Understanding Methods and Analysis - Research Paper Example

Understanding Research Methods and Analysis Introduction Studies have been made on purchasing influences in Thailand, which encompasses a different set of cultural reasons as to why people do make their purchases, particularly when it comes to celebrity endorsements. Consumers are potentially likely to make purchases, based on star-power endorsement marketing, or the use of a celebrity’s picture and name, which enhances the product’s value (Alsmadi, 2006). Indeed, consumers have a more emotional response, and also now investigate products on the Internet first, before making the purchase. However, Thailand respondents appear to be split on whether a celebrity influences their purchasing habits (Nida Poll, 2014). Yet, it is also evident that the message, combined with star-power, may be the culprit, if sales are not working out well. The research project, therefore, aims to look at the effects of celebrity influence on a product or brand’s image, as combined with certain variables, such as gender differences. The product in question would be skincare and cosmetic products and how the message is presented to the public. Such a subject affects both men and women, particularly in the late teenage to young adult age groups, although more women than men may be likely to continue purchasing long after those years have passed, and for different reasons than when they first began making these purchases. The research also attempts a look at whether respondents buy a product, due to the celebrity endorsement, rather than the brand itself. Justification of the Proposed Research Method In the proposal, secondary research was done initially which also helped with developing a questionnaire that will be used in Qualtrics, a professional university-subscribed program. The research was initially conducted through online professional retail and marketing industry journals, trade retail books, and online news articles and conference papers, in order to gather up foundational information for the dissertation and research design process. It was then decided to include and conduct a primary study, utilizing the university Qualtrics survey program, whereby potential respondents could be invited to take a questionnaire concerning the influences of why they buy skin care and cosmetic products (Mitchell & Jolley, 1996). There are a number of ways to ask respondents to take this, which include contacting acquaintances, friends, and family members through social media contact, personally asking university fellow students to take it, and handing out hand designed pre-made invitation cards with the Qualtrics’ web site address on it. As part of the methodology, the snowball method will be used, whereby those who are directly asked to participate, can also ask others to take it, so long as they fit the criteria. Additional cards will be provided to respondents to hand out, and hand-designed print invitation advertisements will be placed in different suitable places that will accept them. In all cases, the potential respondent is asked whether he or she is 18 and older as a main qualifier criteria, including whether they have bought, or currently buy, skin care and cosmetic products (Dillman, Smyth, & Christian, 2008). It is hoped to attain 100 to 200 respondents for a reasonable analysis sample size. All respondents, at the time of the invitation, are informed that their personal information is kept confidential. The Data Collection Instrument Reflection on Predefined Research Questions The questionnaire has been designed to achieve several objectives, particularly when scaling emotions and purchasing influences, such as a celebrity presence, which heavily influences the consumer to buy the product (Bearden, Netemeyer, & Haws, 2010; Dillman, Smyth, & Christian, 2008). The first five questions deal with personal attributes such as gender (M, F), age group (six selections), education level, current occupation, and average monthly income (Buckingham & Saunders, 2004). In this first five questions, the descriptive of the respondents is later determined through a simple descriptive analysis in SPSS, along with the rest of questions. When conducting more advanced analyses, different attribute variables may be used in connection with another question, or set of questions, to find a correlation, or a prediction in a regression analysis (Alsmadi, 2006; Field, 2013; Mitchell & Jolley, 1996). The second part of the questionnaire, as a rating scale application, asks questions on how often a respondent goes shopping, how often one purchases unexpected products (impulse purchases), and if respondents buy skin care products because of a recognizable celebrity featured on the product packaging or an ad. Other questions continue in that same vein of whether one makes a purchase of a product because of a featured celebrity, which also implies that the celebrity uses it as well. In the third section, this becomes a more experimental process to determine which is more important, the celebrity or the brand name, and which gender may be more affected and why. While a respondent may not remember the brand name, the celebrity picture on a package indicates that this may be the better product than others without a celebrity endorsement. There are also questions about whether the respondent will purchase a product that is celebrity-endorsed, even if the prices are higher than other similar surrounding products (Mitchell & Jolley, 1996). Even how celebrities are shown (pictured) on two different products, may make a difference in which one is purchased (sex appeal/design layout/design colour choices). Other reasons why respondents may make a choice to purchase one brand over another is the reviews attached to the product. If more reviewers say that a skin care product has worked wonderfully for their skin, then respondents may purchase