Although the quality of beef is differently interpreted among consumers, nutritional as objective entity and eating quality aspect of quality have been recorded as the most important dimensions in user based quality approach [9] [16] . This has been described based on the human factors that optimize selection of cuts in order to fit dietary and palatability needs of consumers. The categorization of consumer needs in beef is based on beef cleanliness, estimated yields, eating quality and cuts. Although this categorization has been used as an important selling factor in Australia, Europe, Asia and North America, it is too unclear to be applicable in beef trading across different regions [16] .

Quality in user-based dimension is determined by one among the four main approaches, i.e., economics of information, multi-attribute, hierarchical, and integrative [16] . In economics of information the subjective quality of a product is determined based on the economic theory on product quality during which a major distinction between search, experience and credence characteristics is made [16] . The weakness in this approach is that the characteristics are neither visible nor validated by consumers even after trying the product.

On the other hand, the multi-attribute approach of quality determination for a product is considered as a multi-dimensional evaluation, during which, its overall quality is described by a set of cues that are perceived by the consumer [17] . In this approach the interrelationship of attributes is not taken into account, i.e., all the attributes are treated at the same level. For example, consumer may either infer taste from price or healthiness from fat content [16] . In addition to that the importance of the attributes is assumed to be constant although it may be dependent on purchase situations, e.g., weighting of taste and convenience may be different for weekday and weekend use.

The next one is hierarchical approach with means-end chain model as a well-established model. This approach describes the means at which consumers infer some attributes from other attributes. The approach is elaborated by the means-end chain model that describes the links, which a consumer establishes between product perceptions and abstract purchase motives or values. For example, a consumer will inspect the color of a piece of meat (a product characteristic) because he believes it to be related to the taste of meat when prepared (expected quality), and the taste will lead to enjoyment while eating (abstract purchase motive). The product attribute (in and by itself) is only relevant to the extent that the consumer expects to lead to one or more desirable or undesirable consequences [17] .

The last one is integrative approach that focuses on the determinants of experienced as opposed to expected quality and the technical product specifications, which will determine both the intrinsic quality cues the consumer can perceive and the quality finally experienced. The approach is silent on credence attributes like safety, environmental quality, and health although they are recognized as important in food choice motivations. This study draws data from consumers’ perceived beef quality as users of the product. Such users were found at points of sale of beef (butchers).

In connection to this, other literature reported the means in which the described inaccuracy in determination of the quality of products is corrected by using the concept of usability [18] [19] . The quality of the product in user-based dimension is inaccurately measured following the inclusion of consumers’ perceptions, needs and goals in its definition [20] [21] [22] [23] . This concept is embracing the user-perceived quality by connecting the quality to the needs of the user of an interactive product and the quality of use [18] [23] . Quality of use is an extent to which a product satisfies the stated and implied needs when used under the given conditions [19] . This moves the focus of quality from the product in isolation to the particular users of the product, the tasks and the context in which it is used [18] . Since the purpose of the noted product is to help the user to achieve particular goals, the measures of quality of use can be termed as quality of use measures, i.e., the effectiveness, efficiency and satisfaction based on which users can achieve specified goals in different environment [19] [22] .

The determination of beef quality in user-based dimension involves the optimum needs and motivations based on economic and deeper levels involving emotions, cultural norms and values, and group affiliations [20] . In connection to this, it was highlighted that the term optimum in beef quality should not be considered as a single unchangeable state, but it depends on the end use of the meat and even the way the meat is cooked. For, example, the popular beef based cuisine in different continents as reported in literature were observed to be similar for grilled, deep-fried and stewed beef, but differences arise among different regions on case of smoked, raw soup and burger uses as shown in Table 2. Thus, consumer preferences are also dependent on expected uses and preparation.

User-based approach is connected to the fact that, any product can be conceived as an array of its related cues during which, each cue provides a basis for various impressions [1] [4] [18] . The quality aspects that are unobservable at purchasing point are known as quality attributes and normally linked to the product attributes [1] [13] . They are sometimes known as quality criteria that are functional and psychological benefits or consequences provided by the

product, and they are observable prior to consumption [1] . Which product attributes are quality attributes thus depends on the priorities of the individual based on one among these and or all the following criteria, i.e., satisfaction, worries and trust... The quality attributes are described as experience quality attributes, i.e., those ascertained on the basis of actual experience on consumption of a product, e.g., in case of beef we have tastes, tenderness and leanness [13] . The quality attributes are also described as credence quality attributes, the one that cannot be ascertained even after consumption of product that are alternatively known as attributes of concern to the consumer although they are not accessible cues at purchasing point [13] leading to a loose connection between beef quality cues at the sell point and those at end-use. Despites the noted description, the situation was simplified based on the definition of quality convention in specifying rules to characterize products as per user-based approach [17] . Some of the existing forms of conventions are quality traits based on origin or animal breed, on production methods, on specific supply-chain structure [13] . For, example in a case of production methods, the practices involved during meat production are animal feeding, the use of hormones, friendly environmental as well as animal welfare [13] . The success on the noted simplification is quantified by parameters like the market size, the price differential for producers/consumers, and by raising consumer expectations based on their preferences.

