PaPer Ital. J. Food Sci., vol. 27 - 2015 1 - Keywords: Ice cream attributes, manufacturer/national brand, principal component, cluster and multiple regression analyses, Turkish consumers’ purchase decisions - TURKISH CONSUMER DECISIONS AFFECTING ICE CREAM CONSUMPTION YAVUZ TOPCU Department of Agricultural Economics, Collage of Agriculture, Ataturk University 25240-Erzurum, Turkey Tel. 90 4422311393, email: yavuztopcu@atauni.edu.tr AbstrAct the aim of the study is to determine the main factors affecting the national-branded ice cream preferences of turkish consumers, and to analyze the relationships between their preferences and consumption amounts. the data obtained from 400 households in Kahramanmaras, turkey were used for Principal Component, K-Means Cluster and Multiple Regression Analyses. the re- sults of the study highlighted clearly that the consumers of the high (C1), middle (C2) and low-in- come users (C3) satisfied with the manufacturer brand, the individual private label, and the re- tailer brands on the ice cream purchase decision and consumption amounts, respectively. the manufacturers, retailers and marketers of the ice cream, therefore, should implement effectively the manufacturer brand, the private label and the retailer brand for C1, C2 and C3, respectively, and then they could also increase the demand trends of the target consumers segments satisfied. 2 Ital. J. Food Sci., vol. 27 - 2015 IntroductIon Consumers purchase decisions towards the food products are a complex phenomenon influ- enced by a numerous factors classified as prod- uct-related (intrinsic and extrinsic food attributes), consumer-related (demographic, personal, psy- chological and physiological characteristics) and marketing environmental-related (economic, cul- tural, natural, technological, political and social environments) (REALINI et al., 2013; TOPCU and UZUNDUMLU, 2012; TROY and KERRY, 2010; TOP- CU et al., 2009). Traditional sensory analyses fo- cused on the intrinsic product attributes are not sufficient to meet not only the requirements of today’s fast developing food markets but also the food choices of the consumers at the sale points. In order to be able to reply the consumers’ need and willingness to buy, therefore, the suppliers must evaluate the food products with integrated approaches covering their intrinsic and extrinsic attributes by thinking of the consumers’ socioec- onomics characteristics. In order to design the actual food product im- age optimized, the suppliers have focused on not only the intrinsic ice cream1 attributes such as the sensory, structural, visual, nutritional, chem- ical and confidential properties including in aro- ma, taste, flavor, viscosity, color, texture and its content (REALINI et al., 2013; MENICHELLI et al., 2012; TOPCU and UZUNDUMLU, 2012; SOUKOU- LIS et al., 2010; SIMEONE and MAROTTA, 2010; TOPCU et al., 2009) but also the extrinsic ones consisting of hedonic quality attributes such as the price, country of origin, actual product im- age and quality, brand, labelling, packaging, pro- motion, advertising, etc. (TOPCU and UZUNDUM- LU, 2012; TROY and KERRY, 2010; TOPCU et al., 2009; SIRO et al., 2008; McCARTY et al., 2003; ORTH and FIRBASOVA, 2003) along with the con- sumers’ socioeconomic characteristics such as in- come, food expenditure, education, age and life- cycle, occupation, etc. (TOPCU, 2012; TOPCU and UZUNDUMLU, 2012; TOPCU et al., 2009). While the consumers have considered the he- donic and visual quality attributes of the ice creams at point of sale before purchase and the sensory ones after consumption; the manufactur- er, retailer and marketers have also interested with their various attributes to develop, improve and design the innovative ice creams based on their need and willingness to buy. Providing an important communication among the manufacturer, retailer and the consumers, and then establishing a strong bridge between each group; the brands, one of the utmost important pieces of the information read a foodstuff, play an important role on the consumers’ purchase deci- sion making processes (TOPCU et al., 2008; WULF et al., 2005; GUERRERO et al., 2000). Their ef- fects on the purchase decisions, therefore, could be explained by several functions including in their identification and attributes, the reference function formalizing their purchase models, the guarantee function providing the quality image by reducing the feeling of risk, the personalization function al- lowing the consumers to locate themselves in their social environment, the entertainment function get- ting the consumers to motivate into the different brand choices, and the practical function allowing the consumers to learn and evaluate the results of different shopping experiences (TOPCU and UZUN- DUMLU, 2012; AILAWADI et al., 2011; ALDAZABAL et al., 2006; GUERRERO et al., 2000). Not only has been consumed increasingly the ice cream by many consumer segments such as chil- dren, adolescents, adults and elder people during all the lifecycle due to their refreshing, sensational, nutritional and sanitarian attributes, on the oth- er hand, but also the ice cream industry becom- ing a much profitable sub-sector owing to rapid- ly developing technological progresses has intro- duced 240 different types of ice cream resulting from the diverse ingredients and methods of freez- ing to the domestic markets under a strict com- petition between manufacturer and retailer in re- cent years (TOPCU and UZUNDUMLU, 2012; TUR- GUT and CAKMAKCI, 2009). Therefore, the upmost motivation drives of the fundamental selection criteria in the ice cream pur- chase decision making process of all the consumers under various product depths and varieties are the brands. Therefore, it makes possible to distinguish among the consumers in terms of their sensitivity and loyalty to the brand highlighting the persistent purchase of a specific brand within a well-defined context and having a positive attitude towards it. In other word, there is a strong relationship between their varieties and brand types impacting on their purchase decisions (TOPCU, 2012a; ENNEKING et al., 2007; GUERRERO et al., 2000). The brand type or names of the food products could be generally explained in two categories as the manufacturer brands created by manufactur- ers and bearing their chosen brand name (the lo- cal, national, international and global brands) and private label products derived and owned by re- tailers functioning at a distribution channel (the store/retailer, store sub-brand, generic and indi- vidual product brands) (TOPCU et al., 2008). The effects of the brand types associated with the food choice and acceptability, and the consumers pur- chase attitude and behaviors were widely studied in marketing researches (FORNERINO and HAUTE- VILLE, 2010; ZHOU et al., 2010; GEHLHAR et al., 2009; KUMAR et al., 2009; TOPCU et al., 2009; DIMOFTE et al., 2008; ENNEKING et al., 2007). 