# Research question: # Explore whether consumers’ income and the number of consumption at McDonald’s will affect consumers’ willingness to use "McDonald’s APP" Author Business and Management Taipei University of Technology Wu Donghan 107570047 # chapter 1. Research question In this era of advanced Internet, most people’s lifestyles are inseparable from smartphones, so many companies see this and use various applications for marketing, discounts, and fan interaction to help build The brand image and the establishment of a broad customer base create more business opportunities. In this topic, we will focus on whether consumer income and consumption time affect consumers’ willingness to use "McDonald’s APP", and hope that the results of this topic will become the application for each of us in the future. Or as a reference for various marketing research activities. First of all, let me introduce the background of McDonald's. McDonald’s is a multinational fast food chain in the United States and the world’s largest restaurant chain. It was founded in Chicago, the United States in 1955. McDonald’s has a corporate footprint in 119 countries on six continents and has approximately 36,000 branches worldwide. It provides catering services to customers in more than 100 countries and regions. As early as January 28, 1984, McDonald's established its first restaurant in Taiwan and officially entered the Taiwan market. Due to the advancement of the Internet and the emergence of the current Internet media, brands can break through the one-way communication model of the mass media in the past, making it easier and easier to obtain information. The interactive nature of the Internet media strengthens the mechanism of circular dialogue. It is more helpful for the establishment and maintenance of the relationship between consumers and brands. After the application became popular, it changed the business model of the original social network website and also changed the main purpose of mobile phone users. For enterprises, the Internet has gathered a large number of users and stimulated potential consumers. On May 31, 2016, McDonald's launched a brand new application, and it is the protagonist in this marketing research topic, "McDonald's Newspaper". Looking back, "McDonald's News" was opened for download on May 31, 2016. This application combines the function of an alarm clock and uses the beginning of the day as a connection with consumers. When the alarm sounds, you can get daily surprise discounts. The user can store the discounts in the surprise collection. It also incorporates the weather function. When the user opens the application, he can see the weather and temperature in the area, allowing the application The program is not only a tool for receiving discounts, but also uses daily information to get close to the lives of users. As of December 11, 2020, the "McDonald's Newspaper" has been downloaded more than 7 million times. McDonald's used the "McDonald's Newspaper" and the McDonald's Happy Delivery app to publish new product launches, promotional offers or store opening messages, allowing fans to rank first Hand information, Accumulate popularity quickly, and then achieve the purpose of publicity and marketing. However, after introducing the background of the McDonald’s company and the overview of the "McDonald’s APP", we come back to discuss how consumers’ income and consumption periods reflect their social identity and status. It was mentioned in the documentary "Food Company" about McDonald's fast food culture that many poor people in the United States have to frequent McDonald's because of insufficient meal budgets. In this example, the consumer's income and dining choices are inextricably linked. Therefore, in this special study, under the premise that the McDonald’s APP provides various discounts, we will specifically study whether the income of consumers in the Taiwan market will affect consumers’ willingness to use the "McDonald’s APP". In addition, McDonald's has already launched a 24-hour service in the early years. Who has to go to work at night, McDonald's undoubtedly provides them with alternative catering service options. And often on midnight, who has worked hard at night, is mostly blue-collar. Perhaps this group of people is accustomed to having a simple meal in the face of limited dining options after get off work hours. In this situation ,below the various discounts offered by McDonald’s APPs are even more attractive to them. Therefore, in this special study, we will also specialize in the Taiwan