food & nutrition research ORIGINAL RESEARCH Groundnut spread likability, sensory properties, and intent to pay for quality certification Ozgur Kaya1, Wojciech J. Florkowski2*, Daniel Sarpong3, Manjeet S. Chinnan4 and Anna V. A. Ressurrecion4 1Department of Economics, School of Business Administration, American University of Sharjah, Sharjah, United Arab Emirates; 2Department of Agricultural and Applied Economics, University of Georgia, Griffin Campus, Griffin, GA, USA; 3Department of Agribusiness Management, University of Ghana-Legon, Legon, Ghana; 4Department of Food Science and Technology, University of Georgia, Griffin, GA, USA Popular scientific summary • Urban lifestyles, rising incomes, nutrition, and quality awareness drive development of novel spread. • Domestic food manufacturers can offer a quality-certified groundnut spread that promises success in marketplace by using local ingredients. • S pread likability and aroma are major factors behind intent to pay for quality certification. • Quality certification payment intent is higher among educated and married people, while it is lower among young and panelists with children. Abstract Quality-certified, nutritious novel groundnut spread has great commercialization possibilities due to evolving urban lifestyles in Africa, but lack of information about likability, sensory attributes, and consumer safety awareness is a severe barrier for small enterprises. This paper examines a novel groundnut spread, made of sorted kernels deemed free of aflatoxin, intended for use on bread in a fashion similar to groundnut paste or groundnut butter, but with modified sensory charac- teristics. In particular, it seeks to measure the effects of sensory attributes of the novel spread on the intent to pay for safety certification and the role of consumer awareness of aflatoxin. A novel spread was prepared with groundnut paste from sorted kernels (to eliminate the risk of aflatoxin con- tamination) and cocoa. Adults intercepted at Ghana’s International Fair in 2012 volunteered to sample the spread and complete a questionnaire. Results from a tasting panel of untrained participants established that sensory attributes and panellist characteristics are relevant to the intent to pay for quality certification. Spread likability, aroma, education, knowledge about aflatoxin, packaging and being married were identified as major factors increasing the probability of intent to pay for quality certification whereas young age and the presence of children in a household lowered the probability. Results also identified income, education level, and having young children at home as increasing the chances of knowing about aflatoxin. Groundnut paste available in Ghana is often contaminated by aflatoxin as it is in other countries in the region and consumers cannot visually assess paste quality. Under the circumstances, quality certification is necessary. Keywords: Groundnut; quality certification; untrained panel; likability; aroma; aflatoxin awareness Received: 30 May 2019; Revised: 31 October 2019; Accepted: 24 November 2019; Published: 9 January 2020 Groundnuts are a relatively inexpensive protein sauces (1, 2), but nowadays groundnut paste is also con-source produced by African farmers and well- sumed as a spread on bread and other baked foods. The liked by consumers. Traditional ways of eating use of groundnut paste as a spread is relatively new and groundnuts involve the use of groundnut paste as an practiced more often by Ghanaian urban consumers, ingredient to various dishes, including soups stews, and who, similar to many Africans, have gradually departed Food & Nutrition Research 2020. © 2020 Qzgur Kaya et al. This is an Open Access article distributed under the terms of the Creative Commons Attribution 4.0 International License (http:// 1 creativecommons.org/licenses/by/4.0/), allowing third parties to copy and redistribute the material in any medium or format and to remix, transform, and build upon the material for any purpose, even commercially, provided the original work is properly cited and states its license. Citation: Food & Nutrition Research 2020, 64: 3600 - http://dx.doi.org/10.29219/fnr.v64.3600 (page number not for citation purpose) Ozgur Kaya et al. from the traditional diet (3) as this form of consumption link between buyer preferences for sensory attributes and fits the evolving urban lifestyle (4). The combination of product safety assurances, typically limiting their market groundnut protein and cereal has been suggested as the success (9, 10). way to improve the protein and nutrient density in Afri- can diets (5) and to combat malnutrition in societies (4). Materials and methods With consumer incomes increasing in Ghana, the opportunity arises to develop and promote novel foods Samples based on domestically available ingredients. Consumers Shelled groundnuts used to manufacture the spread were are interested in functional food products, and ground- purchased from an open-air market in Accra agglomera- nuts are a functional food (6–8). However, food processors tion. Kernels were sorted on the processor premises using have little knowledge of consumer preferences, including the recommended manual sorting method (15). In contrast product taste and preferences (9, 10). The novelty asso- to developed economies, groundnuts traded in Ghana are ciated with the modification of an existing food product not identified by variety but rather by their appearance, enhances its public understanding and its ingredients and primarily the color of testa and kernel size. A type of assures acceptability. A groundnut paste with the addition groundnuts commonly used for making groundnut paste of domestically produced cocoa powder has been used to are small-kernel groundnuts, called ‘Chinese’ by trad- develop a novelty: a chocolate spread intended for use on ers, resembling the ‘Florunner’ variety. This type differs bread or bakery products in Ghana. The spread has two from the large kernel ‘Bugla’ used mostly to