1 Ice cream has a complex food colloidal system includ- ing in air bubbles, ice crystals and partially destabilized fat globules dispersed in a continuous aqueous phase. therefore, its manufacturing as the most popular frozen dairy dessert has fairly complex processes and follows the subsequent steps such as preparation of the mix- ture, pasteurization, homogenization, cooling, aging of the mixture, addition of fermented milk, freezing, pack- aging and storage of the final mixture (souKouLIs et al. 2010). Ital. J. Food Sci., vol. 27 - 2015 3 The brands of the leader manufacturer manu- facturing the ice cream with the national and in- ternational brands in Turkey consist of Unilever’s Algida, Ulker’s Natura, Has Food’s Panda, Dina- mik Food’s Alaska, Izmir Dairy Products’ MEMO, Nestlé’s Nestle, Yasar Food’s MADO and Ferah Food’s EDO (FOOD, 2009). Their annual produc- tion, export, import, consumption and per capi- ta consumption amounts of the ice cream along with some macroeconomic indicators in Turkey are indicated in Table 1. According to Table 1, as considered the annual increasing production (7.1%), consumption (7.5%) and export (8.3%) trends between 2005 and 2012; it should be increased dramatically the domestic production amount responding to the needs and willingness to buy of Turkish consumers. Addi- tionally, the annual increase rate in the consum- er population (0.6%) and Gross Domestic Product (GDP) per capita (1.9%) between 2005 and 2012 in Turkey could increase considerably its consump- tion trends. Furthermore, it has been estimated that the ice cream consumption at selected Euro- pean markets reported by MGMN for Turkey will also increase about 64% between 2012 and 2016. Annual ice cream consumption amounts per capita of the leader countries such as New-Zea- land, US, Finland and Sweden in the ice cream consumption in 2012, on the other hand; were calculated as 25, 21, 14 and 12 liters, respective- ly (FIIND, 2013). Furthermore, those per capita in Turkey, EU and the world in 2012 were calculated as 4, 12 and 6.0 liters, respectively (EICA, 2013; FAOSTAT, 2013; TUIK, 2013). As a result of all this, compared with the con- sumers of the leader countries in the ice cream consumption, Turkish consumers have consumed about 3-6 times less ice cream than them, but this difference has continued to increase steadily. Ac- cording to all the indicators, the ice cream pro- duction should be increased to be able to meet the increasing demands of Turkish consumers at domestic food markets. In order to increase considerably the annual ice cream consumption amounts of Turkish consum- ers, traditional sensory analysis should be com- bined with integrated marketing approaches fo- cused on the intrinsic and extrinsic ice cream at- tributes, and then they should develop/design the marketing tactic and strategies under homogenous consumer segments based on their socioeconomic characteristics. There is not any integrated scien- tific research taking into consideration all the con- ceptual frameworks with regard to the ice cream purchase decisions for the target consumers seg- ments in the scientific literature. With the present study focused on integrated marketing approach- es combined with all the factors influencing on the ice cream purchase decision making process of Turkish consumers, it could be filled an impor- tant gap in the literature by contributing consid- erably to the scientific literature. This study, therefore, was designed to reach all the objectives mentioned above. In this scope, the main aims of the study are to explore the core fac- tors related to the intrinsic and extrinsic product attributes impacting on the national-branded ice cream purchase decisions of Turkish consumers; and then to determine the target homogenous con- sumer segments based on their socioeconomic char- acteristics, and finally to analyze the effectiveness of the factors effecting on their consumption amounts. MAterIAL And Methods Material The primary data used in this study which in- clude in the variables based on the national-brand- ed ice cream attributes and Turkish consumers’ socioeconomic characteristics influencing on their consumption decisions and amounts were ob- tained from a face-to face questionnaire in Kah- ramanmaras2 from autumn of 2012 until winter of 2013. The questionnaire was conducted with the heads of the households in their houses. The participants consuming the national-branded ice 2 Kahramanmaras is located in the southeastern part of turkey. the province lies on plain at the foot of the tau- rus Mountains and has a total population of 1.063.174 in 2013, and about 42% of those live in the city centre. table 1 - Annual production, export, import, consumption (million liters) and per capita consumption amounts (liter) of ice cream, and macroeconomic indicators in turkey. Items 2000 2005 2010 2012 Annual change(%) and macroeconomic indicators 2005-2012 Production 60.0 113.0 324.0 260.0 7.1 Export - 4.0 8.0 12.0 8.3 Import - 0.8 0.2 1.5 5.8 Consumption - 99.0 229.0 247.0 7.5 Per capita consumption 1.0 1.5 3.0 4.0 7.8 Population (million) 67.8 72.1 73.7 75.6 0.6 Grow rate of GDP (%) 6.8 8.4 9.2 2.1 2.9 Grow rate of GDP per capita (%) 5.3 7.1 7.5 0.8 1.9 Sources: (TUIK, 2013 and 2013a; ASUD, 2013). 