Quality of use is determined by two main components, namely, product and context in which it is used, i.e., the particular users, tasks and environment [18] [23] . It is deduced from the interaction between the user and product while carrying out a task in a technical, physical, social and organizational environment as shown in Figure 2. This brings the understanding to the conviction that, the quality of any product is determined based on the measure of its quality of use in particular context under the influence of the situation in which it is used [18] .

Currently, the quality of use and or usability is measured by the methods of the European Metrics for Usability Standards in Computing (MUSiC) project

[18] . This model fits well the beef quality determinants and consumer preferences. These methods provide valid and reliable means of specifying and measuring, while diagnosing feedback which enables the design to be modified to improve usability [18] . The MUSiC methods proved to be successful since they include tools and techniques for measuring that satisfy consumers [18] [24] as shown in Table 3.

The study was carried out in Mbeya and Dar es Salaam and conducted based on the established hypotheses along the pre-slaughtering, slaughtering and post slaughtering chain between January and April 2017. According to 2012 population and housing census, as the Dar es Salaam city with a total area of 1800 km² has a population of 4.36 million in 2012 United Republic of Tanzania (URT), Mbeya city with 19,098 km² has a population of 0.69 million. Dar es Salaam and Mbeya cities are characterized by high level of poverty despite the various efforts on interventions. Among other poverty indicators, the difficultness for their respective residents on accessing the basic needs in 2011 was at 5.2% and 24.3% levels, respectively. As Dar es Salaam has six slaughter facilities with capacity to slaughter 720 cattle a day, Mbeya has three slaughtering facilities with daily capacity of 200 cattle [25] . The stock that is slaughtered in these cities is distributed to butchers that are classified based on income class of the clientele into high, middle, low and extra-low levels. During the study, data that is connected to beef quality management were collected from slaughtering facilities, carcasses distributing vehicles and beef retailing premises. In Mbeya, data was specifically collected from three slaughtering facilities, i.e., Ilemi, Uyole and Mbalizi that were purposely pointed as they being the only registered facilities around that location.

Dar es Salaam city was selected because of its unusually high population due to its metropolitan characteristics which attract many people looking for employment in the industries. Mbeya city was also targeted due to the availability of consumers from working that includes government departments, Institutions, private companies and entrepreneurs [26] . In addition, these cities were chosen as study areas as they have direct access to other countries through Julius Nyerere and Songwe International Airports, respectively, being potential for export of meat from Tanzania.

Beef quality determination indicators were formulated based on the data that were collected from both secondary (assessment of literature records) including scientific reviews and research articles and primary sources of information (field data collection) as conceptualized in the data gathering model in Figure 3.

In this part of the study, beef quality attribute cues were identified, listed and analyzed in connection to consumers’ preference from literature. Consumer preferences on beef varieties have been detailed by many studies in terms of scientific reviews and articles around the world. In this study, systematic review was used to both identify these studies and organize the outcomes in a comprehensive interpretation on the beef quality attributes cues those impacting consumers’ preferences. Systematic review that describes it as the transparent and systematic process rather than specific method for data aggregation and interpretation was carried out following five main steps [14] [28] [29] , i.e., framing the question, identifying relevant studies, appraising the quality of the studies, summarizing the evidence and interpreting the results.

In the first step, the focus question was stated as a query in free form before explicit structured by relating its components and designed. As it was mentioned in literature [14] [28] , the framed question was used as framework for inclusion and exclusion criteria, the search and selection of relevant studies, as well as in synthesize of studies and presentation of the findings. The question was designed based on a population of interest, an intervention/phenomenon of interest, comparators, and outcomes/context of interest acronymic as PICO.

¹ScienceDirect (https://www.sciencedirect.com/).

²HINARI Access to Research in Health Programme (http://www.who.int/hinari/en).

³Access to Global Online Research on Agriculture (http://www.fao.org/agora/en).

