Increasing Consumer Acceptance of Baby Leafy Greens Grown in a Controlled Environment

Results

1.0. Taste Study

A consumer test was conducted to determine consumer acceptance of six varieties of baby greens. Approximately 85% of consumers reported consuming leafy greens at least once per week. Most participants were between the ages of 18 and 34 (77%), had high food adventurousness (84%), and were medium or high variety seeking (81%). Table 3 shows the mean differences for the sensory attributes of each sample. Mean liking scores differed significantly (p<0.01) across samples in overall liking, flavor, and texture. No differences in aroma were observed between the samples.

Mean hedonic scores for liking are shown in Table 3. Across all sensory attributes, Pac Choi was consistently rated as the highest liked sample. Arugula was the least liked sample overall liking and in flavor liking. Differences in liking of flavor and overall liking followed similar patterns, where both Pac Choi and Red Romaine were liked the more than Arugula; Ruby Streaks, Green Kale, and Red Kale fell between those samples, respectively.  Purchase intent closely followed patterns for both flavor liking and overall liking. The texture of Pac Choi, Arugula, and Green Kale were liked more than that of Ruby Streaks.

Differences were observed in the CATA data for liking/dislike of color and spiciness, liking of sweetness, and dislike of bitterness (Table 4). A higher proportion of participants disliked the color of Red Kale (29%) Ruby Streaks (18%) compared to Arugula (1%) and Pac Choi (1%). More participants recorded liking sweetness of Red Romaine (52%) than all other samples, and more reported disliking the bitterness of Arugula (55%) than Ruby Streaks (23%), Pac Choi (26%), and Red Romaine (16%). Spiciness was liked, and also disliked, by more participants in Ruby Streaks (36%; 19%) and Arugula (34%; 15%) than all other samples.

1.2. PROP Classification

Participants were screened for their sensitivity to the compound PROP with 21% of participants (n=15) classified as non-tasters, 49% (n= 36) classified as medium tasters, and 30% (n=22) classified as supertasters.  This distribution was consistent with normative data from similar populations and frequencies observed in previous studies (Bartoshuk, Duffy, & Miller, 1994; Tepper et al., 2001; Tepper & Ullrich, 2002; Zhao et al., 2003). No differences in liking for any sensory attributes were observed across taster groups.

1.3. Cluster Analysis

AHC grouped consumers into three clusters based on overall liking. Within class variance represented 61.8% of the variance. Within class variance ranged from 12.2% for Cluster 1 (n=38) to 20.7% for Cluster 3 (n=13). Cluster 3 was most dissimilar to the other clusters, according to the dendrogram of the AHC analysis.

Figure 2 shows the mean flavor liking ratings (± SEM) for each cluster, grouped by variety. Generally, Cluster 1 gave high liking ratings to all samples, each falling within the ‘Liking’ range. Liking patterns for Cluster 2 reflected those of Cluster 1 apart from Arugula, in which Cluster 2 disliked the sample. Cluster 3 generally disliked the samples but had higher variability in ratings compared to the other clusters. Arugula was liked the most by Cluster 1 compared to Cluster 2 and Cluster 3, whose liking of arugula flavor was within the ‘Dislike’ range. For Ruby Streaks, Green Kale, and Pac Choi, both Cluster 1 and Cluster 2 gave similarly high liking ratings compared to Cluster 3, whose liking ratings were in the ‘Dislike’ range. For Red Kale, Cluster 1 gave the highest liking compared to Cluster 3, while ratings for Cluster 2 fell between the other two clusters.

Fisher’s exact test of independence revealed relationships (p<0.05) in the consumer clusters for income, ethnicity, and consumption frequency. Cluster 3 contained higher than estimated frequencies of Asian consumers and consumers whose household income was less than $70,000 per year, while Cluster 1 contained higher than estimated White/Caucasian consumers and consumers whose household income was greater than $70,000 per year. No differences were observed between clusters for PROP status, age, gender, food adventurousness, variety seeking, or any TFEQ-R18 category (Cognitive Restraint, Emotional Eating, and Uncontrolled Eating).

2.0. Online Survey

2.1. Demographics

A total of 862 participants attempted the study; 374 did not qualify, and an additional 89 began the survey but did not complete it. A total of 399 consumers completed the online survey based on self-reported consumption of leafy greens at least once per month. Post hoc, 2 participants were classified as ‘careless responders’ and removed from the data set based on the speed of response.

Of the respondents who completed the survey, 55% identified as female. The largest ethnicity demographic was Caucasian (46%). In household income, 39% earned less than $50,000 per year, 34% earned between $50,000 and $90,000, and 26% earned more than $90,000. The participants were primarily highly food adventurous (69%) and were responsible for grocery shopping some of the time or always (73%).

2.2. CBC Analysis

Data from the CBC attributes are reported as importance scores; higher percentages indicate a more important attribute. When all participant data was grouped, consumers placed the most importance on flavor, followed by leaf texture, and then color. Within each attribute, individual utility scores for each level were analyzed. Higher utility scores represent the more attractive level within each attribute. Overall, consumers preferred leafy greens which were described as sweet, were of mixed colors and mixed leaf textures.