4 Ital. J. Food Sci., vol. 27 - 2015 cream, and accepting the voluntary contribution were selected randomly. They have demographic and socioeconomic characteristics such as average 46 years old, $1282 income, 3.7 liters ice cream consumption, 4.3 family sizes 51% male popula- tion and 61% high school education. Methods Method used in determination of the sample size In order to determine the sample size, while minimizing sample bias and representing cor- rectly the population; the city center was divid- ed into four parts covering the west, east, south and north-sides of Kahramanmaras with 73.929 households consisting of about six family mem- bers (APDK 2013; TUIK 2013). In order to calculate the sample size for each district, the following formula was used (TOPCU et al. 2010). Where n = sample size Z = z value, (1.96 for 95% confidence level) p = percentage making a choice, (0.5 used for sample size needed) c = confidence interval, (used 0.05 = ±5) the minimum sample size having the capabil- ity of representing the main population was cal- culated as 385 households (but the study was conducted with 400 households). by consider- ing the information obtained from the food sci- ence and marketing literature under the expert consultancy and the prior experiences of the researcher, a draft questionnaire was prepared (toPcu, 2012). In order to check out non-sam- pling error which occurs due to ambiguous def- initions, unclear instructions, questionnaire wording, format and length, a pre-test was car- ried out 15 consumers selected randomly in the target regions. the flow and nature of the ques- tionnaire were tested, and the order and timing of the questions were re-arranged. the ques- tionnaire was then modified and refined before starting the fieldwork. Methods used in the preparation of the questionnaires It was asked the participants in the survey to respond to each statement indicating the signifi- cance levels of the ice cream attributes by using a Likert-format with 1-5 scale (where 1 refers to the least important and 5 refer to the most im- portant attribute). of eighteen ice cream attrib- utes, five are related to the intrinsic ice cream attributes (ingredient quality, texture and aro- ma, taste and flavor, hardness and viscosity and organic content); six are covered by the extrinsic food attribute referring the marketing mix (the region of origin, advertisement and promotion, package material and appearance, product as- sortment, quality-price relation and price); two are determined by manufacturing process (the reliability of manufacturing process and food safety and hygiene); five are stated by the rela- tionships between the consumer and marketing environments (brand recognition, store satisfac- tion and confidence, manufacturer brand satis- faction and image). the sources in the literature related to the intrinsic and extrinsic ice cream at- tributes and their items were showed in table 2. on the other hand, of the two string and three numeric variables referring to the consumer de- mographic and socioeconomic characteristics, three referred to the consumer demographic characteristics (age as a numeric variable; con- sumer education (0: primary and 1: high school and college graduate) and the occupation (0: oth- ers and 1: officer) as string variables); two in- cluded in the consumer socioeconomic charac- teristics (monthly consumer income ($) and the share of the ice cream expenditure within total food one as numeric variable). Methods used in the statistics analyses After editing and coding, the primary data were first used in Principal Component Analy- sis (PCA)3 to determine the main factors related to the product attitudes influencing on the na- tional ice cream purchase patterns of turkish consumers. PCA is a data reduction technique that reduces the number of variables used in an analysis by creating new variables (called fac- tors) that combine redundancy in the data (sPss 15.0 2006). the first step in PCA is to determine the number of relevant factors. this was con- ducted by PCA using Varimax Rotation Method (VRM)4. PCA was used initially to identify under- lying aspects explaining a correlation among a set of the food product attributes. the purpose of PCA was to identify those attributes account- ing for a relatively large proportion of the vari- ance in the sample. In the second and final steps of the statis- tics analyses, the main factors obtained from PCA were used for k-means cluster and multi- ple regression/correlation MRC Analyses, re- spectively. In the second step, according to 3 A factor extraction method used to form uncorrelated lin- ear combinations of the observed variables. the first com- ponent has the maximum variance. successive compo- nents explain progressively smaller portions of the var- iance and are all uncorrelated with each other. PCA is used to obtain the initial factor solution. It can be used when there is a single correlation matrix. 4 this method is an orthogonal rotation method that mini- mizes the number of variables that have high loading on each factor. It simplifies the interpretation of the factors. Ital. J. Food Sci., vol. 27 - 2015 5 table 2 - the sources in literature related to the macro variables and their items. Macro variables Items Sources Texture and aroma SOUKOULIS et al., 2010; CRUZ et al., 2009 OZDEMIR et al., 2005; AIME et al., 2001 Taste and flavor TOPCU and UZUNDUMLU 2012; SOUKOULIS et al., 2010; TURGUT and CAKMAKCI, 2009; ENNEKING et al., 2007 Ice cream content SOUKOULIS et al., 2010; TURGUT and CAKMAKCI, 2009; ALVAREZ et al., 2005 Hardness and viscosity ATSAN and CAGLAR, 2008; KAYACIER and DOGAN, 2006; GRANGER et al., 2005 Ingredient quality KARAMAN et al., 2011; FARHOOSH and RIAZI, 2007; GRANGER et al., 2005 Food safety and hygiene TOPCU and UZUNDUMLU, 2012; GOFF, 2008; ROININEN et al., 1999 Actual image quality TOPCU and UZUNDUMLU, 2012; SIRO et al., 2008; WILDMOSER et al., 2004 Ice cream assortment SOUKOULIS et al., 2010; LANGE et al., 1999; GOFF, 2008 Brand recognition AILAWADI et al., 2011; GEHLHAR et al., 2009; WULF et al., 2005 Store satisfaction TOPCU and UZUNDUMLU, 2009; TOPCU et al., 2008; WULF et al., 2005; GUERRERO et al., 2000 Store confidence TOPCU and UZUNDUMLU, 2009; SOBERMAN and PARKER, 