make roasted ingredients (cocoa and groundnut) with bioactive compo- groundnuts or another type (‘manipintar’) characterized nents creating functional food product, a good source of by high oil content and initially introduced by the govern- protein, fatty acids, and important micronutrients. ment to produce groundnut oil. None of the traded types The disadvantage of groundnuts as a protein source is is genetically uniform because they are grown from seeds their susceptibility to contamination by aflatoxin, a me- saved by farmers and traded at regional markets, while tabolite of Aspergillus spp. Ingestion of aflatoxin leads certified variety-specific seeds were not available. to severe health problems, and various procedures have The process of manual sorting involves the removal of been recommended to reduce aflatoxin presence (11). The testa (blanching), and this step differs from traditional emerging food manufacturers in Ghana could refer to sort- processing of groundnuts into paste in Ghana during ing as a feasible method of assuring consumers that the which testae are left in the batch. Blanching allows the vi- groundnut product is aflatoxin-safe and establish a claim sual inspection of kernels and the removal of visibly dam- to label a product as quality certified. Novelty-seeking con- aged kernels or splits. Once sorted, kernels are roasted to sumers who pay attention to food safety will purchase the develop the flavor. The type of groundnuts used is im- quality-certified product, especially if the product is well- portant because groundnut germplasm and variety effect liked. Choosing the novel spread shows preferences for the taste of roasted groundnut (4), and also because the quality, but the cost of certification has been viewed as a uniformity of kernel size assures even roasting. Roasted limiting factor (9), while sorting kernels implies that a por- kernels were ground in a plate mill, a commonly available tion of kernels is removed and represents a loss to a proces- equipment in mills providing grinding services in Ghana. sor. The loss of potential raw material has to be recouped Later, cocoa and other ingredients were added to obtain by processor to assure economic viability of the firm. the final product, the groundnut–choco spread. This article examines novel groundnut spread, made from sorted kernels deemed free of aflatoxin, intended for Sensory testing and panelist information use on bread in a manner similar to groundnut paste or To test how well a product is liked, this study applies the groundnut butter, but with modified sensory characteris- hedonic or affective type of test from among the available tics. The addition of cocoa and the safety certification will sensory evaluation methods (16). This method implies the differentiate the product from a plain groundnut paste use of untrained panel screened for product use. In the commonly found in Ghana. In particular, the study seeks case of groundnut spread, the data collection consisted to measure the effects of sensory attributes of the novel of two parts. The untrained panel (naïve panel) was cho- spread on the intent to pay for safety certification and the sen based on the availability and interest from among the role of consumer awareness about aflatoxin. There is no visitors at Ghana’s International Fair in Accra in July and scientific published study on the role of attributes of that August 2012. The panelists were likely representatives of type of spread that help predict the consumer decision in the consumers who are familiar with groundnut paste and countries located in sub-Saharan Africa. Since the ma- capable to indicate the level of acceptability of its sensory jority of food processing companies are small, owned by attributes, while their personal and household features are female entrepreneurs in Ghana (12–14), knowledge gen- consistent with the ability to purchase a novel spread. The erated by the study offsets the gap in understanding the testing took place throughout the duration of the fair at 2 Citation: Food & Nutrition Research 2020, 64: 3600 - http://dx.doi.org/10.29219/fnr.v64.3600 (page number not for citation purpose) Commercialization possibilities of quality-certified, nutritious novel groundnut spread the booth operated by the small food processing company. type of equipment used by small food processors (a plate Organization of product testing in a well-controlled en- mill) in Ghana yields groundnut paste that is commonly vironment remains prohibitively expensive for small food gritty and could be crunchy. Particle size in groundnut processors even if it is available in the country. butter has been researched (20), but panelists are accus- Each visitor who volunteered to participate in the test tomed to locally accessible products, which are made using was offered a teaspoon of the spread followed by an op- available equipment, and the perceptions are inherently portunity to complete a questionnaire. None of the indi- subjective. A question of practical implications probed viduals who tested the spread was obligated to participate panelists about their preferred jar size for the novel spread in the second part of the data collection. The questionnaire offering several size choices. A common size of a jar visible was self-administered. Previous studies have indicated in food stores in Ghana was the 450-g jar and reflected that naïve panelists may have difficulty in describing their a common preference for a relatively large size container sensory experience (17), and hence a list of the sensory among price-sensitive consumers. That size plastic jars attributes included in the posed questions facilitated that with twisted lid were commonly available for food proces- task. Since the study focused on likability of the spread, sors and could be easily purchased from local suppliers. the question probed for that specific attribute using a nine- Likability and other attributes of the spread do not ex- point balanced scale with the centered