Investigating Consumers’ Preference on Fresh Vegetables in Bangladesh: Best-Worst Scaling Approach
BAU Research System (BAURES), Bangladesh Agricultural University, Bangladesh
Submission: March 10, 2020; Published: March 17, 2020
*Corresponding author: PK Sarma, BAU Research System (BAURES), Bangladesh Agricultural University, Bangladesh
How to cite this article: PK Sarma. Investigating Consumers’ Preference on Fresh Vegetables in Bangladesh: Best-Worst Scaling Approach. Agri Res& Tech: Open Access J. 2020; 24(1): 556253. DOI: 10.19080/ARTOAJ.2019.22.556254
Safety vegetables are nutritionally essential for our good health, but we are daily fighting against to take “Safe and Fresh Vegetables” for “sake of “Good Health” due to contamination, adulterations and pesticides residues effects. The aim of the study was to examine the consumers’ preference for food quality and safety attributes of fresh vegetables by using quality and safety attributes of fresh vegetables using the best-worst scaling technique. The balanced incomplete block design procedure was employed to obtain a total 180 sample from Dhaka and Mymensingh city of Bangladesh. The data were collected through survey monkey software by using the pre-determined structured questionnaire and analyzed by the MaxDiff scaling and multinomial mixed logit regression model. The results revealed that fresh vegetable attributes were a combination of size, appearance, texture, freshness, and flavor as well as nutritional ingredients and safety aspects that determine their value to the consumer. The fresh vegetables contained vitamins, minerals and dietary fiber that satisfied consumer demand and helped to keep healthy and protect from diseases. The researchers found that nutritional value, purity, freshness, shelf life and safety are the most important attributes of fresh vegetables and variety, size, content, colour are the least important for the consumers. The study concludes that demand for can be significantly stimulated in the study area when these most important attributes have been considered by producers, processors, and marketers.
Bangladesh is the 3rd position in global vegetable production, but it remained in the lower position in intake. The total vegetables production is 3.73 million but demand annually 13.25 million metric tonnes in Bangladesh (FAO, 2015). According to the World Health Organization (WHO), a healthy individual should take 220 grams of vegetable daily of which Bangladesh is intake only 70 percent of need. Vegetables are the main sources of essential vitamins such as A, C, niacin, riboflavin and thiamin, minerals such as calcium and iron and dietary fiber. They added to eat of crucial nutrients from other foods by making the more taste and dietary fiber necessary for digestion and maintaining health curing nutritional disorders . Vegetables are not only minimizing the malnutrition but also maximize the economic returns.
Farmers are widely using toxic chemicals for vegetables production and pesticides uses to fight against pests. That’s why vegetables are adulterated by various harmful chemicals and pesticides. On the other hand, traders are using toxic artificial
ingredient and colors to extend its appearance and shelf life. Using chemical preservatives without concern for the health
of the consumer is rampant. Now a day, food adulteration is
becoming a “Silent Killer” in Bangladesh. Because, vegetables are contaminated by toxic chemicals and pesticides like carbide, formalin, heavy metal, chemical, textile colors, artificial sweeteners, DDT, urea and so on pose a serious threat to public health especially damaging vital organs like liver, kidney, pancreas etc. It is very difficult to find fresh and adulteration free vegetables in the market. For this reason, the worldwide fresh vegetables market has growing rapidly in recent years to a multi-billion-dollar sector, largely driven by increasing population growth, urbanization and changing consumption pattern as a result increasing consumer demand for healthy, freshly prepared convenient vegetables . The quality of fresh vegetable includes a combination of attributes like appearance, texture and flavor, as well as nutritional and safety aspects that determine their value to the consumer. But challenge for producers, traders and retailers in market is to predict consumer preferences for vegetables, because
consumers represent the essential demand for vegetables and
this demand is always not stable. Moreover, several studies [3,4]
have been undertaken to determine consumer preferences for
vegetable attributes and respective willingness to pay for various
characteristics, thereby providing vegetables with desirable
attributes to consumers and keeping the market fresh and alive.