2006; GUERRERO et al., 2000 Brand satisfaction AILAWADI et al., 2011; DIMOFTE et al., 2008; STRIZHAKOVA et al., 2008 Brand image TOPCU, 2012; KUMAR et al., 2009; BRAKUS et al., 2009; LI and HOUSTEN, 2001 Package appearance TOPCU and UZUNDUMLU, 2012; TOPCU and ISIK, 2008; ENNEKING et al., 2007 Promotion mix KARRAY and MARTIN-HARRAN, 2009; TOPCU and ISIK, 2008; LI and HOUSTEN, 2001 Price and quality relation TOPCU and UZUNDUMLU, 2012; AMROUCHE and ZACCOUR, 2009; KUMAR et al., 2009; WOLK and SPANN, 2008; ENNEKING et al., 2007 Region/country of origin REALINI et al., 2013; BATRA et al., 2010; TOPCU et al., 2010; ORTH and FIRBASOVA, 2003 In tri ns ic pr od uc t a ttr ib ut es Ex tri ns ic pr od uc t a ttr ib ut es monthly income levels of turkish consumers, therefore, the target consumers were separat- ed to three homogeneous clusters including in low-income users (less than $500 per month), middle-income users ($500-1250 per month) and high-income users (more than $1250 per month) (toPcu 2012), and then the main fac- tors were allocated to the homogeneous con- sumer clusters based on monthly income lev- els of the target consumers by k-means clus- ter analysis. In the final step, the main factors obtained from PCA were used Multiple Regression/Corre- lation (MRC) Analysis. MRC analysis was used to measure the effects of variable factors deliv- ered from the ice cream attributes and turkish consumers’ socioeconomic characteristics effect- ing on their ice cream amounts. In order to test whether the normal distribu- tion of the main factors delivered from PCA and the socioeconomic variables collected from the consumers exhibited or not was applied the vari- ous transformations techniques, and it was test- ed that the closest distribution to the normal of all the factors provided. on the other hand, the coefficient estimations were estimated by us- ing ordinary least squares (OLS). Individual and group significance of these coefficients were test- ed using t and F tests, respectively. In order to evaluate whether to be any econo- metrical problem among the variables, it was tested the overall multicollinarity and auto-corre- lation problems by considering Variance-Inflating Factor (VIF) and Durbin-Watson d statistics, re- spectively. Multicollinearity among variables was detected by calculating (VIF) (GuJArAtI 2005; sPss 15.0 2006). the analysis techniques have been widely used in many marketing researches with regard to the dairy food product attributes (dAVIes and cLIne, 2005; IshII et al., 2007; bAtrA et al., 2010; cAdenA and boLInI, 2011). SPSS 15.0 statistical software program was used to run the PCA and MRC analyses. MRC mod- el could be written as follows: Icc = f (rPrs, tMnF, Mbst, hQLt, sQLt, eth, Inc, eXP,ocu, edu, AGe, ) Dependent Variable Icc: Monthly national ice cream consump- tion amount per household (l/month) Independent Variables rPrs: retailer prestige tMnF: trust to manufacturer Mbst: Manufacturer brand satisfaction hQLt: hedonic quality sQLt: sensory quality eth: ethnocentrism Inc: Monthly consumer income ($) eXP: the shares of the ice cream expenditures within total food ones ocu: consumer occupation edu: consumer education AGe: consumer age resuLts And dIscussIon Demographic and socioeconomic profiles of the participants The results of descriptive statistics related to the gender, marital status, tasks in the family, edu- cation levels, occupation and age of the household heads, and monthly income, national-branded ice cream consumption and expenditure of the house- holds attaining in questionnaire were indicated in Table 3. The results of the statistics showed that 51.0, 90.2, 47.1, 61.3 and 28.0% and 46.2 years 6 Ital. J. Food Sci., vol. 27 - 2015 of the participant household heads consisted of the male, married, husband, high school gradu- ate and retailer, respectively. On the other hand, monthly average income, ice cream expenditure, ice cream consumption amount and family size of the households were calculated as $1281.9, $48.6, 3.9 l and 4.3 individuals, respectively. The results of PCA related to the ice cream consumption satisfaction Kaiser Normalization (KMO) which compares partial correlation coefficients with observed ones was calculated as 0.87 for the ice cream attrib- utes, and this means that the data set was at a perfect level for the factor analysis since the test score was greater than 0.50 (Table 4). The principal component analysis using VRM grouped the eighteen variables related to the ice cream attributes into the six factors with Eigen- values greater than 1.0, which these factors ex- plained the 71.96% of the total variance. F1 being the first of these factors was explained by 19.00% of the total variance, and consisted of the retailer/ store positioning. F1, therefore, could be called by retailer prestige (RPRS) (TOPCU et al., 2009). Ex- plaining the 12.15% of the total variance, F2 gave us information about the manufacturing process and the used material quality, and thus this fac- tor could be determined by trust to manufactur- er (TMNF) (TOPCU, 2012). Reporting the 10.99% of the total variance, F3 gathered together the variables related to the brand satisfaction, and it could be named as manufactur- er brand satisfaction (MBST) (TOPCU and UZUN- DUMLU, 2012). Referring to the 10.28% of the total variance, F4 could be represented hedonic qual- ity (HQLT) covering the relationship between the product quality and its price (TOPCU, 2012, 2012a). Considering the 10.08% of the total variance, F5 was constituted by the sensory quality attributes of the ice cream, and could be entitled by sensory quality (SQLT) (TOPCU, 2012). Finally, referring the 7.20% of total variance, F6 stated the relationship between the country of origin and the ice cream consumption with manufacturer brand, and thus it could be denominated by ethnocentrism (ETH) (ORTH and FIRBASOWA, 2003). The results of cluster analysis related to the ice cream consumption satisfaction The main factors derived from the PCA and ef- fecting on the ice cream purchase decisions of Turk- ish consumers were separated into three homoge- neous consumers segments through k-means clus- ter according