market, whether the consumption period will also affect consumers' willingness to use the "McDonald's APP". # chapter 2. Methodology 1, the definition of questionnaire survey method Works Cited : https://www.sciencedirect.com/topics/earth-and-planetary-sciences/questionnaire-survey V. Preston, in International Encyclopedia of Human Geography, 2009 When Is a Questionnaire Appropriate? A questionnaire survey is only appropriate for certain research questions. Its suitability depends on the types of information needed to answer a research question and the people from whom the researcher wants to elicit information. Questionnaires are inappropriate for collecting information about sensitive topics such as sexual orientation and illicit activities. People will rarely talk about actions that put them at legal risk. The structured nature of the questions and the brevity and superficiality of the social encounter between researcher and respondent do not encourage the intimacy and trust that are prerequisites for people to reveal behaviors, beliefs, and attributes that might be unsanctioned. Questionnaires that rely on people's abilities to convey information accurately are also often ineffective for learning about the past. Over time, memories evolve so that responses about the past are often incomplete. Panel studies in which the same respondents participate in a series of surveys are a good method for learning about past actions and attitudes; however, they are expensive and difficult to implement. Panel studies also pose unique geographical difficulties. To reduce costs, samples are often small so they provide little information about populations in specific places. For example, in Canada and Australia, panel studies of immigrants are rich sources of information about settlement processes. Unfortunately, the small samples mean information is available only for the largest metropolitan areas in each country. Finally, a questionnaire survey is effective only when respondents have knowledge of the topic and they are competent to answer the questions. Questions must be relevant to respondents, and respondents must have the information and the ability to answer. Surveys of households usually exclude children as potential respondents because they often lack the information needed to answer the questions, the issues under study are often not relevant to them, and young children may lack the cognitive ability to answer complex questions. Despite these limitations, numerous research questions and topics may be addressed using information collected with questionnaire surveys. Geographers have used questionnaire surveys to learn about the attributes of many different populations, including employees and owners of firms, neighborhood residents, the homeless, antiglobalization activists, politicians, and refugees. Equally varied behaviors ranging from international migrants’ remittances, men's and women's working conditions, and their travel patterns to food practices and access to health services have been investigated using questionnaire surveys. Survey information has also been collected about residential preferences, mental maps, regional identities, and political opinions. In all of these cases, the success of the survey depended on questionnaire design and administration. We use this method to understand whether consumers' income affects consumers' willingness to use "McDonald's APP". # chapter 3. data analsis ### 1. Load libraries I'm just going load my libraries and data here. ```{r} library(readxl) # read readxl package library(vcd) # read vcd package library(gmodels) # read gmodels package library(knitr) # read knitr package library(dplyr) # read dplyr package library(Hmisc) # read Hmisc package ``` ```{r cars} #continue..... library(vcd) # read vcd package library(dplyr) # read dplyr package library(readxl) # read excel package m1 <- read_excel("D:/m1.xlsx") # read excel file name # rstudio can only read English file name View(m1) # View excel file data <- m1 # excel file assigned to data View(data) # View excel file #Excel file in R, data column name change data <- rename(data, ID = "時間戳記", sex = "您的性別", age = "您的年齡" , habits ="請問您平常是否有吃速食的習慣?", job = "您的身份", income = "請問您一個月的可支配所得?", timelevels ="請問您最常去麥當勞消費的時段是?" , times = "請問您通常一個月到麥當勞消費幾次?", everused = "請問您是否使用過麥當勞報報APP?", coupons = "請問您一個月內使用麥當勞報報app裡的優惠卷次數為多少？") #data <- rename(data, #ID = "Timestamp", ##representing the time when the form was completed #sex = "Your sex", ##demographic variable statistics #age = "Your age", ##demographic variable