2.3. Cluster Analysis of CBC Data

Four clusters were identified based on individual utility scores: Cluster 1 (n=88), which placed importance on both color, then flavor; Cluster 2 (n=80), which placed importance on texture, then color; Cluster 3 (n=100), which placed importance on flavor; and Cluster 4 (n=129), which placed similar importance on color and texture. Figure 4 shows the average importance scores for each cluster. Table 5 shows the breakdown of utility scores for each cluster.

Within Color, Cluster 3 placed the most importance on the flavor description, followed by Cluster 1, Cluster 4, and Cluster 2. Within the color attribute, Cluster 1 placed the more importance on sample color compared to Cluster 4 and Cluster 2, which had similar importance scores for color. Cluster 4 placed the lowest importance on color overall. For leaf texture, Cluster 2 placed the highest importance on visual texture appearance, followed by Cluster 4. Clusters 1 and 3 placed similar importance on texture, and had lower importance than both other clusters.

A large group (Cluster 4; n=129) placed high importance on visual characteristics, with high utility placed on greens which had flat texture and those which were only green in color. Flavor was of less importance to this group, and utility values for both levels were very close to 0, with higher utility for greens described as sweet. A group which made most of their decisions based on the flavor of the product concepts (Cluster 3; n=100) placed 62% importance on the flavor description of the sample. This cluster was attracted to samples which were described as sweet as compared to those described as spicy. This cluster placed equally low importance on visual attributes, and placed preference on mixtures of color and texture.

A smaller cluster (Cluster 1; n=88) placed most importance on color and flavor, respectively. Of these attributes, higher utility was placed on greens that were a mixture of green and purple, and those that were described as spicy. Leaf texture was of less importance, and this group was more attracted to mixes that varied in texture. The smallest cluster (Cluster 2; n=80) placed relatively equal importance on both texture and color, more often selecting greens which were mixed color and mixed texture.

Fisher’s exact test showed demographic differences between the clusters in age (p<0.05). Clusters 2 and 3 contained a higher frequency of participants whose age was above 55 years old, while Cluster 4 contained a lower frequency of participants in this age range. A trend (p=0.09) in food adventurousness was observed between the consumer segments.

Additionally, there were significant correlations between bitter food liking score and utility scores for spiciness and sweetness (r² = 0.2, p<0.01). In the CATA questions for liking of sensory attributes, Cluster 1 had a high frequency of consumers who liked spicy and pungent characteristics. Cluster 3 had a high frequency of liking sweet and mild characteristics, and lower frequency of liking spicy and bitter characteristics. Cluster 4 had a higher level of participants liking bitterness and toughness, while a lower frequency of reported liking for crispy leaves. There were no differences in sensory attribute liking for Cluster 2.

The MFA biplot (Figure 4) shows the associations between utility and importance scores, demographic information, and clusters. The analysis explained 95.7% of the variance. Factor 1 (62.7%) shows the variation between Cluster 2 and Cluster 4. These two clusters had high utility scores for color and texture appearance characteristics. Cluster 4, which placed high utility on flat leaves with green color, was associated with disliking of crispy leaves, low consumption frequency, low household income (≤$50,000 per year), and low food adventurousness (rarely try new foods). Cluster 2, which placed high utility on mixed (green/purple) color and mixed (flat/curly) texture. This cluster was associated with disliking tough leaves, high household income (≥$90,000 per year), and high consumption of leafy greens (4-6 times per week).

Factor 2 of the MFA (23.0%; Figure 4) shows the variation between Cluster 1 and Cluster 3. Cluster 1 placed high importance on color and leaf texture appearance, and was characterized as having high utility for spicy greens. Cluster 1 was associated with reported liking of spiciness, having a high bitter food liking score, being male, and having high food adventurousness (try new food most of the time). Cluster 3 placed very high importance on flavor and had high utility for sweet greens. This cluster was associated with dislike of spiciness and liking of sweetness, and female gender.

Discussion

1.0. Taste Study

In this study, we conducted an in-lab test to assess consumer liking of the sensory characteristics of a variety of baby leafy greens. Overall, most of the baby leafy greens were well liked in all sensory attributes tested, however we found differences in liking between the varieties. Liking of flavor was most closely associated with overall liking.

Generally, the mild/sweet greens, Pac Choi and Red Romaine, were liked the most over the spicy and bitter/astringent greens. This is consistent with other reports on liking of leafy vegetables, where Fouladkhah et al. (2011) used lettuces as a control against new leaf cultivars. The control was liked as much or more in flavor liking and overall liking compared to the other samples. Traditionally, lettuces have been widely available across the United States and are the most popular among consumers (Tarwadi & Agte, 2003). The Red Romaine baby lettuce tested in the present study has a relatively simple sensory profile (O'Brien et al., 2021), and familiarity with its flavor may positively influence liking (Köster & Mojet, 2016).