to their income levels including in the low, middle and high-income users (Table 5). table 3 - demographic and socioeconomic characteristics of the participants. Demographic characteristics Frequency Percent Cumulative Gender 400 100.0 Male 204 51.0 51.0 Female 196 49.0 100.0 Marital status 400 100.0 Married 360 90.2 90.2 Not married 40 9.8 100.0 Tasks in the family 400 100.0 Husband 188 47.1 47.1 Wife 212 52.9 100.0 Education level 400 100.0 First school 40 10.0 10.0 High school 245 61.3 71.3 College 115 28.7 100.0 Occupation 400 100.0 White-collar state employ 107 26.8 26.8 Blue-collar state workers 50 12.5 39.3 Businessman 40 10.0 49.3 Retailer 112 28.0 77.3 Pensioner 67 16.8 94.1 Others 24 5.9 100.0 Socioeconomic characteristics Minimum Maximum Mean Std. Dev. Age 23.0 77.0 46.2 12.4 Monthly average income ($)* 190.0 3810.0 1281.9 1448.4 Ice cream consumption amount (l) 1.0 15.0 3.7 2.1 Monthly ice cream expenditure ($)* 2.0 200.0 48.6 36.5 Average family size 1.0 9.0 4.3 2.1 *The prices of the products were converted from Turkish Lira (TL) to US Dollar ($) using the exchange rate on February 25, 2014. The conversion rate used was 2.15 TL/$. Ital. J. Food Sci., vol. 27 - 2015 7 table 4 - Factors and correlated variable loadings related to the ice cream attributes. Variables Factor loadings* F1 F2 F3 F4 F5 F6 Retailer prestige (F1:RPRS ) The ice cream assortment 0.845 0.048 0.032 -0.205 -0.061 0.249 Brand recognition 0.727 -0.343 -0.304 0.198 -0.005 -0.151 Store satisfaction 0.640 0.273 -0.086 -0.362 -0.222 -0.075 Store confidence 0.558 0.121 -0.214 -0.370 -0.051 -0.435 Advertisement and promotion 0.532 0.258 0.340 -0.210 0.070 -0.031 Trust to manufacturer (F2: TMNF) Ingredient quality of the ice cream -0.056 0.879 0.126 0.054 -0.035 0.134 Food safety and hygiene -0.009 0.741 -0.189 0.091 0.080 -0.210 Reliability of manufacturing process -0.284 0.729 0.169 -0.052 -0.092 0.017 Manufacturer brand satisfaction (F3: MBST) Manufacturer brand satisfaction -0.021 0.090 0.828 -0.053 0.058 -0.021 Package material and appearance -0.090 -0.051 0.790 0.021 -0.200 0.091 Manufacturer brand image 0.368 0.220 0.581 0.230 0.450 0.098 Hedonic quality (F4:HQLT) Quality-price relation -0.096 0.177 -0.186 0.765 0.031 -0.081 Product price 0.058 -0.088 0.196 0.745 -0.145 0.139 Sensory quality (F5: SQLT) Texture and aroma 0.112 0.035 0.094 -0.157 0.866 -0.097 Taste and flavor -0.166 0.028 0.038 -0.097 0.735 -0.187 Organic content of the ice cream mix -0.160 -0.172 -0.358 0.024 0.731 0.311 Hardness and viscosity 0.220 -0.202 0.220 -0.409 0.516 0.209 Ethnocentrism (F6: ETH) The region of the origin 0.025 0.070 -0.027 0.098 -0.002 0.932 Eigen-value 3.230 2.449 1.868 1.748 1.713 1.223 Share of explained variance (%) 19.00 14.41 10.99 10.28 10.08 7.20 Cumulative share of explained variance (%) 19.00 33.41 44.40 54.68 64.76 71.96 KMO (Kaiser-Meyer-Olkin) Statistic 0.873 Bartlett’s test of Sphericity (Chi-square, df: 136): 2552.21 (p: 0.000) *Bold numbers indicate the largest loading for each variable. table 5 - Final cluster centres and the number of cases in each cluster. Main Factors Clusters* High-income Middle-income Low-income users (C1)** users (C2)** users (C3)** RPRS (Retailer prestige) -0.266 -0.586 0.991 TMNF (Trust to manufacturer) 0.276 0.256 -0.109 MBST (Manufacturer brand satisfaction) 0.813 -0.375 -0.599 HQLT (Hedonic quality) 0.301 -0.375 0.381 SQLT (Sensory quality) -0.408 0.514 0.105 ETH (Ethnocentrism) -0.332 0.380 -0.414 Number of total cases in each cluster *** 140 187 73 % of total cases in each cluster 35% 47% 18% * Bold numbers indicate the largest final cluster centre scores for each factor. ** According to F statistics, the final cluster center scores were found very importance (p<0.01) *** The total number of the cases (n): 400 8 Ital. J. Food Sci., vol. 27 - 2015 The results of the study showed that high-in- come users of the national-branded ice cream (C1) formed their consumption satisfaction ac- cording to trust to manufacturer (TMNF) and manufacturer brand satisfaction (MBST) by tak- ing into consideration the manufacturer-brand- ed ice cream willingness to buy as an indicator of the main component of the hedonic quality at- tributes. The target homogeneous consumers in the C1, therefore, gave a much more attention to the manufacturer brand satisfaction on their purchase decision and satisfactions, and thus it could be designed/developed the manufacturer- branded ice creams for the target consumers at this segment. The results of the study also indicated that the middle-income users of the national-branded ice cream (C2) focused on the purchase patterns con- structed by a combination of the factors such as the sensory quality (SQLT), trust to manufacturer (TMNF) and Ethnocentrism (ETH). The consum- ers in C2 provided much stronger purchase mo- tivations for the private labels under trust to lo- cal manufacturers by emphasizing the core bene- fits of the ice cream based on the sensory quality attributes. It could be introduced the private la- beled local ice creams for the target segments to the region retailers. The results of the study explained that the low-income users of the national-branded ice cream (C3) tended to buy the ice cream with re- tailer prestige (RPRS) under hedonic quality at- tributes (HQLT). The target consumers in C3 fo- cused on the actual ice cream-imaged purchase decision and satisfactions based on the retailer/ store brands. It could be presented, therefore, the retailer branded ice creams for this segment. The results of MRC analysis with regard to the ice cream consumption amounts VIF values calculated as 1.028 and 1.582 in- dicating the scores between 1.00 and 2.50 deter- mining the acceptable reference range for multi- collinearity problem showed that there was not it. Durbin-Watson d statistics, on the other hand, computed as 2.09 was not located between d u (1.89) and 4-d u (1.65). There, therefore, was no problem related to auto-correlation in the MRC model (KALAYCI, 2005). According to these sta- tistical test results diagnosing the econometrics problems, we could directly use this data set for the MRC model. The determination coefficient (Adj.R2) was cal- culated as 0.80 in the MRC model, this means that all the independent variables explained the 80% of the dependent variable. The OLS estimates of the model coefficients and other statistical meas- urements were presented in Table 6. The results of statistical measurements highlighted that the F- statistic rejecting the null hypothesis that makes all the coefficients equal to zero was calculated as 153.76 (p<0.01). On the other hand, the partial regression coefficients of all the independent var- iables, except for those of RPRS, were statistical- ly found to be meaningful (p<0.00 and p<0.05). Their signs, moreover, were also found in con- formity with economic theory. However, RPRS and ETH were not important statistically (p=0.333 and p=0.384), and thus they were not evaluat- ed for compliance with the economic theory. Ad- ditionally, they had no impact on the manufac- turer/national-branded ice cream consumption amounts, and thus there was an inverse (negative) relationship between ETH and the consumption table 6 - the results of multiple linear regression (Mrc) analysis. n: 400 R2: 0.82 Adj R2: 0.80 F (11, 388): 153.76* dL=1.65 du=1.89 DW dh=2.09 Variables Multiple linear regression model Collinearity statistics Correlations Coefficients a Sd.error t h -value p-value Tolerance VIF Zero-order Partial Part Constant 0.672 1.729 4.437 0.000* - - - - - RPRS 0.023 0.099 0.970 0.333 0.866 1.154 0.196 0.049 0.021 TMNF 0.072 0.102 3.218 0.001* 0.973 1.028 0.011 0.161 0.075 MBST 0.045 0.095 1.996 0.047** 0.893 1.120 0.103 0.098 0.042 HQLT 0.066 0.077 2.503 0.010* 0.958 1.043 0.080 0.126 0.066 SQLT 0.090 0.091 4.045 0.000* 0.975 1.026 0.126 0.201 0.089 ETH -0.019 0.089 -0.872 -0.384 0.968 1.033 -0.004 -0.044 -0.019 INC 0.080 0.511 2.944 0.003* 0.648 1.544 0.331 0.148 0.065 EXP 0.736 0.347 26.689 0.000* 0.632 1.582 0.358 0.305 0.385 OCU 0.062 0.379 2.281 0.018** 0.703 1.422 0.023 0.120 0.052 EDU 0.180 0.395 4.738 0.000* 0.648 1.543 0.081 0.106 0.048 AGE 0.424 0.018 18.958 0.000* 0.690 1.448 0.294 0.101 0.046 a Coefficients consist of the standardized coefficients. *p<0.01 **p<0.05 Ital. J. Food Sci., vol. 27 - 2015 9 amounts of the ice cream. The findings were sup- ported by the similar results of some studies (BR- AKUS et al., 2009; TOPCU et al., 2009; DIMOFTE et al., 2008; STRIZHAKOVA et al., 2008; ENNEK- ING et al., 2007). The results of the study indicated that of the six main factors, the four ones such as SQLT, TMNF, HQLT and MBST had an important effect on the national-branded ice cream consumption amount per household in C2; C1 and C2; C1and C3; and C1, respectively. SQLT was regarded as their main determinant on their satisfaction after consumed, and thus it played an important role in this process for C2 by providing much impor- tant information about the consumer satisfaction associating with the intrinsic ice cream attributes. The results were similar with respect to those of the studies carried out by TOPCU (2012); TOPCU (2012a); TOPCU and UZUNDUMLU (2009); ENNEK- ING et al. (2007); ISHII et al (2007); GUERRERO et al. (2000). The results of the study also showed that there was a positive relationship between TMNF giv- ing much important information about the man- ufacturing process to Turkish consumers, reduc- ing their health concern related to the ice cream based on the manufacturer confidence construct- ed with its ingredient quality at the manufacturing process under hygienic conditions and their con- sumption decision and amounts. This caused the consumers to increase dramatically the ice cream consumption trends in C1 and C2 due to the safe and hygienic ice cream manufactured by the man- ufacturers. This findings were quite similar to the results reported by several researchers (TOPCU, 2012; TOPCU and UZUNDUMLU, 2012; TOPCU et al., 2010; SOBERMAN and PARKER, 2006; SCHU- ILING and KAPFERER, 2004). The results of the present study also report- ed that there was a string relationship between HQLT having a bigger impact effect on the con- sumer demands making possible the product differentiation through the quality and price rat- ing in C1 and C3, and their purchase decisions. The marketers could, therefore, separate tar- get Turkish consumer masses into two homog- enous segments, and then they could also stim- ulate much more effective marketing tactic and strategies for each segments (WOLK and SPANN, 2008; KUMAR et al., 2009; TOPCU et al., 2010; ZHOU et al., 2010). The results of the study also revealed that MBST provided attitudinally a positive motivation on the national-branded ice cream consumption amounts in C1, and it had a string relationship between TMNF and HQLT of the manufacturer-branded ice cream. As a result, MBST was of a strong lin- ear relationship with these factors influencing on not only the ice cream purchase decision at the sale points but also its consumption satisfaction and amounts after consumed. There were a lot of the studies referring to the relationship between MBST and the food consumption amounts (TOP- CU, 2012, TOPCU, 2012a; TOPCU and UZUNDUM- LU, 2012; ZHOU et al., 2010; AMROUCHE and ZAC- COUR, 2009; STRIZHAKOVA et al., 2008). The results of this study also provided the im- portant information about how the demograph- ic and socioeconomic characteristics of Turk- ish consumers affected their purchase decision and attitudes towards the national-branded ice cream. In this research considering their distinc- tive characteristics, the results of the study also referred that the share of the ice cream expend- iture within total food ones (EXP), age (AGE), education (EDU), income (INC) and occupation (OCU) of Turkish consumers had a much bigger effect on the ice cream consumption amounts than the other preference factors, especially, EXP, AGE, EDU. These findings were support- ed by results of previous researches based on the consumers purchase attitude and behaviors towards the food products (TOPCU, 2012; TOP- CU and UZUNDUMLU, 2012; TOPCU et al., 2010; ZHOU et al., 2010; TOPCU et al., 2009; DIMOFTE et al., 2008; ENNEKING et al., 2007; LI and HOU- STON, 2001). concLusIons In this study, the integrated approaches pat- terns