Food Quality andPreference
(1991/2)
215-221
PREMIUMHONEYS: RESPONSEOFSENSORYPANELISTS Ruthann B. Swanson
& Carol E. Lewis
Agricultural and Forestry Experiment Station, University of Alaska Fairbanks, Fairbanks, AK 99’775-0080, USA (Received 28 November 1991; accepted 8 February 1993)
honeys in the US market are typically an ungraded blend relatively consistent in flavor and color. A small but growing segment of the retail honey market consists of exotic premium-priced honeys (Humann, 1991). In this gourmet market, single source or natural blends from floral sources of regional significance are commercially important (White, 1978). These products are further characterized by variability in color and flavor depending on seasonal and environmental factors. Environmental conditions influence color more than flavor; flavor is influenced primarily by floral source (White, 1978). Alaskan honeys are relatively new entrants in the gourmet market. In 1990, there were approximately 1000 bee colonies in Alaska which produced 60 000 lb of honey (Petersen, 1991). Important floral sources are fireweed, raspberry, wild rose, willow and clovers (Petersen, 1989). The relative acceptability of these honeys is unknown. The evaluation of the relative acceptability of similar food products may be enhanced by the use of the R-index when ranking is employed. This signal detection method allows detection of differences that may be easily obscured when traditional scaling techniques are used (O’Mahony, 1985). Although psychophysically valid, this technique has been applied to few food products and investigation of its use with products differing in sensory properties has been suggested (Vie et al., 1991). The purpose of this study was to evaluate the relative acceptability and willingness of consumers to buy selected Alaskan honeys using the R-index and the more traditionally applied ranked sums procedures. In affluent Western countries, it is likely that a large number of similar products from various production locations and floral sources will compete for market share. Factors such as nutrition, health and safety, as well as cost and availability, influence the consumers’ selection of a specific alternative product. Therefore, these factors were assessed in a survey. Some physical characteristics of these honeys were also determined using instrumental techniques as an indicator of similarity of the samples.
ABSTRACT Four honeys and one synthetic ‘honey ’ were ranked by 62 panelists for relative acceptability; willingness to buy was ranked after price information was provided. Data were analyzed with the R-index procedure and Friedman ranked sums test. Honey floral source and production location differed. Panelists completed a questionnaire concerningfood selection and perception and selection of honey. Instrumental techniques were used to characterize the honeys. Honey was perceived b these panelists to have a unique$avor and positive image. Alaska wild?ower; a fireweed table honey and Alaska $reweed/clover the most acceptable. Acceptability
were
did not necessarily
correspond to willingness to buy. Similar results were obtained from both the R-index lyses. The instrumental
and rank sums ana-
assessment of color and ac-
ceptability were significantly correlated. Generally, characteristics that distinguish premium from table honeys were unimportant
in the selection process used
by these panelists. Keywords: Premium honeys, Acceptability, R-index, Rank sums test.
INTRODUCTION Honey the
consumption
1980s
has been
in the USA and
increasing
in the other
throughout nine
major
In the USA, the largest increases have been in the processing and food service sectors; however, increases in the use of retail honey have also occurred (Hoff & Phillips, 1989). Retail honey
producing
countries.
0 1993 Elsevier Science Publishers Ltd 0950-3293/93/$06.00 215
216
Ruthann B. Swanson, CarolE. Lewis
MATERIALS
AND
METHODS
Panelists The panelists (N = 62; 36M, 26F) were university employees, students and their friends. All but one panelist had attended college and 56.4% were college graduates; more than half (51.6%) were full-time students. Nearly all resided in households with four or less members; more than half lived in households with one or two members. Age distribution between the ages of 18 and 54 was platykurtic. When compared to the population over 18 years of age in the Fairbanks North Star Borough (Gramstad, 1991), these respondents were slightly younger and better educated. Average household size was about the same. Only panelists who had obtained honey during the last year were selected to carry out the sensory evaluation and complete the survey on a single day in April 1991. Of these, 75.8% used honey several times per month or more. The majority (87.1%) also consumed processed foods that contained honey. Approximately threequarters (75.8%) of them were the principal household food purchaser and 87.1% were the principal household food preparer.