neutral category clusively determine the purchase of a novel spread. The (16). The equal distance between the points is amenable to apprehension about health effects leads to loss of trust regression analysis applied later in the current study. and weakens marketing of nut spreads (18). Subsequently, Liking is the interaction of the consumer and the prod- other questions in the survey instrument probed the pan- uct (18). Lawless and Heymann note that consumers, such elists for their views about intent to pay for quality certi- as those in Ghana who are familiar with groundnut paste fication of groundnut spread and awareness of aflatoxin. commonly used in cooking, render opinions based on the The questions required a straightforward choice of ‘yes’ integrated pattern of perceptions (16). Since the purpose or ‘no’, leading to creation of a binary variable used in of the study includes learning about the intention to pay statistical analysis.The panelists also shared information for quality certification, knowing about the potential rel- about personal and household characteristics pertinent to evance of specific attributes expressed through nonan- the objective of the study and used in regression analysis alytical spontaneous opinions is valuable. Additionally, to generate their measurable effects on the probability of a separate question focused on spread attributes and choosing to pay for quality certification in the context of included a request to rate the importance of the spread knowing about aflatoxin, a major, internationally recog- taste, color, thickness and crunchiness using a five-point nized food safety issue (21). balanced scale (where 1 = not important at all to 5 = very important). These attributes play an important role in Statistical method consumer appeal and in nut-spread buying decision (18), Responses to questions about the intent to pay for qual- and the untrained panelists easily understood the applied ity certification of a groundnut spread and awareness of terms describing attributes. Taste, experienced only during aflatoxin in foods, converted to numerical measures, have the sampling of the spread, is an essential attribute for been coded as one of two possible outcomes reflecting the sustained marketing and repeated sales, which are the ul- choice of ‘yes’ (1) or ‘no’ (0). When the dependent vari- timate goal of the food processor. Its role in the case of able Y is binary, the population regression function is the groundnut–choco spread is particularly important because probability that Y = 1, conditional on the regressors. The the cost of safety certification increases the total cost and logit and probit regressions are nonlinear and specifically is later reflected in the price of marketed product. Color designed for estimation of equations with binary-depen- of the spread could matter because of the cocoa addition dent variables of the following nature: to the groundnut paste and the absence of a compara- Let y* be an unobserved, or latent, variable, determined by ble product on the market at the time of the test. Color of groundnut paste can vary across varieties (4), but the y*= x b + ε, novelty of adding cocoa restricted the ability of compar- y = 1 if y* > 0, isons made by the untrained panelists, although negative y = 0 if y*1≤ 0, perception of the color could create an obstacle to mar- keting. Color was found relevant in previous sensory stud- where x is the explanatory variable vector, β is the co- ies (19), but its importance needs to be verified because efficient vector, and is the random error term. Also, it is it is specific for each spread. Thickness of the spread is assumed that ε is independent of x and that e has either relevant because the panelists, familiar with groundnut the standard logistic distribution (logit model) or the paste, likely compare that attribute to the novel product. standard normal distribution (probit model). There may Crunchiness is not typically expected in spreads, but the be practical reasons for favoring logit or probit in some Citation: Food & Nutrition Research 2020, 64: 3600 - http://dx.doi.org/10.29219/fnr.v64.3600 3 (page number not for citation purpose) Ozgur Kaya et al. cases, but it is difficult to justify the choice on theoretical the nine-point balanced scale. Taste was, by far, the most grounds (22). Therefore, the choice is based on a compari- important attribute compared to other attributes in the cat- son of test results. Accordingly, based on the Akaike’s In- egory ‘very important’. Its mode value was 5 on the five- formation Criterion (AIC) and the Bayesian Information step scale. Among other attributes, the largest portion of Criterion (BIC), the logistic equations produced smaller panelists found aroma, color, and thickness of the ground- values of both tests than the alternative probit equations nut–choco spread to be important (Table 1). and were applied in the current study. Aroma importance (mode 4 on the five-point balanced The evaluation of goodness-of-fit of logistic equations scale), in particular, stands out and suggests that the com- may involve several formulas of a pseudo R2. The current bination of groundnut and cocoa is very attractive to pan- study reports McFadden’s R2, a measure most commonly elists. Crunchiness was not expected to be important, and used in empirical studies. When comparing two equations the results show that its mode and median values were 4 applying the same data, the McFadden’s R2 would be on the five-point balanced scale. Crunchiness, although it higher for the equation with the greater likelihood (23). was ranked relatively high in importance (above 3 on aver- Additionally, this study reports Tjur’s coefficient of dis- age), was the last in importance among the four attributes crimination, that is, Tjur’s R2 (24). and the overall likeability. From a manufacturer and consumer standpoint, the size Results of the container of groundnut–choco spread is important because consumers in Ghana have been conditioned by