The section below concentrates on presenting the conceptual models and research questions that have been formulated to test the models in this research. As this research is a replication and extension of the research of Brakus et al. (2009) the measurement model was based on their four-factor brand experience model. The BBX scale measures brand experience on four dimensions: sensory, affective, intellectual, and behavioral.
However, the scale does not include any reference or items on eco-friendliness.
The first conceptual model in this study replicates directly the four-factor BBX model.
The main research question related to the four-factor model is whether the brand experience of high-tech products, and especially mobile phone brands, consists of the same four brand experience dimensions presented in the BBX model and is the model sufficient for analyzing the data collected in this research without the eco-friendliness items. The original items in the BBX measurement scale are presented below in table 2.
Table 2. The original items from the four-factor BBX model
Sensory
This brand makes a strong impression on my visual sense or other senses.
I find this brand interesting in a sensory way.
This brand does not appeal to my senses.
Affective
This brand induces feelings and sentiments.
I do not have strong emotions for this brand.
This brand is an emotional brand.
Behavioral
I engage in physical actions and behaviors when I use this brand.
This brand results in bodily experiences.
This brand is not action oriented.
Intellectual
I engage in a lot of thinking when I encounter this brand.
This brand does not make me think.
This brand stimulates my curiosity and problem solving.
In addition to the original twelve items of the BBX model, four items on eco-friendliness were designed for the survey, one for each of the four dimensions, the sensory, affective, behavioral and intellectual dimensions. In the two extended conceptual models developed in this study, the four items on eco-friendliness were included in original BBX model Two variations of an extended model were constructed: one had the items on eco-friendliness embedded in the original four dimensions, and in the other extended model a fifth new dimension having only items on eco-friendliness was constructed.
The eco-friendly items are listed below.
• Sensory: This brand makes an eco-friendly impression. (eco-friendly = not environmentally harmful)
• Affective: This brand creates eco-friendly emotions
• Behavioral: This brand makes me behave in an eco-friendly way
• Intellectual: This brand makes me think about the state of the environment The reason why a statement on eco-friendliness was formulated for each of the four dimensions from the original four-factor BBX model was to enable the testing of whether the eco-friendly attribute is associated with and incorporated in the existing four brand experience dimensions or not, or whether it is a totally separate dimension of its own. As it had already been attested with the original BBX model that brand experience is measurable on four dimensions, it was important to see if eco-friendly attributes would fit the same conceptual model. In this study, the items on eco-friendliness were also formulated following the division into four dimensions following the logic of the original BBX model.
The research questions designed to test the three operational models are the following:
1) Can the original four-factor BBX model be replicated with a data set on high-tech brands collected from Finland?
2) Is the eco-friendliness dimension embedded in the four-factor BBX model?
3) Is the eco-friendliness dimension a separate fifth dimension requiring that the original four-factor model is extended into a five-factor model?
Figure 1. Conceptual models tested in this study
3.2 Research Design
The design of the research starts by explaining how the sampling has been done, how the data was collected and how well the sample represents the whole population. In association with the description of the data collection methods, the development of the questionnaire is also covered.
3.2.1 Sampling
The target of the sampling for the actual online survey was to have the respondents represent a wide consumer group who actively use and buy high-tech products, or in this case mobile phones. The mobile phone has become a permanent commodity in peoples’ lives and the usage covers all age groups in nearly all of the households in Finland. The official national statistics on consumers indicated that in August 2013 of all households in Finland 84% had only a mobile phone, 15% of the households had both a mobile phone and a landline phone, and only 1% of all of the households had solely a landline phone. (Official_Statistics_of_Finland, 2013a).
The data was collected in the form of an online web survey in autumn 2013. The respondents were approached with a market research company via an internet-based sampling frame to ensure that a representative sample of population is included in the survey. The age of the respondents ranged from 18-64 years. The data extracted and analyzed for the purpose of this study consisted of 506 respondents from Finland. The place of residence of the respondents was also tracked to ensure that the respondents were distributed evenly across the country similarly as the whole population of Finland is distributed. The representativeness of the sample was validated against the age distribution of the present population in Finland with reference to their place of residence based on the official statistics of Finland maintained by the Statistics Finland organization, which is the Finnish public authority specifically established for statistics.