Samples Three premium Alaskan honeys harvested in the fall of 1990 and one mass-produced table honey were evaluated. A synthetic ‘honey’ processed from Alaskan fireweed and clover blossoms and sucrose also was included.
Instrumental
analyses
Moisture content based on refractive index was determined in triplicate using the AOAC methods (Horwitz, 1975) with a Abbe-3L refractometer (Milton-Roy Co., Rochester, NY). Viscosity was determined in duplicate using a viscometer (Model DV-II, Brookfield Engineering Laboratories, Inc., Stroughton, MA) equipped with a helipath stand at a spindle speed of 12. A T-bar spindle E was used for the analysis. The room temperature was 21°C. A Lovibond 2000 Comparator (Tintometer Co., Williamsburg, VA) equipped with United States Department of Agriculture (USDA) honey color class discs was used to determine color. An Altex pH meter (Model 3500, Beckman Instruments, Inc., San Ramon, CA) was used to determine the pH in triplicate.
Sensory analyses Panelists ranked the five samples in order of relative acceptability. Samples were presented in clear plastic cups coded with three-digit random numbers and
tasted with white plastic spoons. Water at room temperature was provided for rinsing. Panelists were instructed to cleanse their palate by eating a portion of an unsalted cracker. No suggestion of retasting was made. The positions of the honey samples were randomized. Testing took place in a central location (Vie et aZ., 1991). Instructions and interaction with the panelists were both verbal and written; panelists did not interact with each other. After ranking for relative acceptability, the panelists were given price information for the honeys and asked to rank the samples in order of willingness to buy. It was not suggested that the ranking should change. Panelists were reminded that they had just sampled the products being evaluated and could retaste them if they wished. Prices of an 8 ounce jar of honey were $3.00 for the Alaska wildflower, a multifloral blend, $4.00 for the Alaska fireweed, and $3.50 for the Alaska fireweed-clover blend. The mass-produced table fireweed honey was $1.19. The price of the synthetic ‘honey’ was $0.89. After completing the sensory tests, panelists were asked to complete a survey. This survey was pilot-tested, revised and again pilot-tested prior to administering it to these panelists. Questions were either forcedchoice or open-ended. Questions elicited information regarding food selection in general, panelists’ opinions of honey and their honey consumption patterns. Completion of the sensory tests and the survey took approximately 15-20 min.
Data analysis Instrumental data except color measurements were analyzed by analysis of variance; a Student NewmanKeuls test was used for means separation ( p < 0.05). R-index values were calculated for the sensory data (Vie et al, 1991). The R-index estimates the probability that these consumers will choose a given honey sample over the baseline sample. Increasing values between 0.50 and 1.0 indicate increasing likelihood that a given sample will be preferred over the baseline sample. A Friedman ranked sums test for all treatments ( p < 0.05) was also used to analyze the sensory data (Watts et al, 1989). Kendall’s coefficient of concordance ( W) was used to determine the degree to which the panelists agreed with each other’s rankings of the honeys for relative acceptability and willingness to buy. The probability that ranking changed with price information was determined with the Goodman-Kruskal G coefficient (Daniel, 1990). Correlations between instrumental and sensory data were determined with Spearman’s rank procedure, when appropriate (Trant et al, 1981). For the survey data, the frequency of response was tabulated and significant relationships between variables were delineated using chi-square analysis ( p < O-05), when appropriate (SAS Institute, Inc., 1990).