based on not only the intrinsic and extrin- sic attributes of the national-branded ice cream but also the socioeconomic characteristics of Turkish consumers impacting on their ice cream purchase decisions and consumption amounts were evaluated. The measurement results of the study highlighted clearly that the consumers in C1, C2 and C3 satisfied with the actual manu- facturer brands linking between trust to manu- facturer and the hedonic quality attributes, the individual private brands under trust to local manufacturers and the sensory ones, and the re- tailer brands emphasizing the hedonic ones on the ice cream purchase decision and consump- tion amounts, respectively. The manufacturers, retailers and marketers of the ice cream, there- fore, should widely implement the manufacturer brand, the individual private label and the retail- er brand for C1, C2 and C3, respectively in order to be able to create the demand trend increases under the integrated marketing tactic and strat- egies affecting positively the ice cream purchase decisions and consumption amounts of the tar- get consumer segments. Although this study has some scientific merit for the academic and food manufacturing commu- nities, there are some limitations. The results of this study have a limited generalizability since the data were obtained from only one city. If the survey is conducted nationally, more data will give more objective results about the purchase decisions of all the population. In future studies, furthermore, this model could be expanded to incorporate more factors and factor levels into the model. 10 Ital. J. Food Sci., vol. 27 - 2015 reFerences Ailawadi K.L., Neslin S.A. and Gedenk K. (2011). Pursuing the value-conscious consumer: Store brands versus na- tional brand promotions. The Journal of Marketing 65 (1): 71-89. Aime D.B., Arntfield S.D., Malcolmson L.J. and Ryland D. (2001). Textural analysis of fat reduced vanilla ice cream products. Food Research International 34: 237-246. Aldazabal J, Martín-Meizosoa A, Martínez-Esnaolaa J.M. and Farr R. (2006). Deterministic model for ice cream solid- ification, Computational Materials Science 38 (1): 9-21. Alvarez V.B., Wolters C.L., Vodovotz, Y. and Ji T. (2005). Physi- cal properties of ice cream containing milk protein concen- trates. Journal of Dairy Science 88 (3): 862-871. Amrouche N. and Zaccour G. (2009). A shelf-space-depend- ent wholesale price when manufacturer and retailer brands compete. OR Spectrum 31: 361-383. APDK (2013). Annual report, the data of Agricultural Provin- cial Directorate of Kahramanmaras, Turkey. ASUD (2013). Ice cream-ice cream economy, Packed Dairy Products Industry Association (ASUD). Available: http:// www.asuder.org.tr/dondurmaekonomisi.html Atsan E. and Caglar A. (2008). The effect of different levels of stabilizers on some physical and organoleptic parame- ters of icecream. Journal of the Faculty of Agriculture 39 (2): 195-200. Batra R, Ramaswamy V., Alden D.L., Steenkamp J.E.M and Ramachander S. (2010). Effects of brand local and nonlo- cal origin on consumer attitudes in developing countries. Journal of Consumer Psychology 9 (2): 83-95. Brakus J.J, Schmitt B.H. and Zarantonello L. (2009). Brand experience: What is it? How do we measure it? And does it affect loyalty? Journal of Marketing 73 (3): 1-51. Cadena R.S. and Bolini H.M. (2011). Time-intensity analy- sis and acceptance test for traditional and light vanilla ice cream. Food Research International 44: 677-683. Cruz A.G., Antunes, A.E.C., Sousa A.L.O.P, Faria J.A.F. and Saad S.M.I. (2009). Ice-cream as a probiotic food carrier. Food Research International 42: 1233-1239. Davies A. and Cline T.W. (2005). A consumer behavior ap- proach to modelling monopolistic competition. Journal of Economic Psychology 26: 797-826. Dimofte C.V., Johansson J.K. and Ronkainen I.A. (2008). Cognitive and affective reactions of American consum- ers to global brands. Journal of International Marketing 16 (1): 1-43. EICA (2013). Ice cream market report, European Ice Cream Association. Available: http://www.euroglaces.com. Enneking U., Neumann C. and Henneberg S. (2007). How im- portant intrinsic and extrinsic product attributes affect pur- chase decision. Food Quality and Preference 18: 133-138. Farhoosh and Riazi (2007). FAOSTAT (2013). Food and Agriculture Organization of the United Nations (FAOSTAT): Production, consumption and trade. Available: http://faostat.fao.org/site. FIIND (2013). New Zealand-world master in ice cream con- sumption. Available: http://www.food.industryindia.com FOOD (2013). Overview of Turkish ice cream market. Availa- ble: http://www.21food.com Fornerino M. and Hauteville F. (2010). How good does it taste? Is it the product or the brand? A contribution to brand eq- uity evaluation. Journal of Product and Brand Manage- ment, 19 (1): 34-43. Gehlhar M.J., Regmi A., Stefanou S.E. and Zoumas B.L. (2009). Brand leadership and product innovation as firm strategies in global food markets. Journal of Product and Brand Management 18 (2): 115-126. Goff H.D. (2008). 65 years of ice cream science. International Dairy Journal 18: 754-758. Granger C., Leger P., Barey P., Langendorff V. and Cansell M. (2005). Influence of formulation on the structural networks in ice cream. International Dairy Journal 15: 255-262. Guerrero L., Colomer Y., Guardia M.D., Xicola J. and Clotet R. (2000). Consumer attitude towards store brands. Food Quality and Preference 11: 387-395. Gujarati D (2005). Basic Econometrics. ISBN: 975-7860-99, Ref.