Furthermore, the lack of food safety, food security, nutrition
and profitability are a problem no inherent to nature but with
unjust societal systems and institutions. Lusk and Briggeman 
also argued that understanding why consumers prefer a given
attribute is important to determine the presence of market failure
and therefore the need for public policy intervention. That’s why,
researchers set the objective of study is to elicit fresh vegetables
attribute preferences among quality conscious consumers in
Dhaka city and Mymensingh city of Bangladesh.
Best-Worst Scaling (BWS) is also called Maximum Difference
Scaling (MaxDiff.) which is based on random utility theory
(Thurstone, 1927; Cohen 2003) [6-8]. It is defined as a choicebased
measurement approach that reconciles the need for
question parsimony with the advantage of choice tasks that force
individuals to make choices. BWS be a way to overcome some
key shortcomings of common rating approaches (e.g. ties among
items, response style bias, and standardization difficulties .
The BWS is superior to other methods, such as rating scales,
because it forces respondents to discriminate between items,
so results obtained from BWS are easy to interpret . Unlike
the rating scale, the BWS can also be used to compare both the
intra and inter attributes. It builds on a body of items. The BWS
approach effectively permits respondents to evaluate all pairwise
combinations of alternatives presented in a particular subset
leading to the assumption that their “Best” and “Worst” choices
represent the maximum difference in utility between all attributes.
A respondent gets presented a series of sets and is asked to choose
one best item and one worst item in each set. BWS assumes
everyone has a latent scale of “utility” or “importance”, and the
scale value of objects will determine a person’s choice over these
objects. This scale value consists of two parts: one systematic part
which can be understood as “intrinsic”, and one random error
component. Therefore, the probability of a certain project chosen
as the best given other K options can be expressed as:
in which, and represent the scale value, the systematic
component of the scale value and the error term respectively.
McFadden further developed the conditional logit model under
the independently and identical distribution assumption of the
error terms, and the choice probability can be expressed as below
Compared with other designs there are several advantages of
BWS. It is suggested that choosing the “best” and the “worst” from a
set of options can generate relatively consistent results compared
to the ranking task and it offers more information than simply
choosing the “best”. Besides, when attributes under evaluation
have multiple levels, BWS approach enables a comparison of
all attribute levels on a single scale by setting only one level as
The BWS is widely used for collecting data to analyze the
consumers’ preferences and becoming popular over time. It forces
respondents to discriminate between scaled items and uses an
underlying scale ratio of measurement . This BWS method
is consistent with consumer utility maximization. Based on this
method, respondents are presented a set of items and they are
demanded to indicate which one is the best and which is the
worst . For instance, the BWS was applied to investigate the
relative importance consumers place on food values and to study
preferences for sustainable farming practices . Furthermore,
efficient estimates can be obtained when repeated choices are
made by the same respondent, which is the case in this present
study . Finally, the BWS method is gaining more popularity as
a better alternative to the rating system and measuring value .
Therefore, the BWS method reported above fit well in this present
study related to consumers’ preferences for quality and safety
attributes of fresh vegetables. This is captured by the difference
between the most preferred and least preferred items chosen
from the set, when consumers are making a purchase decision.
This can be mathematically represented as in this equation:
Uij is the utility for consumer i choosing vegetables having
attribute, Vij and εij are the deterministic component and the error
term of utility, respectively.
BWS was used to determine values which consumers placed
on quality and safety. Thus, the balanced incomplete block design
(BIBD) method was used to design the questionnaire administered
to respondents. In total 15 (fifteen) attributes of vegetables
have been considered in the study (Table 1). Fifteen blocks or
questions and eight attributes were randomly assigned to each.