Premium Honeys TABLE 1. Characteristics
of Honeys
2 17
Evaluated
Honey type
Moisture‘* (%)
Alaska wildflower Purchased fireweed Alaska fireweedclover Alaska synthetic Alaska fireweed
Viiosity ’ (% Brookfield)
16.1a 16.9b 17.2~ 19.4d 16.2a
Color d
33.7c 29.5b 281b 11.5a 42.5d
pH ’
White White Extra white Extra white Water white
4.23d 3.88b 4.Olc 3.62a 4.03c
“Means of three replicates; means followed by the same letters are not significantly different according to Student Newman-Keuls test ( p < 0.05). ‘Moisture was based on refractive indices determined with an Abbe-3L refractometer. ‘Means of two replicates determined with a Brookfield viscometer equipped with a helipath stand at a T-bar spindle E speed of 12 rev/min; means followed by the same letters are not significantly different according to Student Newman-Keuls test (P
Comparator.
RESULTS Chemical
AND
DISCUSSION
and physical characteristics
of the honeys
The chemical and physical characteristics of the honeys are reported in Table 1. Significant differences were found in the moisture content of the honeys evaluated. However, all except the synthetic ‘honey’ were below the maximum limit (186%) specified by USDA for the top two grades (White, 1978). Significant differences in viscosity were found, despite evaluation of the honeys under constant temperature conditions. Temperature, water content, and floral source determine honey viscosity. The color of honey according to the USDA classification ranges from water white to dark amber (White, 1975) with amber the typical color of retail honeys in the US market (Humann, 1991). The measurement of the color of the honeys in this trial showed it to be lighter than that of the most common retail honeys in the USA. Honey color is related to its mineral content and is characteristic of the floral source; mineral content has been reported to be lower in the two lightest color classes (White, 1978). Significant variation in TABLE 2. R-indices and Friedman
Rank Sums from
Honey
color of the fireweed honeys from different production locations was observed. The pH of these honeys was significantly different. However, all of the samples evaluated including the synthetic product were at the more acidic end of the range (3.42-6.10) reported in the literature (White, 1978). Overall, these honeys had very similar chemical and physical characteristics.
Acceptability
of the honeys
The Alaska wildflower, mass-produced fireweed and Alaska fireweedclover honeys did not differ and were the most preferred (Table 2) according to rank sums analyses. The R-indices also indicated that these honeys were the most acceptable. The Alaska fireweed honey, the least preferred sample, served as the baseline for calculating the R-indices. Both tests indicated that the Alaska synthetic was less acceptable than the three top ranked samples. This product approximated the acceptability of the Alaska fireweed honey, the baseline sample. In general, there was a consensus among the panelists regarding the overall ordering of samples for acceptability; Kendall’s coefficient of concordance (W) was 0.13 [X2 = 31.00 (df= 4, N= 62) p< 0.0051. Rankings
for Acceptability
and Willingness
to Buy when Price was
Known
Degree of acceptability Honey
Willingnessto buy
R-index
Honey
tYPe
we
Alaska wildflower Mass-produced fireweed Alaska fireweed/clover Alaska synthetic Alaska fireweed “Rank sum values within a column
Price (8 02)
0.73 0.71 0.70 0.53 0.50 followed
by different
150b 165b 169b 218a 228a
Mass-produced fireweed Alaska wildflower Alaska synthetic Alaska fireweed-clover Alaska fireweed
letters are significantly
different
( p < 0.05).