man- Panel summary results ufacturers of other foods sold in plastic jars or cans to seek Product features are essential, especially in a novel product a specific container size. The presence of jars of specific size in spite of containing the well-known and accepted ground- shapes buyer expectations, and manufacturers consider con- nut paste and cocoa. A question listing five attributes of sumer desires seeking a competitive advantage (25). More- groundnut paste aimed at collecting information about the over, if a manufacturer decides to buy packaging equipment, relative importance of each attribute was included in the the investment is likely to pay off if the size of containers it survey. More than four out of five (83%) panelists liked handles corresponds to consumer expectations. In Ghana, the groundnut–choco spread, and the mode value was 9 on like in many other sub-Saharan countries (9), small food Table 1. Descriptive statistics of variables in the sample applied in the empirical analysis Variable Variable description Mean Std. dev. Mode Dependent variables Assur_qual Intent to pay for quality assurance = 1 if yes 0.15 0.35 0.00 Aflatoxin Knowledge about aflatoxin = 1 if yes 0.11 0.31 0.00 Independent variables Likenessa How did you like groundnut–choco spread? 7.10 2.48 9.00 Colorb 3.50 1.13 4.00 Thicknessb 3.56 1.02 4.00 Tasteb 4.10 1.01 5.00 Crunchinessb 3.35 1.14 4.00 Size Preferred groundnut–choco spread package size; 100 g = 1; 250 g = 2; 350 3.13 1.15 3.00 g = 3; 450 g = 4; 550 g = 5; above 550 g = 6 Gender =1 if male 0.43 0.49 0.00 Age Age (in years) 29.61 9.57 23.00 Marital status =1 if panelist is married 0.35 0.47 0.00 Household size Number of household members including panelist 3.70 2.73 1.00 Less than 18 years =1 if there are household members younger than 18 years 0.50 0.50 1.00 Income Household income in month preceding survey in Ghanaian cedis 621.28 619.99 300.00 Education 1 = no formal education; 2 = junior high school/middle school; 3 = senior high/GCE 5.16 2.66 8.00 O-A level; 4 = vocational school; 5 = technical school; 6 = teacher training; 7 = nursing training; 8 = university; 9 = postgraduate Government employee =1 if panelist is government/civil/public employee 0.44 0.49 0.00 aA nine-step scale ranging from 1 = did not like it at all to 9 = liked it very much and 5 = neither liked nor disliked it.bScale: 1 = not important at all; 2 = not important; 3 = neither important nor unimportant; 4 = important; 5 = very important. 4 Citation: Food & Nutrition Research 2020, 64: 3600 - http://dx.doi.org/10.29219/fnr.v64.3600 (page number not for citation purpose) Commercialization possibilities of quality-certified, nutritious novel groundnut spread Table 2. Distribution of responses regarding the importance of attributes of groundnut–choco spread using a five-step scale (in %) Attribute Not important at all (%) Not important (%) Neutral (%) Important (%) Very important (%) Color 4.06 22.52 9.74 46.65 17.04 Aroma 2.22 18.38 10.10 49.49 19.80 Thickness 2.02 18.18 17.37 46.06 16.36 Taste 1.62 10.73 5.26 40.69 41.70 Crunchiness 5.33 22.75 18.03 38.93 14.96 Figures in rows may not add to 100% due to rounding. processors can purchase recycled jars, but such containers negatively associated: household size and the presence of pose safety risk. Containers from vendors trading in pack- children aged less than 18 years. aging materials may cost more, and are clean, but they still have to adhere to the size expected by customers. Thus, the Statistical validity of logistic relationships question presented to panelists listed six jar sizes. The most The results in Tables 3 and 4 show the odds ratios, the cor- preferred among panelists (53%) is the 350-g groundnut– responding z-statistics, the marginal effects, and the P-values choco spread jar size. The balance was about equally divided of marginal effects. The odds ratios, where the coefficient is between those selecting smaller or larger jar sizes. the effect of a unit change in an explanatory variable on the Among the untrained panelists, 57% of the panelists were predicted odds ratio while other variables are held constant, male, 35% were married, and 50% had at least one house- carry more practical information than logistic slope coeffi- hold member younger than 18 years (Table 2). The average cients. For example, the odds ratio greater than 1 suggests household size comprised 3.7 persons. About 44% of the that the event is more likely to happen than not to happen. panelists were government employees, 33% were self-em- The interpretation of odds effects has limited practical ployed, and the remaining 23% were unemployed, retired, value, therefore Tables 3 and 4 report the associated aver- or students. The panelists’ age ranged from 18 to 71 years, age marginal effects of each predictor on the probability with an average person being about 30 years old (median of a positive outcome in dependent variable, that is, intent age 27 years). The relatively young age is not a surprise, to pay for quality assurance and knowledge of aflatoxin. given the demographic structure of the country’s popula- The marginal effect of a continuous variable measures tion, where an estimated 57.3% were 24 years old or younger change in the probability of dependent variable as the for- in 2014 (26). Almost 98% of the panelists had formal educa- mer changes by a unit (e.g. 1 year older than the average tion, and 41% received a university or postgraduate degree; age of a panelist). In the case of a binary variable, the therefore, the mode value of 8 (university degree) is not sur- average effect shows how the dependent variable changes prising. The income reported by panelists ranged from 50 to as the binary variable changes from 0 to 1. 