Relevant studies were identified in the second step, an activity that between January and April 2017 from ScienceDirect¹, HINARI² and AGORA³ websites that provide subscription-based to a large database of scientific and medical research licensed at TFDA Offices during which more than 13 million documents and 30,000 e-books were accessible. This was comprehensively and unbiased conducted by using key word identification, search strategies for multiple databases, and hand search journals as shown in other studies [14] [29] . In connection to this, a list of records from all databases was exported directly to EndNote⁴ for removing duplicates. Then, the remained records were screened by reviewing the title, abstract or both, in order to remove, off topic records, citations in non-understandable languages and further duplicates. The remained list of records was further screened by focusing on the citations that were published in journal articles that were peer-assessed, thus, acting as a quality assurance criterion of the data and results presented. As it was mentioned in other literature [14] , papers that were not in the Journal of Citations Report (JCR), that have H index⁵ less than 20 were not included in this analysis.

⁴Is an industry standard tool for publishing and managing bibliographies, citations and references on the windows and Macintosh.

⁵Is calculated by counting the number of publications for which an author has been cited by other authors at least that same number of times (this means that the scientist has published at least 20 papers that have each been cited at least 20 times).

In the next step, the records were classified into reviews and research articles, the classes that were applicable for identification and relative ranking of beef quality attribute cues, respectively. The identified reviews were assessed by reading title and abstract before classifying them based on either only one beef quality attribute cue in detail was considered or several attribute cues simultaneously. The full texts of the identified and separated reviews were analyzed for their relevance and quality. Quality was analyzed based on the key sources, clarity and comprehensive description of the argument and novelty in the reviews. Similarly, the list of the selected research articles was obtained by using the same process as in the selection of appropriate citations for the reviews.

Research articles that describe relative importance of different quality indicators from consumer perspective were then identified. The identified articles were further refined based on the described ranking procedures to remain with papers that are clear enough to quantify relative ranking of indicators. Beef quality determination indicators explored from the refined papers were next classified based on search, experience and credence quality and their individual importance within each study was re-scaled as per steps indicated in Table 4.

The previous ranking, R_R, was re-scaled from 1 (as the least important indicator within the article) to 3 (more important indicator within the article) based on Equation (1).

R R = 1 + [ ( 3 − 1 ) ( m l − 1 ) ] × ( g l − 1 ) (1)

Equation (1). Rescalling the previous ranking

In this regards, m_l, indicates maximum level of indicators, g_l, shows a given level of indicator (The number of levels in this equation is not necessarily the same as the number of indicators if some indicators had the same score.

n i f = 1 + [ ( 3 − 1 ) ( m i − 2 ) ] × ( n i − 2 ) (2)

Equation (2). Determination of number of indicators factor n_if, where, m_i, maximum number of indicators in the reviewed paper, n_i, number of indicators given in the reviewed paper.

f s = 1 + ( R R × n i f )

Equation (3). Final score, f_s of each quality cue within a reviewed scientific publication.

In the final step, data were synthesized in a tabular summary of the all the surveyed studies related to each of the key quality attribute cues across search, experience and credence identified as important and facilitate analysis of comparison across studies form like other descriptive synthesis. In this respect the gathered data were entered in designed database in Microsoft excel version 2010 software before being analyzed and interpreted. Data were further presented in a tabular summary indicating beef quality attribute cues, connected with scientific publication reviewed and their respective ranking in figures as clustered columns. Since this is observational study, data were written up in CONSORT) flow diagram as indicated in other studies [29] .

Beef quality attribute cues that impacting consumer preference explored from secondary sources of information were presented across the five main steps of systematic review. First, it includes the focus question, which is asked that ‘is the customers’ choice on a photo of standard graded beef similar to their responses on quality attribute cues that impacting their preferences’.

In line with this question, the structured components are: customers that responded on both photo of beef and quality attribute cues were noted as the population of interest; photo of USDA standard graded beef, was expressed as an intervention of interest; beef quality attribute cues, were explained as the comparator of interest; and customer preferences were outcome of interest. Then, results with respect to second to the final steps of the systematic review processes were presented in CONSORT diagram as shown in Figure 4.