$1.19 $3.00 $0.89 $3.50 $4.00
R-index
Rank SUIYISfl
0.87 0.84 0.78 0.78 0.50
133c 163bc 173bc 197b 264a
218
Ruthann B. Swanson, Carol E. Lewis
TABLE 3. Panelists’ Perception of Honey
Statement
Concept
Image
No. of respondents
% of respondents saying: Strongly agree
Agree
Slightly agree
Diiee
Is too expensive Is a pure and natural product Usually contains additives and preservatives Is high in calories Is better for you than sugar
62 61
1.6 32.8
6.5 44.3
33.9 16.4
58.1 6.5
61
1.6
3.3
29.5
65.6
61 60
11.5 20.0
57.4 38.3
21.3 23.3
9.8 18.3
Product attributes
Has a unique flavor Improves food flavor Improves food texture Lighter the color, the better the quality
62 62 61 58
58.1 19.4 8.2 3.4
41.9 56.5 29.5 6.9
0 19.4 34.4 15.5
0 4.8 27.9 74.1
Handling and storage
Is difficult to cook with Is messy to use Can’t be stored for a long time
59 62 62
0 9.7 3.2
5.1 29.0 9.7
15.2 46.8 9.7
79.7 14.5 77.4
The Alaska fireweed honey, which was the least preferred
and most expensive
honey evaluated,
the least likely to be purchased was
provided.
honeys
However,
did not necessarily
to buy once
of the panelists nation
the
were
most
middle
should
of
tended price
Goodman-Krnskal association
between
instructions change,
rankings
to move
up or down
to change
However,
G coefficient, the rankings
indicated one
in the to
the strength
when
the
of the
for acceptability
to buy [Kendall’s
that place
Panelists
ranked
according
willingness to buy was O-73. Panelists were also in concordance the samples for willingness
75%
An exami-
was provided. samples
other
to willingness
panelist’s
scale.
the
their rankings.
information
likely the
not necessarily
did change
of individual
when
correspond
for
the price was known. Despite
that the rankings
the ranking
once price information
preference
Panelists’ honey selection criteria
remained
and
ranking
coefficient
W = 0.25; X2 = 62.00 (df = 4, N = 62) of concordance p < 0.0051. In general, there was an increase in the R-index values after price information was provided and the range of rank sums values also increased. The mass-produced fireweed which did not differ in acceptability from the Alaska wildflower and fireweed-clover honeys, but was less expensive, was statistically most likely to be purchased once the price was known. The synthetic ‘honey’ which was less liked but was the least expensive product evaluated, was more likely to be purchased when the price was known (Table 2). These changes suggest that price is used as a criterion when these panelists select among similar appropriately priced products. According to survey responses, honey is not perceived by these panelists to be too expensive (Table 3).
Although the table honey most frequently found in US supermarkets is a branded, ungraded blend, relatively consistent in color and flavor (Humann, 1991), honeys available in urban Alaska are typically graded, with floral source and production location identified. Gourmet or exotic honeys rely on their production location and floral source for market appeal (Hoff & Phillips, 1989). However, less than one-third of these panelists indicated that production location or brand name were either fairly or very important in their selection among available honeys (Table 4). Floral source, a major determinant of flavor, was a fairly or very important selection criterion for 50%. Interestingly, flavor was an important selection criterion for 85% (Table 4). The majority of these panelists further indicated that they were typically brand and label conscious (Table 5). It has been previously suggested that consumers primarily read labels to ascertain whether chemical preservatives and additives are present (Sloan, 1985). In this study, 82% of the respondents agreed to some extent (slightly agree, 34%; agree, 35%; strongly agree, 13%) that they would prefer to purchase organic or natural foods when available. The relative unimportance of label information on honeys may reflect the perception by most of these panelists that honey is a pure and natural product that does not contain additives and preservatives (Table 3). Panelists may also be unaware of the relationship between flavor and floral source. Only 54.7% of those panelists (n = 53) who regarded flavor as important in their selection of honeys also identified floral source as an important selection criterion.
Premium Honqrs
2 19
TABLE 4. Honey Selection Criteria Concept
Parameter
No. of respondents
% of respondentssaying: very important
Fairly important
Slightly important
Not at all important
Quality
Flavor Color Floral source Geographic location of production Grade
62 62 62 62 61
56.5 8.1 8.1 8.1 8.2
29.0 21.0 41.9 19.4 29.5
11.3 43.5 29.0 25.8 36.0
3.2 27.4 21.0 46.8 26.2
Marketing
Price Brand name Type of container
62 62 62
16.1 3.2 8.1
41.9 11.3 21.0
32.3 27.4 40.3
9.6 58.1 30.6
A significant sessment -0.95).