5,000 Ghanaian cedi, with the mean of 621 Ghanaian cedi.11 In the case of the intent to pay for quality certification The empirical examination of the intent to pay for qual- (Table 3), the likelihood ratio Chi2, 100.85, is statistically ity assurance and knowledge about aflatoxin using regres- significant, with a P-value of 0.0000 suggesting the overall sion technique was preceded by the preliminary exploration good fit of the specified empirical relationship with the data. of the related factors. The analysis was conducted using The value of pseudo R2 is 0.301, the same as Tjur’s (Table 3) PROC FACTOR in SAS software, version 14.1. Four fac- and common in studies using the cross-sectional data (27). tors are associated with paying for quality assurance and The ability to predict correct outcomes is very high (90.49%; knowledge about aflatoxin. Each of the four factor load- Table 3). Solid predictive performance of both equations is ings exceeding 0.6 that was associated with either variable important from practical standpoint of marketing the novel consisted of a very similar and distinct set of characteris- spread by small food processors as they identified statisti- tics. The first factor was associated with the four attributes cally significant effects truly influencing consumer choice. of the groundnut–choco spread (each loading exceeding 0.75) and the overall likeability of the product; the second factor was positively associated with traditional house- Discussion holds characterized by older and married panelists with Groundnut paste available in Ghana is often contami- children aged less than 18 years. In the case of knowing nated by aflatoxin as it is in other countries of the region about aflatoxin, this factor also is positively associated with (28) or other parts of sub-Saharan Africa (29, 30), and income. Education and knowledge about aflatoxin were consumers are not able to assess visually the paste qual- identified in factor 3, while with factor 4 two measures were ity (30). Under the circumstances, quality certification is necessary as it results in discarding some of the purchased 1Around USD 320 in August 2012. raw groundnuts during sorting. Product attributes appear Citation: Food & Nutrition Research 2020, 64: 3600 - http://dx.doi.org/10.29219/fnr.v64.3600 5 (page number not for citation purpose) Ozgur Kaya et al. Table 3. Logit results identifying factors that affect the intent to pay for quality assurance on groundnut–choco spread label and their individual effects on probability of the intent to pay Variables Odds ratio z Probability change P-value Liked (categories 7 and 9) 3.938* 1.76 0.10* 0.08 Color (1 = very important) 2.084 1.36 0.06 0.17 Thickness (1 = very important) 0.518 –0.95 –0.05 0.34 Taste (1 = very important) 0.957 –0.10 –0.00 0.92 Crunchy (1 = very important) 0.407 –1.28 –0.07 0.20 Aroma (1 = very important) 5.987*** 3.09 0.14*** 0.00 Gender (male dummy) 1.569 1.24 0.03 0.21 Age (1 = 30 years old or younger) 0.381** –2.12 –0.072** 0.03 Low income dummy 1.04 0.10 0.00 0.92 Household size 1.07 0.85 0.01 0.40 Knowledge of aflatoxin (1=yes) 3.614*** 5.60 0.10*** 0.00 Marital status (1 = single) 2.450* 1.80 0.07* 0.07 Children younger than 18 years (1 = yes) 0.438* –1.65 –0.062* 0.098 Education 2.501** 2.29 0.07** 0.02 Jar size (1 = 450 g or larger) 2.709*** 2.64 0.08*** 0.01 Constant 0.007*** –5.56 - - Number of observations 452 LR Chi2 (16) 100.853 Prob > Chi2 0.0000 Pseudo R2 (McFadden’s) 0.301 Tjur’s R2 0.301 Log likelihood –116.93 Proportion correctly classified 90.49% *P < 0.1;**P < 0.05;***P < 0.01. to be important in the intent to pay for the quality certi- paying for certification increases by 6.9% for university or fication of novel spread, and their quantified effects are postgraduate degree holders (Table 3). Younger panelists quite large (Table 3). Certification required for proper had low odds of choosing to pay for quality certification, monitoring of the manufacturing process was found im- and being 30 years old or younger lowered the probability portant to Ghanaian consumers (31). by 7.2% as compared to older panelists. Likability is a prominent factor influencing the inten- A negative effect on the intent to pay for safety certifi- tion to pay for quality certification. The odds suggest that cation is confirmed regarding the presence of children in those who liked the product ranking it high on a nine- a household (Table 3). Young children at home are asso- point balanced scale were 3.9 times more likely to pay as ciated with decreasing odds (0.438) of a panelist’s intent compared to those who liked the spread less. The proba- to pay for quality assurance (Table 3) as suggested by lo- bility of choosing to pay increased by 10.3% for panelists gistic regression results. The probability of intending to who liked the product. It appears that aroma is a critical pay for safety certification of the spread associated with attribute of the novel groundnut–choco spread since the the presence of young children at home decreases by 6.2% panelists who felt it was a very important attribute had al- (Table  3) as compared to panelists not having children. most six times higher odds to pay for quality certification. Because the vast majority of households in Ghana have The probability of intending to pay increased by 13.5%, children, the decreased probability of paying for safety the largest probability change of all statistically significant certification puts pressure on food manufacturers to price factors. The combination of roasted groundnut and cocoa the certified spread, or to market the spread stressing its appears to have a very strong effect on panelists’ decision. sensory attributes and likability, among others. A panelist with a university or postgraduate degree is 2.5 Panelists who know about aflatoxin have 3.6 times times more likely to pay extra for a quality assurance label higher odds (Table 3) regarding the intent to pay for qual- (Table 3) than a panelist with less education. The average ity certification as compared to those who do not know marginal effect of education shows that the probability of about aflatoxin. The high odds are reflected in an increase 6 Citation: Food & Nutrition Research 2020, 64: 3600 - http://dx.doi.org/10.29219/fnr.v64.3600 (page number not for citation purpose) Commercialization possibilities of quality-certified, nutritious novel groundnut spread Table 4. Logit results identifying factors characterizing knowledge of aflatoxin and their individual effects on the probability of such knowledge Variables Odds ratio z Probability change P-value Liked groundnut–choco spread 0.764*** –2.98 –0.0240*** 0.002 Color 0.692 –1.61 –0.0328 0.105 Thickness 1.114 0.33 0.0097 0.743 Taste 2.059*** 2.75 0.0645*** 0.005 Crunchiness 1.533* 1.66 0.0382* 0.094 Education 2.216* 1.84 0.0711* 0.064 Gender 1.791 1.52 0.0521 0.126 Age 1.015 0.57 0.0013 0.566 Household size 0.914 –0.94 –0.0080 0.347 Marital status 0.699 –0.75 –0.0320 0.453 Children younger than 18 years 2.689* 1.93 0.0884* 0.052 Government employee 1.890 1.58 0.0569 0.112 Income 1.001** 2.44 0.0001** 0.013 Constant 0.003 –4.21 Number of observations 330 LR Chi2 50.57 Prob > Chi2 0.0000 Pseudo R2 (McFadden’s) 0.2010 Tjur’s R2 0.2077 Log likelihood –100.50 Proportion correctly classified 89.39% *P < 0.1;**P < 0.05;***P < 0.01. in probability of the intent to pay for quality certification odds value of 0.764 (Table 4). The effect translated into a an extra 9.7% (Table 3). 2.4% decrease in the probability of knowing about afla- The size of the jar of spread also affects the intent to toxin (Table 4). Liking the product could override caution pay a premium. A panelist who prefers a jar size of 450 with regard to food safety, and the effect stresses the impor- g or larger is 2.7 times more likely to pay for a quality tance of assuring the supply of safe groundnut products. assurance label as compared to smaller size jars (Table 3). The spread’s taste and crunchiness are also important Also, the probability of the intent to pay extra for certi- predictors of probability of knowledge about aflatoxin as fication increases by 7.5% (Table 3). A larger jar permits indicated by logistic regression results (Table 4). Ground- the use of a large label to communicate product quality nuts are stored in ambient temperatures throughout the and efforts undertaken to safeguard it; therefore, prefer- marketing chain in Ghana, while many households lack ence for a larger jar allows educating consumers about the refrigeration to store food products. Ambient tempera- distinctiveness of spread. tures encourage the gradual growth of mold on kernels The small proportion of panelists who know about at high relative humidity levels, and auto-oxidation of aflatoxin (11%; Table 1) is comparable to that of earlier groundnut paste is rapid (19), resulting in progression reports from Ghana or other countries in sub-Saha- of rancidity and thus altering the taste. A panelist who ran Africa (e.g. 32–34). Even among groundnut proces- liked the taste and crunchiness has 2.06 (Table 4) and 1.53 sors and distributors in other countries, only 32% admit (Table 4) higher odds, respectively, of reporting to know knowing about aflatoxin’s detrimental health effects (35). about aflatoxin than those who did not like the taste and This study quantifies the changes in probability of know- crunchiness. Specific attributes are characterized by par- ing about aflatoxin with regard to sensory attributes of ticularly heavy loading in preliminary factor analysis, and the novelty spread and socioeconomic and demographic their positive association with knowing about aflatoxin re- characteristics of panelists and their households. It is pre- flects discerning panelists (mold can affect the taste) or is supposed that knowledge about aflatoxin leads to the pur- confounded by other factors. The more the panelist likes chase of quality-certified spread. the overall taste and crunchiness of the spread, the higher If a panelist overall liked the groundnut–choco spread, the probability of aflatoxin knowledge (6.5% for taste and knowing about aflatoxin is less likely as suggested by the 3.8% for crunchiness; Table 4). Citation: Food & Nutrition Research 2020, 64: 3600 - http://dx.doi.org/10.29219/fnr.v64.3600 7 (page number not for citation purpose) Ozgur Kaya et al. The study results identify income, education level, and Results of the current study assign changes in probabili- having young children at home as factors increasing the ties to likability, sensory attributes, and panelist character- odds, that is, the chances of knowing about aflatoxin in- istics associated with preference of quality certification. For crease (Table 4). An increase in income above the average groundnut processors, the key message to communicate to leads to an increase in odds ratio by 1, suggesting that buyers is that the production process adheres to procedures those with high incomes (Table 4) are considerably more that assure that only safe ingredients are used in the final likely to be aware of aflatoxin. Accordingly, the marginal product. Interestingly, the current study’s findings that both effect of income obtained from logistic regression suggests household size and number of children aged less than 18 that if the average income increases by 100 cedis (or about years lower the probability of choosing to pay for quality 16% above the sample mean value shown in Table 1), the certification coincide with similar variables associated with probability of knowing about aflatoxin increases by 1% high levels of aflatoxin in blood samples among Ghana- (Table 4). Per capita income has increased by 15.3% be- ians (37). The effect of sensory attributes, especially aroma, tween 2011 and 2015, possibly and indirectly contributing seems critical for the groundnut–choco spread in encourag- to the effect of income on aflatoxin knowledge. ing purchase. This may require in-store sampling opportu- Higher educational attainment is a likely contributor to nities in promoting sales of the quality-certified product. wider knowledge and appears specifically to be associated Provision of appropriate packaging material helps with knowing about aflatoxin as confirmed by the logistic small enterprises to enter