The Green Kale and Red Kale samples generally were moderately liked and followed the same pattern for flavor liking and overall liking. However, the texture of Red Kale was not as well liked as that of the Green Kale sample. Texture is an important attribute influencing liking of mature kale cultivars (Swegarden et al., 2019), although the texture of baby kale may be more acceptable than that of mature kale (Barker et al., 2021). Additionally, many people disliked the color of the Red Kale sample. Red Kale has a very dark purple color compared to other samples in this study (O'Brien et al., 2021), which may be unfamiliar and unappealing to some consumers. Purple coloring in leaves due to anthocyanins (Zhang et al., 2014) has been associated with moderate astringency in vegetables (Martín, Navas, Jiménez-Moreno, & Asuero, 2017).

The baby greens we categorized as “spicy”, Arugula and Ruby Streaks, had different patterns of liking.  In flavor liking and overall liking, arugula was the least liked sample of all varieties, consistent with other liking studies of leafy greens (Fouladkhah et al., 2011). Differences in liking of arugula were observed between Cluster 1 and Clusters 2 and 3. Ruby Streaks, a mustard green, did not have differences in liking between the largest clusters, despite also having spicy quality. Arugula has been evaluated as having high bitterness, pungency, heat, and pepperiness, whereas mustard greens generally do not have the bitterness associated with arugula (O'Brien et al., 2021; Talavera-Bianchi, Chambers, & Chambers, 2010). Arugula and mustard have different compositions of glucosinolate compounds, which are widely accepted to drive bitterness and pungency in Brassicaceae vegetables (Bell et al., 2018).

We found no differences in aroma liking for the varieties. This is consistent both  with our findings from trained panelists (O'Brien et al., 2021) showing that the intensity ratings for aroma attributes of baby leafy greens were much lower than for flavor attributes and reports from a consumer study on kale showing no differences in aroma liking between consumer clusters (Swegarden et al., 2019). This indicates that flavor, a combination of retronasal aroma and basic taste, should be the focus of crop selection in future programs.

Cluster analysis of the hedonic scores revealed three groups. The largest differences existed between Cluster 1, which liked most samples, and Cluster 3, a small group, which disliked most samples. Cluster 2 disliked both Arugula and Red Kale, which we previously reported as having intense bitterness compared to other baby leafy green (O'Brien et al., 2021). No differences were observed between consumer groups as a function of PROP classification, indicating that PROP tasting alone cannot explain differences in liking. The cluster disliking most of the samples was composed of more consumers who had low household income and those who were Asian. Cultural differences in consumption or preparation could influence liking, and socioeconomic status is well-documented to be a determinant of vegetable consumption (Reicks, Randall, & Haynes, 1994). Due to low frequencies in some groups, it is difficult to determine trends for some demographic data.

2.0. Online Study

In the online survey, we conducted a choice-based conjoint analysis. Although flavor was the most important attribute overall, clustering showed groups which assigned different importance levels to the attributes. The largest cluster (Cluster 4) contained participants who placed similar importance on both flavor and appearance, and preferred greens that were sweet, had a single texture (flat leaf), and were only green in color – consistent with the qualities of most traditional varieties of baby leaf lettuce and/or spinach on the market. This cluster was comprised of participants who were in an older age demographic and who try new foods less often than other clusters. Another large cluster (Cluster 3) placed most importance on flavor and preferred greens which were sweet. This is consistent with previous reports that spiciness (hotness) was the main driver of disliking arugula cultivars (Bell, Lignou, & Wagstaff, 2020), and a consumer group placing high importance on flavor would avoid products with indicated spiciness.

Two smaller consumer clusters placed importance on color/flavor and color/texture (Cluster 1; Cluster 2), respectively, and these groups preferred the non-traditional sensory qualities (spiciness, mixed color, and mixed texture) in attributes of high importance. Despite color potentially being an indicator of nutritional quality, the influence of nutritional packaging claims on purchasing decision did not differ between the groups. Attention should be placed on these clusters to produce salad mix tailored to those preferring variety. Producers of leafy salad mix may decide to incorporate variety across many sensory attributes to satisfy these consumer segments. Across all participants, we observed positive associations between liking of bitter foods and high preference (i.e., utility) for products labeled as spicy. Spicy compounds, such as capsaicin, may also confer bitterness for some population segments. TAS2R38 polymorphism is associated with perception of bitterness in foods which are burning (Green & Hayes, 2003; Nolden, McGeary, & Hayes, 2016).

3.0. Integrated Insights from Taste Study and Online Study

In both hedonic consumer evaluations and online surveying, we saw similarities in consumer behavior. First, most consumers have high liking for baby greens which are sweet with minimal variation in texture from typical baby lettuces. Second, in both studies, we found that consumer clusters have high variability in their liking of spicy varieties. From the conjoint analysis, we determined that consumers who placed very high importance on flavor were averse to baby greens labeled as spicy. However, other consumers who do not find flavor to be highly important may be more concerned with textural variety when making a purchasing decision. Of consumers who preferred mixes with non-traditional sensory attributes, variety across several factors were preferred. However, producers should be conscious of extremely dark purple coloring on leaves, as one-third of consumers in the taste study reported to dislike the color of Red Kale.