As color
creasingly milder, panelists the
(White,
the
product
differs
colors
1978).
Indeed,
responding quality’
instrumental was found from
are
became the
to the
survey
(Table
3).
inwith
sponse
may be task related
the survey immediately characteristics
of
these
as the panelists
after
of reference
honeys
the lightest
honeys
(Table color
of these
panelists
(Table
dis-
However,
this re-
cate
that
color
honey
than
(Table
4).
2). Survey responses is of lesser
most other
by the
do, however, indi-
importance
parameters
the
was the least ac-
and the least likely to be purchased
the color,
The
probably
1). Among
ceptable
query
answered
the honeys.
particular
served as a frame evaluated,
tasting
are more
the majority
that ‘the lighter
as(rs =
general
associated
flavors and therefore
with the statement
better
the
This
lighter
pleasant
when
between
acceptability
darkened,
that
more
acceptable
and
acceptable.
assumption
agreed
relationship
of color
in selecting
that affect
quality
TABLE 5. Panelists’ Attitudesabout Food Selection Concept
Statement
No. of respondents
% of respondentssaying: Strongly
Agree
SWtly ape
Disagree
agree Importance of quality
Brand/label conscious
When making a food purchase, quality is more important than price Quality of products available from speciality shops is usually better than similar products available in the supermarket Nutritional value is an important factor in my food selection Usually read labels to identify processing methods or ingredients used in food products Generic brands are just as good as brand name foods I usually do not purchase a similar product if my favorite brand is not available
62
25.8
32.3
35.5
6.5
60
8.3
25.0
35.0
31.7
62
45.2
51.6
3.2
0
61
26.2
36-l
19.7
18.0
61
9.8
21.3
45.9
22.9
62
4.8
8.1
30.6
56.5
220
Ruthann B. Swanson, CarolE. Lewis
Nearly 73% of the panelists in this study obtained honey most frequently from the supermarket, although 67.1% also obtained honey from other sources. There was a general impression among two-thirds of these panelists that the quality tends to be at least slightly better if the product is available in speciality shops as opposed to supermarkets (Table 5). At present, Alaskan honeys are marketed primarily through farmers’ markets, gift and gourmet shops and health food stores (Petersen, 1991). No significant relationships between purchase point for honeys and frequency of use, production location, floral source, or grade were discernable with chi-square analysis ( p > 0.05). There was a trend, although nonsignificant, toward the increasing importance of production location and grade when the respondents purchased food as gifts. More than half (66.1%) of these panelists purchased food as gifts and more than a quarter (25.8%) had received honey as a gift.
Perception
of honeys
All of these panelists agreed or strongly agreed that honey has a unique flavor. Most agreed that the honey improves food flavor. They were divided on its effects on food texture (Table 3). In national surveys, consumers in the USA claim to use honey for both textural and flavor effects (National Honey Board, 1989). By far, the majority of these panelists disagreed with the statement that honey was difficult to use in cooking, although they agreed that it was messy to use. They also indicated that storage of honey was not a problem (Table 3). US consumers sampled nationally report that honey is added to a wide array of home prepared food items (National Honey Board, 1989). Less than 20% of these panelists disagreed with the statement that honey is better for you than sugar, indicating that honey had a somewhat healthy image among this group despite the perception by most of them that it is high in calories. This result mirrors the opinions of US consumers sampled nationally (National Honey Board, 1989). These panelists were not told that a synthetic product was included among the samples evaluated. In national surveys in the USA, artificial honey is regarded with strong disfavor by consumers (National Honey Board, 1989). Honey has a nutritious image and an artificial product is perceived to lack positive nutritional attributes. Panelists in this study indicated that nutrition was an important quality criterion in food selection in general (Table 5). If these panelists were also to react negatively to an artificial honey product, this impression may have overridden the effect of price in ranking of the synthetic ‘honey’ for willingness to buy. This further suggests the usefulness of survey data in interpreting the results of sensory tests.