This questionnaire was used to collect data from Dhaka and
Mymensingh City such as city market, shops, open vendor where
vegetables are sold to consumers [16-19]. For each question are
asked to choose which option of vegetables attributes they most
preferred and which one they least preferred. A sample question
for this study based on best-worst scaling method is presented in
The study was conducted in Dhaka and Mymensingh City of
Bangladesh because of large number of urban populations, quality
conscious consumers live in the cities that’s why a considerable
amount of the fresh vegetables are comes here from all over the
country. The balanced incomplete block design technique was
used to obtain a total 180 sample (100 Dhaka City consumers and
80 Mymensingh City). Survey Monkey software was used to collect
by pre-determine structured questionnaire and interviewed in
F2F due to lack of reached respondents by e-mail, phone and
internet facilities of Bangladeshi people. Descriptive statistics,
BWS/MaxDiff and multinomial mixed logit regression model were
used to analyze the collected data.
The demographic profiles of respondents are illustrated in the
Table 3. Among the total respondents about 63.50% were male and
36.50% were female. The ages were grouped into five categories
and found that the highest (40.67%) age group under 30-40 years
old and followed by 50-60 years old (25.34%), 40-50 years old
(14.48%) and above 60 years old (2.25%) which is indicated most
of the young respondents involved in household buying activities.
As for matrials status about 83.89% were married, 12.22%
were unmarried and 3.89% were unknown about the marital
information as per survey questionnaire in the study areas. About
24.30% were employed in Government sectors, 57.30% were in
private sectors, 7.50% were unemployed and 4.10% were retired.
The result found that 5.21% of respondents’ income was below
BDT 20000 per month, 18.81% was in BDT 20000-40000, 25.46%
was BDT 40000-60000, 25.46% was BDT 60000-80000, 19.82%
was in BDT 80000-100000 and 12.4% was BDT 100000 and
above, respectfully. The household size was 1-10 member with
a relatively high share of respondent 37.78% were 4-5 persons
per family, 25.00% was 2-3 persons per family, 22.78 % was 6-7
persons per family, 5.00% was 8-9 persons per family, 4.44%
were about 10 persons and 0.56% were more than 10 persons
per family and 4.44% was living alone. The Table 3 also showed
that about 52.22% of the respondent had graduate education.
This was followed by 20.00% had postgraduate, 17.78 % had
higher secondary education, 6.67% had secondary education.
This implies that most of the respondents attained certain level
of formal education, which would facilitate understanding of
improved selection method of fresh vegetable of the respondents
more receptive to advisory services [20-25].
The Table 4 shows that parameter estimates from the MML
model. The coefficients with positive sign indicate that attributes
are preferred, while coefficients with negative signs indicate that
attributes are not preferred by consumers. The results found that
nutritive value, shelf life, availability, purity and safety are positive
and statistically significant, implying that consumers preferred to
have vegetables products with these attributes. It also shows that
varieties, nutrient content, packaging and handling convenience
were negative and statistically significant, indicating consumers
significantly discounted vegetables products having these
attributes and standard deviation regardless of location (Dhaka
city or Mymensingh city) for nutrient content, color, shelf life,
purity, safety, hygiene, availability, handling convenience, type of
animals, origin, food miles and therapeutic value were statistically
significant, implying that these parameters indeed randomly vary
over the population.
The relative importance of 15 (fifteen) attributes of vegetables
is estimated by using MML model (Table 5). Researchers found
that nutritive value, on average, is the most important vegetable
attributes and significantly more important than hygiene, shelf
life, availability, purity and safety are the next most important.
Table 5 summarizes the results of simple count analysis. The
relative importance of each variable is indicated by its B-W score,
and the standard deviation suggests the degree of heterogeneity
of consumers’ perceptions on fresh vegetable attributes for
food safety. The result shows that “standardization” “freshness”
“physical appearance” and “traceability” are substantially more
highly valued than the rest of the attributes, while “audit during
sales life as long as possible” and “cold chain with pesticide and
chemical used when asked by consumers” are underrated by
most respondents. Most consumers are inclined to hold more
positive attitudes towards “physical appearance to traceability”,
“hormone” and “fertilizer residue”, and understate the importance
of “fertilizer residue” and “pesticide residue costs with producers”.