formal markets (9). Panelists regression results (Table 4); if a panelist has a university expressing preference for a 450-g or larger jar have more or postgraduate degree, he or she is 2.22 times more likely than 7.5% higher probability of choosing to pay for qual- to know about aflatoxin (Table 4). The average marginal ity assurance on product label. The jar size allows placing effect of education shows that the probability of aflatoxin a sufficiently larger label with information of why and how knowledge increases by 7.1% (Table 4) for university or safety certification is accomplished. Lowering the costs of postgraduate degree holders. Since there has been a rapid quality certification has been suggested among the devel- increase in college enrolment in recent years, the effect of opment of a system of licensed laboratories. Many small education is likely to improve consumer knowledge about enterprises fail because of nonfinancial reasons (13), but aflatoxin. food scientists could offer guidance about specific infor- The presence of children and greater odds of knowl- mation that can be placed on labels. However, such as- edge of aflatoxin are highly encouraging results confirmed sistance is not yet readily available in Ghana because the by logistic regression (Table 4) because children are par- links between the industry and researchers are incidental ticularly vulnerable to the effects of aflatoxin (5). Specifi- due to enterprises’ limited resources. cally, panelists who have young children at home are 2.69 Such situation opens opportunities for the government times more likely to know about aflatoxin than those with agencies to develop certification procedures, and imple- no young children (Table 4). The increase in probability of menting them in the initial stages facilitates access to knowing about aflatoxin increases by 8.8% if young chil- laboratories of government research institutes. Initially, dren are present at home (Table 4). the government may choose to subsidize the certification of spread for micro and small enterprises to support the Conclusion market entry of their product. Gradually, the government Groundnuts and groundnut products are consumed daily could shift certification to commercial laboratories while by many residents of Ghana and sub-Saharan Africa. In monitoring their performance, pricing of service, and Ghana, the existing standards do not differ from those in assuring consistently high-quality certification services. many other countries but are excluded from compliance The approach would contribute to the broadening of the by numerous micro and small enterprises. Consequently, role of private sector in emerging food sector and may es- processors are free to decide whether to use only pre- tablish grounds for wide retailing and even exporting of sorted groundnuts that, in turn, may impact the extent to certified spread. Retailers in sub-Saharan Africa consider which groundnut products are safe from aflatoxin. certification of products while seeking suppliers (38). Food processors are forced to supply safe groundnut Separately, the government could prepare a framework products (21, 36). This study, using data obtained from allowing university scientists to formally assist private untrained panelists expressing their subjective opinions, micro and small enterprises in the scale-up of novel- shows that panelists who express intent to pay for qual- ty-spread manufacturing. Such services, available in the ity certification readily accept a novel groundnut–choco developed countries, have proved important in accelerat- spread. A domestic food manufacturer using local ingredi- ing the proportion and, consequently, reducing the costs ents can offer a novel groundnut spread that promises suc- and increasing the flow of revenues of enterprises. Par- cess in marketplace. The manufacturing process assuring ticipation of nongovernmental organizations (NGOs) in safety is a matter of concern for Ghanaian consumers (31). some of the activities, including training, could be helpful. 8 Citation: Food & Nutrition Research 2020, 64: 3600 - http://dx.doi.org/10.29219/fnr.v64.3600 (page number not for citation purpose) Commercialization possibilities of quality-certified, nutritious novel groundnut spread References 19. Santos BL, Koehler PE, Resurreccion AVA. Sensory analysis of peanut-based imitation cheese spread. J Food Qual 1987; 10: 1. Osseo-Asare F. ‘We eat with our eyes’: on Ghanaian cuisine. 43–56. doi: 10.1111/j.1745-4557.1987.tb00288.x Gastronomica 2002; 2(1): 49–57. 2 0. How JSL, Young CT. Factors affecting peanut butter pref- 2. Osseo-Asare F. Food culture in sub-Saharan Africa. West- erence. J Am Oil Chem Soc 1985; 62: 538–40. doi: 10.1007/ port, CT: Greenwood Publishing Group; 2005. doi: 10.1525/ BF02542328 gfc.2002.2.1.49 21. IARC. Wild CP, Miller D, Groopman JD,IARC (International 3. Samuel F, Antinmo T. Obesity and cardiovascular diseases: the Agency for Research on Cancer) Mycotoxin control in low- and risk factor in African diets. The Forum on Public Policy. 2008; middle income countries. In: Wild CP, Miller D, Groopman JD, 1–13. Available from: http://www.forumonpublicpolicy.com/ar- eds. IARC Working Group Report, vol. 9. Lyon, France: IARC chivespring08/samuel.pdf [cited May 2019]. Press; 2015, pp. 31–42. 