Influence of seasoning on vegetable selection, liking and intent to purchase

Low vegetable intake continues to be a health concern, and strategies to increase vegetable intake have resulted in only small increases. One strategy that has received less attention is the use of seasonings. This study's objective was to determine the impact of seasoning on vegetable selection, liking, and intent to purchase. We conducted a 3-week study in a public café on a university campus. Customers buying a main dish could select a vegetable side (seasoned [SS] or steamed [ST]) at no cost. Based on café data and power analysis (alpha 0.05, 80% power), 2 days per vegetable pair were conducted with carrot, broccoli, and green bean pairs randomized 3 days/week 1 and 3, with normal service week 2. Selection was greater for SS vs ST, n = 335 vs. 143 for all 3 vegetables combined; n = 97 vs 47 for carrots; n = 114 vs. 55 for broccoli; n = 124 vs. 41 for green beans (p < 0.001 Chi-Square). Liking responses were similar for SS vs ST and were high for all vegetables. Response distribution was not significantly different for SS vs ST vegetables when people were asked if they would purchase the vegetable that they selected. More customers chose the 'somewhat likely' and 'very likely' (n = 353) than the 'not likely' and 'definitely would not' (n = 121) purchase responses. Regression showed that people who did not often consume a vegetable with lunch while dining out were 1.59 times more likely to select the SS vegetables over the ST (p = 0.007). Given a choice, consumers were more likely to select a seasoned vegetable. With low vegetable consumption as a predictor of seasoned vegetable choice, offering seasoned vegetables may increase intake in those with poor vegetable intake in a café setting.


Introduction
Vegetables are an important part of any dietary pattern. Because vegetables are typically high in nutrients and low in calories, they can play an essential role in health promotion and disease prevention (Center for Disease Control, 2015). Increasing consumption of vegetables and fruits has been shown to aid in prevention of Coronary Heart Disease (CHD). A meta-analysis of 12 studies with 13 cohorts concluded that increased intake of vegetables and fruits from under three servings per day to more than five servings per day lowered the rate of developing CHD by 17% (He, Nowson, Lucas, & MacGregor, 2007). Similarly, a systematic review of six studies found that a higher intake of dark leafy green vegetables was associated with a 14% decrease in Type 2 Diabetes Mellitus (Carter, Gray, Troughton, Khunti, & Davies, 2010). In addition to disease prevention, dietary patterns rich in vegetables and fruits have been associated with lower body mass index (BMI). Data from the 2007 Behavioral Risk Factor Surveillance Survey showed that only 23.9% and 21.9% of those classified as overweight and obese, respectively, consumed five or more servings of vegetables and fruits per day, compared to 27.4% of those who were of normal BMI. This negative association between BMI and vegetable and fruit intake remained significant after adjustment for confounding variables (p < 0.0001) (Heo et al., 2011). According to the World Health Organization (WHO), 5.2 million deaths worldwide in 2013 were attributed to diets lacking in vegetables and fruits (WHO, 2015).
Incorporating vegetables into a dietary pattern can take several different forms. Vegetables can be fresh, frozen, canned, juiced, or dried. In the United States (US), people can access vegetables by growing them or purchasing vegetables from a grocery store, farmer's market, restaurant, or other retail outlet. In the US, 65% of vegetables are consumed as a base dish, or "as is". This includes vegetables used as side dishes, snacks, or the entire meal. The next most popular way (29%) to consume vegetables is as an ingredient in another dish. This includes vegetables used in stews, casseroles, and sauces (Produce for Better Health Foundation, 2015).
Daily intake recommendations for vegetables by adults that engage in at least 30 min per day of exercise are 2e3 cups (Center for Disease Control, 2015). These recommendations are dependent on age, sex, and physical activity. Vegetable consumption trends in the US have not been promising over the last few years. The usual quantity of vegetables consumed for Americans at the 75th percentile of intake is two cups (USDHHS, USDA, Dietary Guidelines for Americans, 2015). Eighty-seven percent of Americans did not meet the daily vegetable recommendation set for their age-sex group. These numbers are even higher for adolescents and young adults (Castenson, Dodd, Krebs-Smith, Parsons, & Reedy, 2015). According to the State of the Plate study on America's consumption of fruits and vegetables (FV), the total amount of vegetables consumed from the year 2009e2014 decreased by 7%. When looking at consumption trends over the last decade, fresh vegetables purchased from a retail outlet were the only form of vegetables to show a growth, while intake of canned, frozen, dried, and juice vegetable forms declined (Produce for Better Health Foundation, 2015).
With poor intake of vegetables but known health benefits from greater consumption, allied health professionals are faced with the problem of how to increase vegetable intakes. Erinosho, Moser, Oh, Nebeling, and Yaroch (2012) evaluated the exposure of the More Matters Campaign e a program designed to increase awareness of FV recommendations and actual consumption in adults. Individuals (n ¼ 3021) who consumed the recommended five or more FV servings per day were 1.33 times more likely to have heard about the More Matters Campaign, or the 5-A-Day Campaign, and were 1.55 times more likely to know about current FV recommendations. However, it remains unknown whether increased exposure to FV campaigns increases intake, or if adults who already consume more FV are more aware of FV messages. Indeed, a review of vegetable and fruit promotional campaigns has found results to be low to modest at best (Rekhy & McConchie, 2014). Ungar, Sieverding, and Stadnitski (2013) examined methods to increase FV intake by assigning groups the goal of eating "5 servings a day" (5/day group) versus "1 more serving than you already eat" per day (þ1/day group); the control group was instructed to eat as usual. A one-week randomized, controlled intervention was conducted with mostly female students (n ¼ 84, 85% female). Contrary to expectations, the 5/day group was more effective at increasing their FV consumption. The þ1/day group consumed an average of 3.41 servings (SD ¼ 0.96) compared to the 5/day group at 5.00 servings per day (SD ¼ 0.70). Of note was that FV intake declined to 3.45 (SD ¼ 1.10) servings per day in the 5/day group at one week after intervention compared to 2.72 (SD ¼ 0.84) in the þ1/day group.
In addition to education campaigns, the impact of price on vegetable consumption has been explored. Cost has been reported as a barrier to adequate vegetable consumption (Skuland, 2015). Smith-Drelich, 2015 lowered this barrier by providing reimbursement of up to $50 per week for vegetable purchases in an intervention group (n ¼ 144). While more money was spent on vegetables [þ$8.16 (SD ¼ $2.67)/week, p < 0.01], vegetables consumption did not increase as a result of reimbursement [þ1.3 (SD ¼ 1.2) servings/week]. More recently, price was not found to be a barrier to FV consumption in a national Scottish survey (Mc Morrow, Ludbrook, Macdiarmid, & Olajide, 2016). These findings suggest that financial barriers may only be a minor contributor to inadequate vegetable consumption, although cost may be more influential in food insecure households (Mook, Laraia, Oddo, & Jones-Smith, 2016).
Other research suggests that improving the palatability, or taste, of vegetable or vegetable-based dishes may improve intake. Ghawi, Rowland, and Methven (2014) examined whether herbs and spices could be used to enhance liking of low sodium tomato soup. Adults (n ¼ 148) participated in this five consecutive day study, where they were given three samples of tomato soup: regular salt; low salt; and low salt enhanced with seasonings. After repeated exposure to the three soup variations, participants reported increased overall liking (p < 0.02), flavor liking (p < 0.02), texture liking (p < 0.01), and aftertaste liking of the low salt enhanced with seasonings soup. No significant changes with repeated exposure to the other two soup variations were identified. However, Wang, Lee, and Lee (2014) found that while some level of herbs in tomato soup increased liking, high levels decreased liking. Tomato soup with herbs did lower the amount of salt needed to result in consumers perceiving that the soup tasted right. Other studies with children have found that seasoned dips can influence vegetable intake (Fisher et al., 2012;Savage, Peterson, Marini, Bordi, & Birch, 2013), although this strategy applied to raw vegetables and may increase calorie or dietary fat intakes. Peters, Polsky, Stark, Zhaoxing, and Hill (2014) examined the effects of seasoning foods on restoring the liking of reduced fat (RF) food items. Participants (n ¼ 148) were fed three meals in a randomized order on three different days. One meal was a full fat (FF) meal, one was a RF version of the same meal, and one was a RF version of the meal with the addition of herbs and spices. Reducing the fat content significantly lowered the rating of meal liking (FF ¼ 7.05 vs. RF ¼ 6.29, p < 0.0001). However, the RF seasoned meal was liked as well as the FF condition. Individual food items revealed no significant differences between liking distribution of meatloaf and vegetables between the FF and RF seasoned options.
The use of herbs and spices (i.e., seasonings) to enhance vegetable liking is an area of research that needs further attention. With enhanced liking, we expect that vegetable selection, willingness to purchase, and intake will also increase. This may help individuals reach their daily vegetable recommendation in a way that seems effortless and enjoyable. The purpose of this study was to determine whether seasoned vegetables would be selected more often than unseasoned (plain) vegetables. Secondary aims were to determine whether intention to purchase seasoned vegetables would be reported more often than intention to purchase unseasoned, and whether consumption of seasoned vegetables would be greater than unseasoned vegetables.