that product over another one, even if it is celebrity-endorsed. An excellent example of such an occurrence might be seen on Amazon, whereby a product has 1,000 likes, four to five-star reviews, with many purchasers taking the time to write a review about a product they love. This brand is already popular and there is no celebrity endorsement on the package, although other advertisements for Olay may use current celebrity, Katie Holmes. Figure 1. Olay’s Pro-X Skin Cleansing System with 4.5-Star Positive Reviews Source: (Amazon Online, 2015) The design of the questionnaire responses is relatively simple in scale, using a simple yes or no selection, or an A or B for which picture of a celebrity in an advertisement is more influential in the purchase (Carifio & Perla, 2007). Several questions do use a rating scale, such as ‘several times a week,’ ‘once or twice a week,’ ‘a few times a month,’ or ‘very rarely.’ Another response design uses ‘Always,’ ‘Usually,’ ‘Sometimes,’ ‘Rarely,’ and ‘Never.’ These are similar to the Likert scale which measures the degree of an action, or an emotional response, similar to ‘Strongly Agree,’ all the way down to ‘Strongly Disagree.’ Appropriate Sampling Technique Discussion For the foundation of this proposal and subsequent dissertation, the sample plan reflects respondents in higher education who are 18 years of age, and who have purchased skin care and cosmetic products in the past, particularly within the last year. With this criteria, it can be safely assumed that all respondents will fit the probability sample objective of using only those who have an interest or knowledge of purchasing these products and what influences them most, as reasons for such purchases (Short, Ketchen, & Palmer, 2002; Readman, 2015). The sample goal is up to 200 but may also work well with 100, depending on the ease of gaining such respondents within an appropriate time span. The snowball technique will be utilized to encourage more respondents to take the questionnaire in order to attain a robust group. As the category of skin care and cosmetic products is used by nearly everyone, including men with skin care, the target population is assured to be a relatively good sample of the population overall. As noted in the first section for justification of this research, respondents will be contacted using acquaintances, friends, and family members through social media contact, personally asking university fellow students to take it, and handing out hand designed pre-made invitation cards with the Qualtrics’ web site address on it, to those who become a random selection from casual meetings. In effect, this is called convenience sampling. All must meet the two simple criteria and all are assured of confidentiality. This also provides acceptable confidence levels (95%) and margins of error ( 5%) (Readman, 2015). In research study samples that encompass larger sample sizes, the Krejcie & Morgan Table can be used to determine a suitable size out of a total population group, such as a census population. There are many ways that an appropriate sample of a population can be achieved, such as purchasing a list from a sample collection vendor containing all college mathematics professors, or all colleges which use a certain textbook. Some survey companies, such as Survey Monkey, provide a customer-paid list of respondents who are vetted through a short survey to determine if they fit the required criteria to move on to the questionnaire. The final sample group can be selected through random sampling of a large data list, using Microsoft Excel’s random sampling function tool to select a sample, based on criteria put into the function form. Another process is cluster sampling, using variables to determine similar sets of similar data. Any clusters over a certain criteria, can be eliminated, and then the process is repeated (Romesburg, 2004). For example, there can be sets 1,2,3,4, and 5, but 3 and 4 have the identical points of interest needed to conduct further analyses. Appropriate Analytical Techniques Useful for Testing Research Propositions and Hypotheses The data will be collected, initially, from the Qualtrics program and downloaded into Microsoft Excel, whereupon it will be cleaned and coded as necessary. At the conclusion of cleaning, the coded worksheet will be uploaded into IBM’s SPSS analysis program for further examination and analyses. Some of the component variables that will be looked at are brand recognition and trust, consumer behaviour and decisions, emotional responses, and gender correlation, along with salary, education, job status, and age. Different combinations of these variables will provide answers to the question of the research and to validating (or not) any hypotheses developed. Categorical Data and Cross-Tabulation with Chi-Square This is a process where similar variables, including responses, are put into categories and, in the case of cross-tabulation as found in SPSS, will provide an output that joins several variables together, intersecting at rows and columns with the analysis output (Field, 2013). In determining a statistically significant relationship between two nominal variables, Chi-Square is used when there are two rows and two columns (2 X 2) in the format. When there are more than 2 X 2, then phi and Cramer’s V is selected. Kendall’s tau-b is used for ordinal variables, and eta for one nominal and one variable chosen as a normal or scale variable. These are selected in the Variable View tab when first setting the information into SPSS (Morgan et al., 2012). Fisher’s exact test can also be selected when the sample is small and is a 2 X 2 output, and becomes part of the report instead of using the Chi-Square output (Field, 2013). Descriptive Analysis and ANOVA Descriptive analysis provides a front end view of the statistics of variables by showing a central tendency, the percentages of each group or