CONCLUSIONS Relative preference for the honeys evaluated differed; panelists tended to agree among themselves regarding the overall ranking of the samples. Price may also influence their selection in the marketplace, especially among samples perceived to be similar in acceptability. Product characteristics which typically distinguish premium from table honeys with the exception of flavor were unimportant when these panelists selected among available products. Many panelists were unaware of label information that could help identify a preferred flavor. The R-index and rank sums analyses provided similar information on relative acceptability and willingness to buy these honeys. Both were useful in assessing the effects of price information on willingness to buy similar honey products. The R-index did not provide more information about the relative acceptability and willingness to buy the samples evaluated than was found in the rank sums analysis. Information obtained from the survey data, in some cases, allowed the sensory data from the acceptability and willingness to buy analyses to be qualified and substantiated. This is important as consumers are increasingly using factors other than relative quality and price to make marketplace decisions. The usefulness of these tests in assessing the relative preference of honeys has been demonstrated.
ACKNOWLEDGEMENTS Thanks are expressed to Stephen Petersen, Toklat Apiaries, Fairbanks, AK for providing the honey samples used in this study; Mary Wright provided the synthetic product evaluated. The able assistance of Cathy Birklid in conducting the sensory panel and statistical analyses is appreciated. The refractive indices were determined with the assistance of Dr Mary Keskinen, Department of Geology/Geophysics, University of Alaska Fairbanks.
REFERENCES Daniel, W. W. (1990). Applied Nonparametric Statistics, 2nd edn. PWSKent Gramstad, Research
Publishing
Company,
M. (1991).
Personal
Center,
Fairbanks
banks, AK.
Boston,
MA.
communication. North
Star
Community Borough,
Fair-
Premium Honeys Hoff, F. L. & Phillips, J. K. (1989). Honey: background for the 1990 Farm Legislation. Commodity Economics Division, Economic Research Service, US Department of Agriculture Staff Report No. AGES 89-43, Washington, DC. Horwitz, W. (ed.) (1975). O&ial Methods ofAnalysis of the Association of Oficial Analytical Chemists. Association of Official Analytical Chemists, Washington, DC. Humann, M. (1991). Personal communication. National Honey Board, Longmont, CO. National Honey Board (1989). Consumers love honey. Honey Hotline 4. O’Mahony, M. (1985). Sensory Evaluation of Food: Statistical Methods and Procedures. Marcel Dekker, New York, pp. 38997. Petersen, S. F. (1989). Beekeeping under the northern lights, American BeeJournal, 129(l), 33-5. Petersen, S. F. (1991). Personal communication. Interior Alaska Beekeepers Association, Fairbanks, AK.
221
SAS Institute, Inc. (1990). SAS User? Guide, Statistics. SAS Institute, Inc., Raleigh, NC. Sloan, A. E. (1985). Chemical confusion. Food Engineering, 57(9) 72. Tram, A. S., Pangbom, R. M. & Little, A. C. (1981). Potential fallacy of relating hedonic responses with physical and chemical measurements. Jownal of Food Science, 46, 583-8. Vie, A., Gulli, D. & O’Mahony, M. (1991). Alternative hedonic measures. Journal ofFood Science, 56, l-5,46. Watts, B. M., Ylimaki, G. L., Jeffery, L. E. & Elias, L. G. (1989). Basic Sensory Methods for Food Evaluation. International Development Research Centre, Ottawa, pp. 65-7. White, J. W. (1975). Physical characteristics of honey. In Honqr A Comprehensive Survey, ed. E. Crane. Heinemann, London, pp. 207-39. White, J. W. (1978). Honey. Advances in Food Research, 24, 288-354.