-Publish: 33, Istanbul. Ishii r., Kawaguchi h., Mahony M.o. and rousseau b. (2007). relating consumer and trained panels’ discrimi- native sensitivities using vanilla flavored ice cream as a medium. Food Quality and Preference 18: 89-96. Kalayci s. (2005). sPss Applications, Multi-Variables sta- tistics techniques, Isbn: 975-9091-14-3, First edition, Asil Publish, Ankara. Karaman s., Yilmaz M.t. and Kayacier A. (2011). simplex lattice mixture design approach on the rheological behav- ior of glucomannan based salep-honey drink mixtures: An optimization study based on the sensory properties. Food hydrocolloids 25 (5): 1319-1326. Karray r. and Martin-harran A. (2009). A composition- al study on two current types of salep in Iran and their rheological properties as a function of concentration and temperature. Food hydrocolloids 21: 660-666. Kayacier A. and dogan M. (2006). rheological properties of some gums-salep mixed solutions. Journal of Food en- gineering 72: 261-265. Kumar A., Lee h.J. and Kim Y.K. (2009). Indian consum- ers’ purchase intention toward a united states versus local brand. Journal of business research 62: 521-527. Lange c. rousseau F. and Issanchou s. (1999). expectation, liking and purchase behaviour under economical con- straint. Food Quality and Preference 10: 31-39. Li h.s. and houston J.e. (2001). Factors affecting consum- er preferences for major food markets in taiwan. Journal of Food distribution research 18 (1): 97-109. Mccarthy, M., boer, M., o’reilly, s. and cotter, L. (2003). Factors influencing intention to purchase beef in the Irish market. Meat science, 65, 1071-1083. Menichelli e., olsen n.V., Meyer c., and næs t. (2012). combining extrinsic and intrinsic information in con- sumer acceptance studies. Food Quality and Preference, 23 (2): 148-159. MGMn, 2013. World of ice cream-repositioning strate- gies, Mintel Global Market navigator (MGMn). Availa- ble: http://www.frozenfoodeurope.com/article/markets orth u.r. and Firbasova, Z. (2003). the role of consumer ethnocentrism in food product evaluation. Agribusiness 19 (2): 137–153. ozdemir c., demirci M., ozdemir s. and sagdic, o. (2005). Production and some properties of yoghurt ice cream in turkey. Journal of nutrition research and Food science 60 (4): 353-472. realini c.e., Furnols M.F., sañudo c, Montossi F., oliver M.A. and Guerrero L. (2013). spanish, French and brit- ish consumers’ acceptability of uruguayan beef, and con- sumers’ beef choice associated with country of origin, finishing diet and meat price. Meat science 95: 14-21. roininen K., hteenmaki L.A. and tuorila h. (1999). Quan- tification of consumer attitudes to health and hedonic characteristics of foods. Appetite 33: 71-88. schuiling I. and Kapferer J.n. (2004). real differences be- tween local and international brands: Implications for in- ternational marketers. Journal of International Market- ing 12 (4): 97-112. simeone, M. and Marotta, G., (2010). towards an integra- tion of sensory research and marketing in new food prod- ucts development: A theoretical and methodological re- view. African Journal of business Management, 4 (19): 4207-4216. siro I., Kapolna e., Kapolna b. and Lugasi A. (2008). Func- tional food. Product development, marketing and consum- er acceptance: A review. Appetite 51: 456–467. soberman d. and Parker P.M. (2006). the economics of qual- ity-equivalent store brands. International Journal of re- search in Marketing 23: 125-139. soukoulis c., Lyroni e. and tzia c. (2010). sensory profiling and hedonic judgment of probiotic ice cream as a func- tion of hydrocolloids, yogurt and milk fat content. Food science and technology 43: 1351-1358. Ital. J. Food Sci., vol. 27 - 2015 11 sPss base 15.0 (2006). sPss base 15 user’s Guide 11, p. 161. chicago, IL. Strizhakova Y., Coulter R.A. and Price L. (2008). Branded products as a passport to global citizenship: Perspective from developed and developing countries. Journal of In- ternational Marketing 16: 1-52. Topcu Y. and Isik H.B. (2008). Topcu Y., Isik H.B. and Uzundumlu A.S. (2008). Manufacturing brands vs. store brands: A case of liquid vegetable oil mar- ket. In: Proceedings of International Conference on Busi- ness Management and Economics, 95-115, 26-28 Janu- ary, Erzurum, Turkey. Topcu Y., Isik H.B. and Uzundumlu A.S. (2009). Turkish con- sumer attitudes toward food products: The case of Erzu- rum. Italian Journal of Food Science 21 (1): 37-50, Topcu Y. and Uzundumlu A.S. (2009). Analysis of factors af- fecting customer retailer loyalty in Turkish food market: The case of Erzurum. Italian Journal of Food Science 21 (2): 157-169. Topcu Y., Turhan B. and Uzundumlu A.S (2010). Analysis of factors affecting Turkish sunflower oil consumer behaviour. Italian Journal of Food Science 22 (2): 239-247. Topcu Y. and Uzundumlu A.S. (2012). Turkish consumers’ purchase attitude and behaviors towards Kahramanmaras type ice cream as a local branded product. African Journal of Business Management 6 (34): 9695-9703. Topcu Y. 2012. The effects of the local products on the rural de- velopment under societal marketing orientation: the case of Erzurum Civil cheese. 10th National Agricultural Economics Congress. pp. 916-926. 5 September 2012. Konya, Turkey. Paper Received November 29, 2013 Accepted April 26, 2014 topcu, Y. (2012a). rural development-contemporary Issue and practices, In. r.s. Adisa (eds.), the integrated mar- keting approach as a rural development tool. Intech-open Access Publisher, Isbn: 978-953-307-942-4, 257-282. troy, d.J. and Kerry, J.P. (2010). consumer perception and the role of science in the meat industry. Meat science, 86, 214-226. tuIK (2013). Agricultural production statistics, turkey sta- tistical Institute. Available: http://report.tuik.gov.tr/re- ports/consumption tuIK (2013a). 2013 turkey population census, turkey sta- tistical Institute. Available: http://report.tuik.gov.tr/re- ports/rwservlet?adnksdb2 turgut t. and cakmakci s. (2009). Investigation of the pos- sible use of probiotics in ice cream manufacture. Inter- national Journal of dairy technology 62 (2): 444-451. Wildmoser h. scheiwiller J. and Windhad e.J. (2004). Impact ofdisperse microstructure on rheology and quality aspects of ice cream. Lebensm.-Wiss. u.-technology 37: 881-891. Wolk A. and spann M. (2008). the effects of reference prices on bidding behavior in interactive pricing mechanisms. Journal of Interactive Marketing 22 (4): 3-18. Wulf K., odekerken-schröder G., Goedertier F. and ossel, G.V. (2005). consumer perceptions of store brands ver- sus national brands. Journal of consumer Marketing 22 (4): 223-232. Zhou L., Yang Z. and hui M.K. (2010). non-local or local brands? A multi-level investigation into confidence in brand origin identification and its strategic implications. Journal of the Academic Marketing science 38: 202-218.