Nevertheless, consumer’ perceptions on the importance of
vegetable attributes, such as “hormone”, “fertilizer residue” and
“pesticide residue life as long as possible” tend to diverge as
reflected by the large standard deviations of their B-W scores,
which also suggests the need to further exp lore respondent
Table 6 showed that the average BWS scores with high scores
corresponding to attributes that were like and the low scores
corresponding to attributes that were disliked. The scores reflect
some expected patterns which provide some evidences that our
consumers appropriately read and interpreted the questionnaire.
BWS scores also support quantitative comparison between levels
within an attribute, and even across attributes something that
is not possible with traditional likert-scale ratings. The Table
6 shows that the average best-worst scaling (BWS) score were
-0.806 for food borne pathogens, -0.698 for heavy metals, -0.452
for pesticide residues, 0.415 for food additives, .060 for naturally
occurring toxins and -0.340 for veterinary residues of food
safety attributes. The average best-worst scaling (BWS) score
of nutrition attributes were fat (0.040), calories (0.554), fiber
(0.541), vitamins (0.253), sodium (-0.523), and mineral (0.323).
Under the value attributes of fresh vegetables, the BWS scores
were purity (-0.250), compositional integrity (0.216), appearance
(-0.343), taste (0.251), convenience of preparation (-0.542) and
size (-0.351). The average best-worst scaling (BWS) scores of
production process attributes were genetic modification (-0.351),
environmental impact (0.560) and pesticide use (0.140).
BWS was used to identify the fresh vegetables attribute
preference among the quality conscious consumers in the
selected areas of Bangladesh. BIBD method was used to design
the questionnaires and 180 consumers were purposively sampled
and interviewed. The results found that nutritive value, shelf life,
availability, safety and purity are the most important attributes,
whereas verity, nutrient content, handling convenience and
food miles are the least important attributes. This indicates that
consumers place a high preference for nutritive value, shelf life,
availability, safety and purity attributes of fresh vegetables. It
implies that consumers are more concerned about vegetables
with these attributes and this can significantly stimulate demand.
The output also helps people along the value chain of vegetables
to provide consumers with desirable attributes of vegetables,
thereby enhance not only the profitability of their business,
but also improve the overall welfare of consumers. The BWS
technique, however, appropriate in ranking order of importance,
had never been used in ranking choice of fresh vegetable purchase.
Finally, the study created future direction for ranking of choice
setting of consumer behavior analysis to researchers, academician
and policymaker in context of Bangladesh as well as developing
The authors gratefully thank the anonymous reviewers of
this article for helpful comments for further improvement. The
authors also acknowledge the contributions of other study team
who are involved in a team. All referees and the journal editor are
gratefully appreciated for their valuable comments, consequent
recommendation and suggestions on earlier drafts of this
Abubakar MI, Amadou Z, Daniel K (2014) Best-Worst Scaling Approach in Predicting Seed Attribute Preferences among Resource Poor Farmers in Northern Nigeria. International Journal of Humanities and Social Science 2(9): 304-310.
Bal HS, Göktolga ZG, Karkacier O (2006) Consumer Awareness on Food safety: The Case of Tokat. Journal of Agricultural Economics 12 (1): 9-18.
Cannon GC (1992) Food and Health: The Exports Agree. An Analysis of 100 Authoritative Scientific Reports on Food. Nutrition and Public Health Published throughout the World in Thirty Years, between 1961 and 1991, Consumers Association, London, UK.
Ijabadeniyi OA (2010) Effect of Irrigation Water Quality on the Microbiological Safety of Fresh Vegetables. PhD thesis, Pretoria University of Agricultural and Food Science, Johannesburg, South Africa.