4. Dhamsaniya NK, Patel NC, Dabhi MN. Selection of ground- 22. Greene WH. Econometric analysis, 6th ed. Upper Saddle River, nut variety for making a good quality peanut butter. J Food Sci NJ: Pearson; 2008. Technol 2012; 49(1): 115–18. doi: 10.1007/s13197-011-0361-9 2 3. Freese J, Long JS. Regression models for categorical dependent 5. Temba MC, Njobeh PB, Adebo OA, Olugbile AO, Kayitesi E. The variables using Stata. College Station, TX: Stata Press; 2006. role of compositing cereals with legumes to alleviate protein energy 24. Tjur T. Coefficients of determination in logistic regression mod- malnutrition in Africa. Int J Food Sci Tech 2016; 51: 543–54. doi: els – a new proposal: the coefficient of discrimination. Am Stat 10.1111/ijfs.13035 2009; 63: 366–72. doi: 10.1198/tast.2009.08210 6. Francisco MLD, Resurreccion A. Functional components 25. Rundh B. The multi-faceted dimension of packaging marketing in peanuts. Crit Rev Food Sci Nutr 2008; 48(8): 715–46. doi: logistic or marketing tool? Br Food J 2005; 107(9): 670–84. doi: 10.1080/10408390701640718 10.1108/00070700510615053 7. Sales JM, Resurreccion A. Resveratrol in peanuts. Crit Rev Food 2 6. Factbook CIA. The world factbook; 2015. Available from: Sci Nutr 2014; 54(6): 734–70. doi: 10.1080/10408398.2011.606928 http://www.cia.gov/library/publications/the-world-factbook 8. Arya SS, Salve AR, Chauhan S. Peanuts as functional food: [cited 30 January 2015]. a review. J Food Sci Technol 2016; 53(1): 31–41. doi: 10.1007/ 2 7. Wooldridge J. Introductory econometrics: a modern approach, s13197-015-2007-9 2nd ed. Cincinnati, OH: South Western College Pub; 2003. 9. Mhazo N, Hanyani-Mlambo B, Proctor S, Nazare RM. Con- 2 8. Waliyar F, Umeh VC, Tragore A, Osiru M, Ntare BR, straints to small-scale production and marketing of processed D iarra B, et al. Prevalence and distribution of aflatoxin food products in Zimbabwe – the case of fruits and vegetables. contamination in groundnut (Arachis hypogaea L.) in Mali, Food Africa, Internet Forum. Palais des Congrès, Yaounde, West Africa. Crop Protection 2015; 70: 1–7. doi: 10.1016/j. Cameroon; 2003. cropro.2014.12.007 1 0. Mhazo N, Mvumi BM, Nyakudya E, Nazare RM. The 29. Bumbangi NF, Muma JB, Choongo K, Mukanga M, Velu status of the agro-processing industry in Zimbabwe with MR, Veldman F, et al. Occurrence and factors of aflatoxin particular reference to small- and medium-scale enter- contamination of raw peanuts from Lusaka district’s mar- prises. Afr J Agric Res 2012; 7(11): 1607–22. doi: 10.5897/ kets, Zambia. Food Control 2016; 68: 291–6. doi: 10.1016/j. AJAR10.079 foodcont.2016.04.004 1 1. Torres AM, Barros GG, Palacios SA, Chulze SN, Battilani P. 3 0. Njoroge SMC, Matumba L, Kanenga K, Siambi M, Waliyar F, Review of pre- and post-harvest management of peanuts to Maruwo J, et al. A case for regular aflatoxin monitoring in pea- minimize aflatoxin contamination. Food Res Int 2014; 62: nut butter in sub-Saharan Africa: lessons from a 3-year survey 11–19. doi: 10.1016/j.foodres.2014.02.023 in Zambia. J Food Protect 2016; 79: 795–800. doi: 10.4315/0362- 12. Agyapong D. Micro, small and medium enterprises’ activities, 028X.JFP-15-542 income level and poverty reduction in Ghana – a synthesis of 31. Owureku-Asare M, Kingsly Ambrose RP, Oduro I, Tortoe C, Saa- related literature. Int J Bus Manag 2010; 5(12): 196–205. doi: lia FK. Consumer knowledge, preference, and perceived quality of 10.5539/ijbm.v5n12p196 dried tomato products in Ghana. Food Sci Nutr 2017; 5: 617–24. 1 3. Abor J, Quartey P. Issues in SME development in Ghana and doi: 10.1002/fsn3.439 South Africa. Int Res J Finance Econ 2010; 39: 1450–2887. 32. Jolly CM, Bayard B, Awuah RT, Fialor SC, Williams JT. Exam- 14. Ababio PF, Lovatt P. A review on food safety and food hygiene ining the structure of awareness and perceptions of groundnut studies in Ghana. Food Control 2015; 47: 92–7. doi: 10.1016/j. aflatoxin among Ghanaian health and agricultural professionals foodcont.2014.06.041 and its influence on their actions. J Socio-Econ 2009; 38: 280–7. 15. Lustre A, Paloma L, Francisco M, Resurreccion A. Impact as- doi: 10.1016/j.socec.2008.05.013 sessment of the Peanut CRSP project in the Philippines Part 2. 33. Ezekiel CN, Sulyok M, Babalola DA, Warth B, Ezekiel VC, USAID – Peanut CRSP Monograph Series No. 9. Griffin, GA: Krska R. Incidence and consumer awareness of toxigenic Peanut CRSP, University of Georgia; 2007. Aspergillus section Flavi and aflatoxin B1 in peanut cake from 16. Lawless HT, Heymann H. Sensory evaluation of food: Nigeria. Food Control 2013; 30: 596–601. doi: 10.1016/j. principles and practices. Second edition. New York: Springer; foodcont.2012.07.048 2010. 3 4. Matumba L, Monojerezi M, Kankwamba H, Njoroge SNC, 17. Heymann H. A comparison of free choice profiling and multidimen- Ndilowe, P, Kabuli H, et al. Knowledge, attitude, and practices sional scaling in vanilla samples. J Sens Stud 1994; 9(4): 445–53. doi: concerning presence of molds in foods among members of the 10.1111/j.1745-459X.1994.tb00258.x general public in Malawi. Mycotoxin Res 2016; 32: 27–36. doi: 18. Shakerardekani A, Karim R, Ghazali HM, Chin NL. Textural, 10.1007/s12550-015-0237-3 rheological and sensory properties and oxidative stability of nut 35. Azaman NNM, Kamarulzaman NH, Shamsudin MN, Selamat J. spreads – a review. Int J Mol Sci 2013; 14: 4223–41. doi: 10.3390/ Stakeholders’ knowledge, attitude, and practices (KAP) towards ijms14024223 Citation: Food & Nutrition Research 2020, 64: 3600 - http://dx.doi.org/10.29219/fnr.v64.3600 9 (page number not for citation purpose) Ozgur Kaya et al. aflatoxins contamination in peanut-based products. Food Con- 38. Nandonde FA, Kuada J. Modern food retailing buying be- trol 2016; 70: 249–56. doi: 10.1016/j.foodcont.2016.05.058 haviour in Africa: the case of Tanzania. Br Food J 2016; 118(5): 36. Udomkun P, Wiredu AN, Nagle M, Bandyopadhyaya R, 1163–78. doi: 10.1108/BFJ-09-2015-0335 Müller J, Vanlauve B. Mycotoxins in sub-Saharan Africa: present situation, socio-economic impact, awareness, and outlook. Food Control 2014; 72: 110–22. doi: 10.1016/j. *Wojciech J. Florkowski foodcont.2016.07.039 Department of Agricultural and Applied Economics 37. Jolly P, Jiang Y, Ellis W, Awuah R, Nnedu O, Phillips T, et al. University of Georgia, Griffin Campus Determinants of aflatoxin levels in Ghanaians: sociodemo- 1109 Experiment Street graphic factors, knowledge of aflatoxin and food handling Griffin, GA 30223-1797 and consumption practices. Int J Hyg Environ Health 2006; USA 209: 345–58. doi: 10.1016/j.ijheh.2006.02.002 Email: wojciech@uga.edu 10 Citation: Food & Nutrition Research 2020, 64: 3600 - http://dx.doi.org/10.29219/fnr.v64.3600 (page number not for citation purpose)