3.2.2 Data Collection
The data collection method in this consumer research was a web survey that was sent out in an e-mail to 4681 people living in Finland. The survey was open for the respondents from the 19th of Sep to the 4th Oct 2013 in Finland. As a token of appreciation, all the respondents had the possibility to participate in a lottery with a prize of 100€.
The survey questionnaire was done on the basis of the original questionnaire used in the research of Brakus et al. (2009). One clarifying addition was made to the three statements under the sensory dimension: the five senses were listed in parentheses at the end (sight, touch, hearing, taste, and smell) to ensure that this statement was fully understood by the respondents and could also be translated accurately into another language. In addition to the original twelve statements of the BBX model, four additional statements on eco-friendliness were designed for the survey. All of the statements were repeated for five major mobile phone brands used globally: Nokia, Samsung, Apple, Sony and hTc. In the actual survey, the order of the questions for the five brands was random and rotated in order to avoid any ordering effects in the responses.
The questionnaire was pretested with a small group (N=7) of respondents representing the middle-age category in the target age group (25-54 years) with varying educational backgrounds, different nationalities (Finnish and Indian), and four of the respondents were male and three were female. In the pilot survey, the response options were according to the 7-point Likert scale and the options were anchored only at the ends of the options, at the response options 1 and 7. When answering the pilot survey, one of the respondents requested for a new copy of the questionnaire as the person did not understand the ratings in the scale especially for the negative statements that the person wanted to disagree with. Also for one respondent some of the answers were not logical when comparing the negative statements to the positive statements. On the basis of the results of the pilot survey, some of the statements on eco-friendliness were also slightly reformulated and all of the numerical response options were anchored to a worded response. The anchoring of all of the response options clarifies the response options and reduces the possibility to misinterpret the numerical options and thus enables the respondents to find the correct answer in a more reliable way.
In the pilot survey, a few of the respondents answered with the rating 4 to all of the statements for some brands which was interpreted so that this person probably did not know these brands. The neutral option seems to have been used by some of the respondents as an option to indicate that the respondent does not have any experience of the brand and really does not know what else to answer. Based on this response style, an eighth option “Do not know” was added in the response options. The “Do not know”
option has also been included in brand image measurement surveys (Dolnicar and Grun,
2014). In the online web survey, the respondents are not given the possibility to skip a question or a brand, instead the option “Do not know” needs to be selected in order to be able to proceed to the next brand which is on the next page of the survey. This eighth option also allows the respondent to be honest about their experience and to avoid conflicting statements that came up in the pilot when one of the respondents had selected the option “I do not use the brand”, but has still added a rating for the statement
“I engage in physical actions and behaviors when I use this brand.”
In the actual survey, all the numerical options of the 7-point Likert scale were anchored to a written response, and also there was an 8th option “Do not know”. The following wordings were used to anchor the numerical options: 1= Strongly disagree, 2= Disagree, 3= Somewhat disagree, 4= Neither disagree nor agree, 5= Somewhat agree, 6= Agree, 7= Strongly Agree, 8= Do not know. The eighth option “Do not know” helps respondents to go on answering the questions even though they do not have the kind of experience called for in the statement. The option 4 in the 7-point Likert scale worded as “Neither disagree nor agree” option is a neutral response, which can also indicate that the person has some kind of an experience of the brand, but that s/he does not have a strong stand to the statement. On the other hand, when a respondent answers that s/he does not know something, it is a clear indication that the person does not at have the experience and is not able to evaluate whether the statement applies.