statistics #habits ="Do you usually have the habit of eating fast food?", ##Behavioral variable statistics #job = "your identity", ##demographic variable statistics #income = "What is your disposable income for a month?", ##demographic statistics #timelevels ="What time do you spend most frequently at McDonald's?", ##Behavioral variable statistics #times = "How many times do you usually spend at McDonald's in a month?", ##Behavioral Variable Statistics #everused = "Have you ever used the McDonald's newspaper app?", ##Behavioral variable statistics #coupons = "How many times have you used the coupons in the McDonald's newspaper app in a month?" ##Statistics on behavioral variables) #Excel file in R, data column name change data <- rename(data, couponstimes = "承上題，假如麥當勞報報中沒有優惠卷的功能，會不會影響您到麥當勞消費的次數", Features = "您最滿意此APP裡的哪項功能?", Buyonegetonefree = "您最希望在麥當勞報報中收到何種優惠? [ 買一送一]", Redeemoffer = "您最希望在麥當勞報報中收到何種優惠? [銅板加價優惠（多加 1 元/10 元兌換優惠 ）]", buyamealtogetfood = "您最希望在麥當勞報報中收到何種優惠? [附贈餐點類型（買套餐送點心/買主餐送飲料）]" ) data <- rename(data, Packagespecials = "您最希望在麥當勞報報中收到何種優惠? [套餐特價(例:麥克雞塊方享特價180元 )]", motivations = "您曾經使用這個APP的動機大多是?", Highpricegood = "您平常會去消費高單價商品(100元以上非套餐單品)嗎?", Buyhighpricedgoodsatadiscount = "承上題，您會因為抽到優惠，而去購買平時不會消費的高單價商品嗎?", ) # bug - - - - Couponisthebiggestreason = "優惠劵是你下載這個APP的最大原因" - - -because recode the UTF - 8 data <- rename(data, spendsmoneygetacoupon = "您是個因為抽到喜歡的優惠券而去消費的人", usedspendsmoneygetacoupon = '您曾經因為抽到/兌換到自己喜愛的優惠券而"實際"去消費' ) #data <- rename(data, #couponstimes = "Continuing from the above question, if there is no coupon function in the McDonald's newspaper, will it affect the number of times you spend #at McDonald's", ##behavioral variable #Features = "Which feature in this app are you most satisfied with?", ##psychological variables #Buyonegetonefree = "What kind of discount do you most hope to receive in the McDonald's newspaper? [Buy one get one free]", ##psychological variable #Redeemoffer = "What kind of discount do you most hope to receive in the McDonald's newspaper? [Copper plate price increase (add 1 yuan/10 yuan to redeem the #discount)]", ##psychological variable #buyamealtogetfood = "What kind of discount do you most hope to receive in the McDonald's newspaper? [Bonus meal type (buy a set meal to get snacks/buy a meal to get #drinks)]" ##Psychological variables) #data <- rename(data, #Packagespecials = "What kind of special offer do you want to receive in the McDonald's newspaper? [Package specials (for example: Mike's chicken nuggets will enjoy #a special price of 180 yuan)]", ##psychological variable #motivations = "Most of your motivations for using this app?", ##psychological variables #Highpricegood = "Do you usually go to consume high-priced goods (not package items above 100 yuan)?", ##psychological variable #Buyhighpricedgoodsatadiscount = "Continuing from the question above, will you buy high-priced goods that you would not normally consume because of the discount?", ##psychological variable #Couponisthebiggestreason = "Coupon is the biggest reason for you to download this app", ##psychological variable) #data <- rename(data, #spendsmoneygetacoupon = "You are a person who spends because you get a coupon you like", ##behavioral variable #usedspendsmoneygetacoupon ='You used to "actually" spend because you drew/redeemed your favorite coupons', ##behavioral variables ) ``` ### 2. data Cleanup ```{r} data data$sex <- recode(data$sex, "女" = "F", "男" = "M") #Excel file in R, data column sex change gender data to F and M data$age <- recode(data$age, "18-30歲" = "young", "61歲以上" = "old") #Excel file in R, data column AGE changes age data to young and old data$habits <- recode(data$habits, "有，麥當勞" = "TRUE", "無" = "FALSE", "有，其他品牌" = "NONE") #Excel file in R, the data column habits will be changed to TRUE/FALSE/NONE data$Features <- recode(data$Features, "以上均不滿意" = "none", "填寫餐廳滿意度回饋送優惠卷" = "a", "點點卡帳戶查詢與管理" = "b", "每日抽優惠券/歡樂貼" = "c") #Excel file in R, the data column Features change the consumption hobby data to none/a/b/c data$job <- recode(data$job, "學生" = "student", "在職" = "work") #Excel file in R, change the data column job to student and work data$motivations<- recode(data$motivations, "抽到/兌換到喜歡的優惠券而去消費" = "a", "因為剛好去吃麥當勞而使用這個APP" = "b") #Excel file in R, the data column motivations change the data to a and b data$timelevels <- recode(data$timelevels, "21:31-隔天早餐時段前" = "midnight", "16:31-21:30" = "afternoon", "5:00-10:30" = "morning", "10:31-16:30" = "noon") #Excel file in R, the data column timelevels change the data to midnight/afternoon/morning/noon data$income <- recode(data$income, "20000以上" = "more", "10001-15000" = "more", "5001-10000" = "more", "5000以下" = "less", ) #Excel file in R, the data column income change the data to more and less data$times <- recode(data$times, "9次以上" = "more", "5-8次" = "more", "2-4次" = "more", "0-1次" = "less", ) #Excel file in R, the data column times change the data to more and less data$coupons <- recode(data$coupons, "9次以上" = "more", "5-8次" = "more", "2-4次" = "more", "0-1次" = "less", ) #Excel file in R, the data column coupons change the data to more and less data$everused <- recode(data$everused, "是" = "yes", "否" = "no") #Excel file in R, the data column everused change the data to yes and no data$couponstimes <- recode(data$couponstimes, "會" = "yes", "不會" = "no") #Excel file in R, the data column couponstimes change the data to yes and no data$Highpricegood <- recode(data$Highpricegood, "會" = "yes", "不會" = "no") #Excel file in R, the data column Highpricegood change the data to yes and no data$Buyhighpricedgoodsatadiscount <- recode(data$Buyhighpricedgoodsatadiscount , "會" = "yes", "不會" = "no") #Excel file in R, the data column Buyhighpricedgoodsatadiscount change the data to yes and no data$Buyonegetonefree <- recode(data$Buyonegetonefree, "第一順位" = "the first", "第二順位" = "second", "第三順位" = "third", "第四順位" = "fourth") #Excel file in R, the data column Buyonegetonefree change the data to the first/second/third/fourth data$Redeemoffer <- recode(data$ Redeemoffer, "第一順位" = "the first", "第二順位" = "second", "第三順位" = "third", "第四順位" = "fourth") #Excel file in R, the data column Redeemoffer change the data to the first/second/third/fourth data$buyamealtogetfood <- recode(data$buyamealtogetfood , "第一順位" = "the first", "第二順位" = "second", "第三順位" = "third", "第四順位" = "fourth") #Excel file in R, the data column buyamealtogetfood change the data to the first/second/third/fourth data$Packagespecials <- recode(data$Packagespecials, "第一順位" = "the first", "第二順位" = "second", "第三順位" = "third", "第四順位" = "fourth") #Excel file in R, the data column Packagespecials change the data to the first/second/third/fourth data$income_reordered <- ordered(data$income, levels = c("more","less")) # order two level on income data$times_reordered <- ordered(data$times, levels = c("more","less")) # order two level on times data$coupons_reordered <- ordered(data$coupons, levels = c("more","less")) # order two level on coupons ``` ### 3. Descriptive Statistics This is basic descritive statistics of the relevant categorical variables. ```{r} head(data) describe(data$sex) describe(data$age) describe(data$habits) describe(data$Features) describe(data$job) describe(data$motivations) describe(data$timelevels) describe(data$income) describe(data$times) describe(data$coupons) describe(data$everused) describe(data$couponstimes) describe(data$Highpricegood) describe(data$Buyhighpricedgoodsatadiscount) describe(data$Buyonegetonefree) describe(data$Redeemoffer) describe(data$buyamealtogetfood) describe(data$Packagespecials) ``` This is basic descritive statistics of the relevant variables by subgroup ```{r} by(data$age, data$sex, describe) by(data$sex,data$habits, describe) by(data$Features,data$habits, describe) by(data$Features,data$job, describe) by(data$motivations,data$job, describe) by(data$motivations,data$timelevels, describe) by(data$income,data$timelevels, describe) by(data$income,data$times, describe) by(data$coupons,data$times, describe) ``` ### 4. Relationship between 2 categorical variables (Chi-sq test) This tests the relationship between 2 categoricial variables. ```{r} CrossTable(x=data$age, y=data$sex, prop.r=TRUE, prop.c=FALSE, prop.t=FALSE, prop.chisq=FALSE, chisq=TRUE) CrossTable(x=data$sex, y=data$job, prop.r=TRUE, prop.c=FALSE, prop.t=FALSE, prop.chisq=FALSE, chisq=TRUE) CrossTable(x=data$age, y=data$job, prop.r=TRUE, prop.c=FALSE, prop.t=FALSE, prop.chisq=FALSE, chisq=TRUE) CrossTable(x=data$age, y=data$everused, prop.r=TRUE, prop.c=FALSE, prop.t=FALSE, prop.chisq=FALSE, chisq=TRUE) ``` #### 4A. Testing males and females separately ```{r} data$IsMale <- recode(data$sex, "M" = TRUE, "F" = FALSE) CrossTable(data$income[data$IsMale],data$job[data$IsMale], prop.r=TRUE, prop.c=FALSE, prop.t=FALSE, prop.chisq=FALSE, chisq=TRUE) ``` This shows that whether a male income [is/is not] significantly related to whether the male job ```{r} CrossTable(data$income[!data$IsMale],data$job[!data$IsMale], prop.r=TRUE, prop.c=FALSE, prop.t=FALSE, prop.chisq=FALSE, chisq=TRUE) ``` However, whether a female received a income [is/is not] significantly related to whether the