In the flow a list of 930 duplicates was excluded from 3199 records that were obtained from ScienceDirect, HINARI and AGORA by using EndNote. The systematic review was further, elaborated the means in which, a list of 1971 citations among of 2269 records was identified as irrelevant, hence excluded and remaining with a total of 298 documents. The remained records were further classified into 79 reviews and 219 research articles. The assessment and classification of reviews as per either only one indicator in detail or several indicators simultaneously involved resulted to a total of 5 reviews. Similarly a total of 18 articles addressing several indicators were selected for ranking and excluding 32 that dealt with 1 - 2 indicators. Fifty three articles that contained information on relative importance of different beef quality determination indicators as per consumers’ perspective were initially identified. Among the identified papers, 32 were excluded in ranking due to the ambiguous on ranking of the identified indicators. The remaining 28 publication papers resulted to a total of 25 quality attribute cues that impacting customers’ preferences whose individual importance within each study was rescaled as shown in Table 5. The systematic review was finally resulted to 7 reviews and 22 articles as the appropriate records for identification and ranking of beef quality cues.

The identified quality attribute cues that affecting customers’ preferences on beef were ranked based on their relative positions and percentage of scientific

publication referred. In connection to this, Figure 5 presents percentage of scientific publications referred for each of the beef quality attribute cues. The top most referred quality attribute cues were seller’s information (96%), taste (91%) together with tenderness and juiciness that each ne has 83%. The least referred quality attribute cues in the surveyed scientific publication on the other hand were advertisement and smell that each of them has 13% together with traditional image at 9%

The final score for individual beef quality attribute cues were presented in terms of both overall ranking and relative ranking in ascending order across all quality attributes search, experience and credence as shown in Figure 6. The former case was described by the summed of final scores for beef color that was referred in 57.1% of the reviewed scientific publications. This analysis was overall ranked in 5^th position. Similarly the overall three top most ranked beef quality cues were tenderness, price and breed’s information that were analyzed from 46.4%, 57.1% and 32.1% of the respective scientific publication. On the other hand, the lowest three ranked parameters were marbling, branding/packaging and pathogens which were analyzed from 35.7%, 39.3% and 28.6%, respectively. The letter case was described based on price, tenderness and breed’s information which were the top most relatively ranked summed final scores beef quality attribute cues across search, credence and experience quality attributes that were reviewed in 57.1%, 46.4% and 32.1%

Analytical results in this study is similar to other literature [14] in which the topmost relatively ranked beef quality cues were in the highest ranks within the respective beef quality attributes.

Personal relation between sellers and consumers in beef retailing was noted as one of the consumer influencing factor on their respective preferences on beef. The assessment of this parameter was presented as the major factor that influencing consumer preference on beef. It is a customer’s responses on their respective personal relation with sellers as the top most influencing factor distributed in the degree of agreement as shown in Figure 10.

The strength of relation is observed to be about 69.7% and 69.3% of visited consumers in Dar es Salaam and Mbeya, respectively, which positively accept it as reliable beef quality determinant. This quality determination indicator is absolutely not accepted by 30.3% and 30.7% of customers in Dar es Salaam and Mbeya, respectively. On the other hand, approximated discrepancy of 10% of the

consumers is observed in terms of both absolutely rejection and acceptance relation as the highest determinant. The findings of this research are similar to earlier studies that consider personal relationship as one of the beef quality determination indicator in which consumers place much value on being served by butchers of the same ethnic race and religion in traditional market [43] . It was indicated that good relationship between vendors and customers in the traditional market (23%) was important motive compared to those shoppers who purchase meat in modern retail outlet (3%) [43] . It was also reported in other literature that the Arab Islaelis prefer to buy fresh meat from a known and trusted source [43] [44] . Additionally, relation is further described as the concept of convenience as respondents indicated that their preferred retail outlet was close to where they either lived or worked.

Fat content in retail beef is noted as one of the main influences of for determination of beef suitability. This presentation assumes fat content as the only available quality cue to be assessed by consumers in meat retailing shops. The applicability of fat content as determinant of quality in m1eat retailing premises is presented based on the consumers’ responses on the Likert scales on its applicability as judgment criteria. The results are arranged in clustered columns as percentage of consumers’ responses against the categorized Likert scales in Figure 14.

Preference on the retail beef is presented based on fat content during which, about 69% of consumers in each of the visited city were positively accept it as appropriate intrinsic cue. Fat content on the other hand is not accepted by 31% of the customers from each of the visited city. The general responses of the consumers on fat in beef are not evenly distributed due to the observable variation particularly on absolutely rejection and considering fat as the highest determinant in both cities. This is also indicated in literature [47] , where visible fat was used as one of the main criterion for grouping the University meat consumers into three main preference groups, i.e., lean meat (42.2%), moderate fat (18.4%) and meat with fat (39.5%). Society classification based on fat preferences in beef was also observed in the study conducted in 183 villages in Addis Ababa, Ethiopia where the relative dislike for high fat beef was higher among higher income households. On the other hand the findings on the assessed influence of consumers characteristics on the choice of beef quality attributes in Tanzania revealed that fat is the only eating attribute that is favored by consumers [26] .