The questionnaire and response options were translated into Finnish by an experienced translator, and to ensure translation equivalence the translation was checked by a qualified English teacher at the Tampere University of Technology. The e-mail survey was sent to 4681 people, of which 814 recipients opened the questionnaire, and 506 completed the questionnaire. The response rate calculated from all of the invitations sent to participate in the survey is 10.8%. From all the opened questionnaires, 62.2%
were completed. The share of Finland’s population aged 16 to 74 that had used the Internet in 2013 in the past three months was 92% of the whole population (Official_Statistics_of_Finland, 2013b).
Overall the response representativeness is considered to be more relevant than the response rate in survey research, and the response rate is only important if it has an effect on the representativeness (Cook et al., 2000). For certain populations, e-mail and Web surveys may have only minor coverage problems. Response rates for e-mail surveys tend to be lower than for traditional mail surveys (Cook et al., 2000). Some individuals have reached a point where they do not read all their e-mail messages and they may resist to being reminded about a survey, if they receive too many reminders (Cook et al., 2000) The response rate results of offline surveys cannot be generalized for online surveys (Deutskens et al., 2004). It is worthwhile conducting even long and elaborate surveys over the internet, if respondents are adequately rewarded (Deutskens et al., 2004). The types of population participating in the survey has an impact on the variation of response rate differences between Web and mail surveys, so that for college
populations, Web survey response rates can be higher than for mail surveys, and for other populations (e.g., professionals, employees, and general population), Web survey response rate can be lower than the mail survey response rate (23%, 10%, and 13%
lower, respectively, for the three population types) (Shih and Fan, 2008). The meta-analysis of Web and mail survey modes done by Shih and Fan (2008) indicates that for Web surveys the response rates are about 10% lower on the average than for mail surveys and that the population types can statistically account for some of the variation between the Web and paper survey response rate differences. From the 39 web surveys analyzed by Shih and Fan (2008) 10 had a response rate of under 15%, which is one fourth of all the analyzed surveys. Based on this, it can be stated the response rate for the current survey was good enough.
3.2.3 Representativeness of the Sample
The representativeness of the sample is validated against the distribution of gender and age of the present population in Finland. Table 3 shows how the respondents are distributed demographically according to gender. To verify that the data set is a representative sample of the whole population, the corresponding figures for the demographics of the whole country are listed in the table beside the figures of the data set.
In the sample 46.6% were females and 53.4 were males. The target would have been to have close to 50% - 50% ratio. However, based on the results of binominal tests it can be said that the sample is representative for gender (See Table 3).
Table 3. The distribution of genders in the survey
Gender Respondents Population of Finland
in 2013
Female 236 46.6 % 50.8 %
Male 270 53.4 % 49.2 %
Total 506 100 % 100 %
Binominal test: Exact Sig. 1-tailed = .034
The distribution of the respondents in the age groups represents the age distribution among the whole population of Finland ideally, the very small differences range between 0,21% - 0,8% (Official_Statistics_of_Finland, 2014a). The fit of the sample in comparison with the overall population of Finland was very good as the residuals indicate in the Table 4 below.
Table 4. The distribution of different age groups in the survey in comparison with the population of Finland in 2013
Age
Data set
N %
Population in Finland
Residual
18-24 68 13,4 % 14,2 % - .8
25-34 104 20,6 % 20,8 % - .2
35-44 102 20,2 % 19,8 % .4
45-54 116 22,9 % 22,4 % .5
55-64 116 22,9 % 22,8 % .1
Total 506 100 % 100 %
The respondents are from four different regions in Finland and the coverage corresponds to the distribution of the population in Finland (Official_Statistics_of_Finland, 2014b) fairly well as the differences are ranging from 2,4% - 6,7% as indicated by the residuals as presented in Table 5 below.
