nominal data (Lee & Lings, 2008, p.336). According to Tabachnick & Fidell (2007, p.11-12) in order for a correlation to be considered significant, it must have a Sig. value of less than or equal 0.05 ,
According to Pallant (2007, p.126) the value of Pearson r ranges from -1 to 1; the sign before the value indicates whether the correlation is negative (as one variable increases, the other decreases), or positive (as one variable increases, so too does the other). A perfect correlation would be either -1, or 1, meaning that a perfect predictability of one variable when the other is known (Tabachnick & Fidell, 2007, p. 56). In order to interpret the values between 0 and 1, Cohen (1988, pp. 79-81, cited in Pallant, 2007, p.132) proposed the following “rule of thumb”:
r = 0 – .29 weak relationship
r = .30 – .49 moderate relationship
r = .50 – 1.0 strong relationship
The data analysis adopted the use of a parametric testing of Pearson r as the data was either interval or ratio. Furthermore, in regards to the generalisability, the Pearson r was chosen for its accurateness, and also the choosing of other testing methods would not have significantly altered the final result as the large quantity of respondents (N=250) enabled an accurate generalization for the population of Les Roches, regardless of the parametric test selected.
Jupp (2006, p.96) states that research justifies itself as being done for the benefit of all, but even so, ethical concerns arise because of the apprehension between this objective and the concern of those affected by it. In this study the individuals directly affected by the research are the Les Roches students the local authorities and the venue owner. Tharenou, Donohue, & Cooper (2007, p.317) insist that researchers consider possible ethical concerns their study might face before actually carrying out a research project. Savin-Baden & Howell Major (2010, p.20) actually mention that many Universities now require their students to first inquire about “ethical approval” of their chosen research before they beginââ‚¬â„¢ before they even initiate their practical analysis of their study.
Particularly when humans are the participants of a researches’ study (Tharenou, Donohue, & Cooper, 2007, p.317), many ethical principles must be accepted, such as being “respectful to the individuals”, entitling them to be autonomous, as well as providing protection to those who lack autonomy (Marczyk, DeMatteo, & Festinger, 2005, p.255) in addition to ensuring confidentiality and anonymity (Fisher, Buglear, Lowry, Mutch, & Tansley, 2010, p.93). The first page of the questionnaires offered information about the topic of research; the principle of confidentiality and anonymity was as well as pointed out in the questionnaire. Additionally, when the questionnaires were distributed, the students were offered further information about the topic. Prior to conducting the semi-structure interviews, formal e-mail requests were sent to the local authorities and the venue owner to inquire on their participation in the study. Following, whilst the semi-strucured interviews were taking place, the venue owner and police authorities were reminded that their identity will not be discolsed in this research to protect their individual and institutional privacy (Swanson & Holton III, 2005, p. 109).
In addition, the study should be made in manner as to “avoid all unnecessary physical and mental distress and/or harm” (Henn, Weinstein, & Foard, 2006, p.71). Considering the delicate topic covered in this study, the research strategy employing an experiment was not utilized, nor was a case study examined in order to avoid physical damage to the researcher. For this reason the less harmful strategy to adopt in this study was surveys, which relied on participants recollection of past violent events rather than present ones. Because culture was a main factor involved in this analysis, attention was given to not incline any discrimination or biasness towards one specific answer. According to the pilot test and supervisory advice, drug consumption and its effect on violent behaviour were removed from the questionnaire although have been identified by many scholars (Powell, 2011, p.1; Goldstein, 1985, p.1; Homel & Clark, 1994, p.24) as influential factors to aggression. Furthermore, due to ethical consideration related to this matter, in addition to the “Zero Tolerance” associated with the Les Roches International School of Hotel Management policy.
Following, the study must abide to the principle of honesty, in terms of its participants, it must be fair and avoid any type of discrimination in the selection and use of the respondents (Thorpe & Holt, 2008. p.30). As mentioned earlier, the sampling method applied was simple random sampling to Les Roches students that were part of the campus life for over one semester. Other than that no discrimination on either participants nationality, sex or age was made.
Limitations and Credibility
When considering the data in the research, the limitations of the study must be kept in mind. Marczyk, DeMatteo, & Festinger (2005, p. 35) explain that altough research has the main function of describing the association of two things, “The most important limitation pertaining to a study, is that a connection between two things does not allow a presumption of causality”.
The main concern in this research, when acquiring data from the students of Les Roches, is that respondents might provide false information in reference to their ranking of the factors that influence violent behavior due to uncertainty and desire to boast themselves, possibly also due to a sense of anxiousness to complete the survey, they might involuntarily miss questions, thus providing the study with inaccurate results. Even so, the study’s reliability was strengthened with the use of a pilot test Cronbach’s alpha score of .864, that according to Bryman & Bel (2007, p.164) implies a very good level of internal reliability, as it is higher than the recommended 0.7 minimum value stated by Nunnally (1978, cited in Pallant, 2007, p.20).
The study employed a cross-sectional analysis, so in this case, it will not have the benefit of examining the changes over time. Also pertaining to this fact is that the study will be made during the winter season of the school semester, and for this reason there will not be many students who frequent venues, and thus exposed to much violence, considering the majority of students will not be leaving their dorms due to the severe cold weather, for this reason their opinions will not be as valued as if a sample from the summer population of Les Roches Int. School of Hotel Management would be used for this research. Another limitation would be involved in the subjectivity of accidental and deliberate contacts between patrons. Another factor to keep in mind is that violence is more recurrent in some settings than others because of the type of clientele it attracts, also because of a special theme night, violence may occur more frequently.
In reference to representativeness, since a non-probability method of convenience sampling was chosen for the venue owner and local authorities, the data received from the interviewees will be generalized to the limited amount of individuals interviewed. For these reasons the final analysis must take into account that the results may not be generalizable to licensed venues or authorities on a national or global scale. As well in consideration to the Les Roches students, because of the simple random sampling strategy employed, it may occur that because there is a majority of a certain nationality or course, the studied sample might illustrate a heterogeneous population (Jupp, 2006, p. 171) of the Les Roches students, which often occurs when the study concerns human beings (Jill & Johnson, 2002, p.60), thus causing a bimodal distribution to occur (Hatcher, 2003, p. 198).
Validity and Reliability
To promote the validity of the data obtained, a triangulation method was employed as it encouraged the study to obtain the information from different perspectives, by addressing different opinions on the chosen topic, thus allowing the final analysis to be examined thoroughly and therefore allowing for a comprehensive analysis to have been completed (Henn, Weinstein, & Foard, 2006, pp.20-12). Although the inconvenience in this method, which strengthens the reliability of the study, is that it requires time and expertise in both qualitative and quantitative methodology (Swanson & Holton III, 2005, p. 319). Additionally, due to the time sensitivity of this paper, it can be difficult to analyze, as it produces a great amount of statistical data allowing a great deal of observable correlations (Henn, Weinstein, & Foard, 2006, p. 215). Furthermore the analyzed data was regarded as statistically reliable, as it was evaluate through the parametric testing of Pearson r, as mentioned in section 1.5 Data Analysis.