female job. ###5. PASS CAUSE NO Our questionnaire has no continuous variables for analysis ### 6. Logistic Regression, with Binary dependent variable This tests the relationship between 2 categorcial variables. ```{r} data$couponstimes_2levels <- recode(data$couponstimes, "yes" = TRUE, "no" = FALSE) #I created a binary variable called couponstimes_2levels to store whether no coupon function in the McDonald's newspaper, will it affect the number of times to spend data$everused_2levels <- recode(data$everused, "yes" = TRUE, "no" = FALSE) #I created a binary variable called everused_2levels to store whether consumers ever used the McDonald's newspaper app. data$sex_2levels <- recode(data$sex, "M" = TRUE, "F" = FALSE) #I created a binary variable called sex_2levels to store whether consumers's gender will affect to spend. data$timelevels_2levels <- recode(data$timelevels, "midnight" = TRUE ,"morning" = TRUE , "afternoon" = FALSE, "noon" = FALSE) data$timelevels <- ordered(data$timelevels, levels = c( "midnight" ,"morning", "afternoon", "noon" )) #I created a binary variable called sex_2levels to store whether consumers's go to McDonald's time level. summary(glm(data$couponstimes_2levels ~ data$income + data$times + data$everused + data$job , family = "binomial")) summary(glm(data$everused_2levels ~ data$income + data$times + data$job , family = "binomial")) summary(glm(data$sex_2levels ~ data$income + data$times + data$job + data$everused , family = "binomial")) summary(glm(data$timelevels_2levels ~ data$income + data$times + data$job + data$everused , family = "binomial")) ``` # Chapter 4. Conclusion This is my research conclusion. The result is that the p value is too large to be judged. Call: glm(formula = data$couponstimes_2levels ~ data$income + data$times + data$everused + data$job, family = "binomial") Deviance Residuals: Min 1Q Median 3Q Max -1.9836 -0.7419 0.4608 0.7263 1.6880 Coefficients: Estimate Std. Error z value Pr(>|z|) (Intercept) -1.1495 1.1436 -1.005 0.3148 data$incomemore -2.0905 1.3489 -1.550 0.1212 data$timesmore 0.3724 1.2046 0.309 0.7572 data$everusedyes 2.9662 1.6233 1.827 0.0677 data$jobwork 1.0996 1.4147 0.777 0.4370 --- Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1 (Dispersion parameter for binomial family taken to be 1) Null deviance: 30.789 on 22 degrees of freedom Residual deviance: 23.335 on 18 degrees of freedom AIC: 33.335 Number of Fisher Scoring iterations: 4 We collected too few samples. We hope to obtain results from our research topics, that is, whether consumers’ income and McDonald’s consumption will affect consumers’ willingness to use the “McDonald’s APP”, which cannot be seen from the results of our questionnaire. If the sample size is large enough, the analysis should show correlation. In addition, McDonald's has already launched a 24-hour service in the early years. Who has to go to work at night, McDonald's undoubtedly provides them with alternative catering service options. data$timelevels Value afternoon midnight morning noon Frequency 10 4 3 6 Proportion 0.435 0.174 0.130 0.261 Judging from the results above, in this survey, consumers prefer afternoon meals Below, the various discounts offered by McDonald’s APPs are even more attractive to them. Therefore, in this special study, we will also specialize in the Taiwan market, whether the consumption period will also affect consumers' willingness to use the "McDonald's APP". data$everused n missing distinct 23 0 2 Value no yes Frequency 5 18 Proportion 0.217 0.783 Judging from the results above, in this survey, consumers prefer to use apps In this special study, under the premise that the McDonald's APP provides various discounts, in the Taiwan market will affect consumers' willingness to use the "McDonald's APP". data$Buyonegetonefree n missing distinct 18 5 4 Value fourth second the first third Frequency 2 1 13 2 Proportion 0.111 0.056 0.722 0.111 data$Redeemoffer n missing distinct 18 5 4 Value fourth second the first third Frequency 1 12 2 3 Proportion 0.056 0.667 0.111 0.167 data$buyamealtogetfood n missing distinct 18 5 4 Value fourth second the first third Frequency 6 3 2 7 Proportion 0.333 0.167 0.111 0.389 data$Packagespecials n missing distinct 18 5 4 Value fourth second the first third Frequency 9 2 1 6 Proportion 0.500 0.111 0.056 0.333 From the discount ranking above, we can see that buy one get one free is very attractive. Although the results of the questionnaire cannot be used as strong evidence to implement consumer decision-making, it can be used as a small R data analysis exercise.