category, frequency distribution, and the shape of the distribution of a selected variable. These can be easily seen in histograms and graph outputs. The more complex descriptive output can show two or more variables, such as the percentages of males versus females in the Islamic religion, Catholics, and Jews (Morgan et al., 2012). Skewness of a distribution is also shown in a bell-curve graph which can show a symmetric or a skewed distribution (Field, 2013). A one-way ANOVA t-test can be run when two categorically independent variables are considered in combined analysis. The two-way ANOVA is applied when using three variables simultaneously, such as asking if there is a significant interaction statistically, with independent variables one and two, with dependent variable one (Morgan et al., 2012; Field, 2013). It is also important to note a Type 1 error which rejects a null hypothesis, and indicates that there is no difference between groups, when the assumption was that this was true. The Type 2 error accepts the null hypothesis when there is a difference between groups, and the assumption was that there was no difference (Field, 2013). Correlation and Regression Analysis Correlation analysis measures how two variables are related and also the response patterns across variables, For example, if females tend to purchase those skin care and cosmetics products which are endorsed by one particular celebrity, then it might be safe to assume that the new product coming out, using that same celebrity with the same consumer target, will also be very successful. The correlation process determines what is currently in place as being connected, showing degrees of freedom, while the regression analysis will tell us whether that assumption or hypothesis is true, using the linear plot line (Morgan et al., 2012). Regression analysis occurs when the researcher wishes to make a prediction, based on the current data as seen in correlation. Scatterplot outputs provide a visual method of determining if the prediction will happen, such as in a tightly gathered series around, or along a line, providing an assumption of linearity. The Bivariate regression process puts the independent variable on the X axis line, while the dependent variable goes on the Y axis line. The slope of the line and its intercept is also analysed Tightly plotted points on the line, which runs from lower left to upper right (high positive), would show a distinctly positive outcome. When it is a negative outcome, then the line runs from the upper left to the lower right, and with zero correlations, the regression line is flat, with the points scattered out, away from the line ((Morgan et al., 2012). Multiple regressions uses more predictor variables. A good example of this might be if the education level of the mother and father, directly relates to how well the child will do in school by gender. This presents four predictor variables: the mother and father’s education levels (2) and the gender and grade level of the child at school (2) (Morgan et al., 2012). Conclusion There are many ways to understand the information given by respondents, and analyses, therefore, must be a precise art in how the form in SPSS is set up, and what particular selections must be made in order to get the right outcomes. The use of SPSS will provide a far more in-depth view of the data and it is hoped that the findings in this research study will add to current research studies in how skin care and cosmetic products are being purchased when they are combined with the endorsement of a popular celebrity (Alsmadi, 2006). Admittedly, different cultures see such advertising and marketing in different ways, such as in the Asian countries, like Thailand. Therefore, it may be that celebrity endorsements, using the right pictures to present a message, is indeed, the best way to market such products in this culture, but far different elsewhere. Resources Alsmadi, S. (2006) The Power of Celebrity Endorsement in Brand Choice Behaviour: an empirical study of consumer attitudes. Journal of Accounting, Business & Management. Vol.13, pp69-84. Bearden, W.O., Netemeyer, R.G., & Haws, K.L (2010) Handbook of Marketing Scales: multi-item measures for marketing and consumer behaviour research. 3rd ed, Thousand Oaks: SAGE Publications. Buckingham, A., & Saunders, P. (2004) The Survey Methods Workbook: from design to analysis. Bristol: Polity. Carifio, J.M., & Perla, R.J. (2007) Ten Common Misunderstandings, Misconceptions, Persistent Myths and Urban Legends about Likert Scales and Likert Response Formats and their Antidotes. Journal of Social Sciences, Vol. 3, No. 3, pp106-116. Retrieved from http://thescipub.com/PDF/jssp.2007.106.116.pdf Dillman, D.A., Smyth, J.D., & Christian, L.M. (2008) Internet, Mail, and Mixed-Mode Surveys: the tailored design method. 3rd ed, Hoboken: Wiley. Field, A. (2013) Discovering Statistics using IBM SPSS Statistics. 4th ed, New York: SAGE Publications. Mitchell, M., & Jolley, J. (1996) Research Design Explained. 3rd ed, Fort Worth: Harcourt Brace. Morgan, G.A., Leech, N.L., Gloeckner, G.W., & Barrett, K.C. (2012) IBM SPSS for Introductory Statistics: use and interpretation. 5th ed. Florence: Routledge. (Morgan et al., 2012) Nida Poll. (2014) Presenter and purchase intention [Online]. Retrieved from: http://nidapoll.nida.ac.th/top_news-พรีเซ็นเตอร์ดารากับการเลือกซื้อสินค้า-68-151.html#.VSuWmYe27sE Readman, J. (2015) ‘Quantitative Research Methods Sampling’. [Lecture PowerPoint Slides] Brighton: University of Brighton Business School, Unpublished. Romesburg, H.C. (2004) Cluster Analysis for Researchers. Raleigh: Lulu Press. Short, J.C., Ketchen, Jr., D.J., & Palmer, T.B. (2002) The Role of Sampling in Strategic Management Research on Performance: a two-study analysis. Journal of Management, Vol. 28, No. 3, pp363-385. Retrieved from http://www.sciencedirect.com/science/article/pii/S0149206302001320 Read More