Fat content in retail beef is noted as one of the main influences of for determination of beef suitability. This presentation assumes fat content as the only available quality cue to be assessed by consumers in meat retailing shops. The applicability of fat content as determinant of quality in meat retailing premises is presented based on the consumers’ responses on the Likert scales on its applicability as judgment criteria. The results are arranged in clustered columns as percentage of consumers’ responses against the categorized Likert scales as indicated in Figure 15.

Preference on the retail beef is presented based on certification during which, 60.7% and 65.3% of consumers in Dar es Salaam and Mbeya were positively accept it as appropriate indicator. Retail and beef certification on the other hand is not accepted by 39.5% and 34.2% of the customers in Dares Salaam and Mbeya. Generally, the consumer variation on responses for certification as quality determinant is not high due to the observable range of about 5%. The finding in this research is parallel with literature that indicated non conformances in beef retailing, during which more than half of the surveyed selling points were operated without certification [12] .

Analytical results with respect to the reliability of customers’ responses on the assessed beef quality attribute cues in Dar es Salaam and Mbeya were summarized in ANOVA Table as shown in Table 6.

With respect to the first hypothesis the output shows that F = 133.8 (p < 2.5 × 10⁻⁵⁰) for Mbeya and F = 91.5 (p < 2.8 × 10⁻⁷⁵) for Dar es Salaam indicating that we should clearly reject the null hypothesis that beef quality cues collectively have no effect on the decision made during beef quality determination in retailing shops. This situation that is also denoted as (β_i ≠ β_j) as indicated in Table 7 proves that each of the beef quality cues has influence on the other in multiple regression analysis. For example, the findings show that the beef quality cue

marbling in Dar es Salaam is significant controlling for the variable beef presentation (p = 0.001), as is beef presentation controlling for the variable marbling (p = 0.001).

Table 7 was also presents results based on the second hypothesis, on the correlation between customers’ responses on quality cues that impacting their respective preference on beef and their respective choice on the photos of standard graded beef.

Looking at individual tests in Table 7(a), five beef quality cues, i.e., price of beef (β = 0.167, p < 0.005), beef color (β = 0.099, p < 0.001), beef presentation (type of cut) (β = 0.085, p = 0.001), Freshness (β = 0.081, p < 0.001) and marbling (β = 0.098, p = 0.001) have significant influence on the consumers’ responses on the displayed photo of beef. Other parameters have not significant influence since they have larger p ? values than 0.05 as recommended for individual tests.

The Table 7(b) shows price of beef (β = 0.190, p = 0.05), beef color (β = 0.129, p = 0.008), type of cut (β = 0.097, p < 0.005), beef cleanliness (β = 0.074, p < 0.005), marbling (β = 0.079, p = 0.048), butcher’s location (β = 0.104, p < 0.005) and fat content (β = 0.118, p < 0.001) have significant influence on the responses on the displayed photo of beef in retailing shops. On the other parameters are not significant, implying that the choice of beef do not depend on these parameters.

The data in Table 8 was further indicated that the model is reliable for the intended use as the adjusted squares were 0.850 (R² = 860, standard error = 0.298) and 0.947 (R² = 954, standard error = 0.198) for Dar es Salaam (N = 208) and Mbeya (N = 98), respectively.

Results with respect to the third hypothesis in this study that compared the surveyed cities with respect to customers’ response on beef quality cues against the portrayed photo of beef are presented in Figure 16. It was indicated that all the assessed quality cues were positively accepted as the factors that influence consumers’ preferences at more than 50%, with smell (aroma) being the lowest accepted parameter in the cities. In addition to that, all the assessed beef quality

cues, except freshness were responded as appropriate beef quality cues that impacting the preferences of customers in Mbeya at higher rate than in Dar es Salaam.

It was further observed that, the top three highest pointed quality cues were fat content (89.4%), personal relation (88.4%) and beef presentation (type of cut) (85.7%) that were responded in Mbeya. On the other hand, the top three scored quality cues in Dar es Salaam were freshness (79.2%), butcher’s location (72.6%) and beef cleanliness (70.2%) together with beef color (70.2%). Similar to this study, butcher’s location was also reported in literature [3] [36] as one of the strong quality cues that impacting customer preferences on beef. This was observed in Gauteng Province, South Africa together with Nairobi and Isiolo counties in Kenya, where customers in lower and middle class areas prefers meat in retail outlets with in lower food safety and quality standard.