Table 5. The distribution of respondents according to their location in comparison with the population of Finland in 2013
Region in Finland
Data set Population
in Finland
Residual
Helsinki area 150 29,6 % 25,3 % 4.3
Rest of Southern Finland 114 22,5 % 29,2 % - 6.7
Western Finland 132 26,1 % 21,4 % 4.7
Northern and Eastern Finland 110 21,7 % 24,1 % - 2.4
Total 506 100,0 % 100 %
3.2.4 General background on the respondents
All of the respondents replied to the same questions for the five mobile phone brands in the survey. Three brands were selected to be analyzed in this study from the five brands included in the web survey, due to the fact that the majority of the respondents, 65.8%, owned a Nokia/Lumia branded mobile phone, a Samsung mobile was owned by 17.4%,
and an Apple branded mobile phone as owned by 9.3% of the respondents. Only 6.3%
of the respondents had a mobile phone of another brand and there were several brands in this group of respondents. Also the idea was to concentrate on comparing brands that represent the biggest mobile phone Operating Systems (OS) on the market according to market share in 2013: Samsung has an Android OS with 78% market share, Apple has iOS with 17.5% market share and Nokia Lumia has Windows Phone OS with 3%
market share (IDC, 2014).
Table 6. Smartphone OS Market Share in 2014(IDC, 2014)
Period Android OS iOS Windows
Phone OS
BlackBerry
OS Others
Q4 2014 76.60% 19.70% 2.80% 0.40% 0.50%
Q4 2013 78.20% 17.50% 3.00% 0.60% 0.80%
The data set consists of the combined responses of 506 respondents to the same questions on three different brands (Nokia, Samsung, Apple) summing up to a combined set of 1518 responses in total. By analyzing the combined responses to all of the three brands, the effect of the brand bias was reduced to a minimum and the result can be considered to represent a more general brand experience. Also separate CFAs were conducted individually for each of the brands, so that the 506 responses for Nokia, 506 responses for Samsung, and 506 responses for Apple were analyzed separately to verify that the brand experiences of the individual brands are also in line with the general brand experience.
3.2.5 Missing Data
All the answers with the option “Do not know” were coded as missing data in the data set. Below Table 7 lists the ratio of missing data for all of the responses. The data consists of the responses for all of the three brands (N= 3x 506=1518). Only one of the questions has a missing value for over 25% of the responses, the negatively worded statement on the behavioral dimension has missing values for 28,3% of the responses.
Otherwise, the share of missing values is 11-22% for the rest of the items, the mean being 15.7%.
Table 7. Missing data per item
Item in the measurement model N
Valid N Missing
% Missing ThreeBrand_1-a- This brand induces feelings and sentiments. 1341 177 11,7 ThreeBrand_2-I- This brand makes me think about the state of the
environment.
1279 239 15,7
ThreeBrand_3-i- This brand stimulates my curiosity and problem solving. 1289 229 15,1 ThreeBrand_4-s- I find this brand interesting in a sensory way. (sight, touch,
hearing, taste, and smell).
1299 219 14,4
ThreeBrand_5-a- This brand is an emotional brand. 1331 187 12,3
ThreeBrand_7-i- This brand does not make me think. 1302 216 14,2
ThreeBrand_8-i- I engage in a lot of thinking when I encounter this brand. 1292 226 14,9
ThreeBrand_9-b- This brand is not action oriented. 1089 429 28,3
ThreeBrand_10-s- This brand does not appeal to my senses. (sight, touch, hearing, taste, and smell).
1265 253 16,7
ThreeBrand_11-S- This brand makes an eco-friendly impression. (eco-friendly
= not environmentally harmful)
1177 341 22,5
ThreeBrand_12-a- I do not have strong emotions for this brand. 1350 168 11,1 ThreeBrand_13-A- This brand creates eco-friendly emotions. 1230 288 19,0 ThreeBrand_14-s- This brand makes a strong impression on my visual sense or
other senses. (sight, touch, hearing, taste, smell).
1286 232 15,3
ThreeBrand_15-b- This brand results in bodily experiences. 1251 267 17,6 ThreeBrand_16-B- This brand makes me behave in an eco-friendly way. 1272 246 16,2 ThreeBrand_17-b- I engage in physical actions and behaviors when I use this
brand.
1232 286 18,8