Design
This was an observational cross-sectional study over a threeweek period (NoveDec 2015), with two testing weeks and one wash out week in between to ensure that each vegetable tested had the same entr ee pairing. Carrots, green beans, and broccoli were selected as test vegetables, based on their high consumption frequency by US adults (Produce for Better Health Foundation, 2015). One vegetable was offered per test day as both a seasoned and unseasoned choice, and vegetables were randomly assigned to a day of the week. Industry experts in culinary sciences developed the seasoning blends that were unique for each of the vegetables tested in the caf e. Data were collected on previous vegetable purchase patterns for comparison to selection in the current study. The University Institutional Review Board approved this study; written consent was not required for this exempt protocol (IRB #16360).

Previous purchase patterns for vegetables
From purchasing data collected from the previous school year (SepteMay 2015), mean broccoli purchase was 39 servings/day over a 30-day period, (range 15e64). Mean green bean purchase was 23 servings/day over 25 days, (range 15e34). Mean carrots purchase was 17 servings/day over 36 days (range 8e33).

Participants
All customers purchasing a hot entr ee in the caf e, located on a large, Midwestern campus, were offered a seasoned or unseasoned vegetable at no additional cost. According to historical purchasing data, 40% of diners choose a vegetable; thus, we estimated a 20% difference in choosing the seasoned versus the unseasoned vegetable. For 83% power, alpha 0.05, and effect size of 20%, 140 diners were needed.

Procedures
The seasoned and unseasoned carrots were steamed for five minutes and tossed with soybean oil (14 g) and salt (1.5 g). Additionally, seasoned carrots received 0.48 g of ground cinnamon. Seasoned and unseasoned broccoli was steamed for four minutes and tossed with soybean oil (14 g) and salt (3 g). Seasoned broccoli received 0.42 g of garlic powder, 0.27 g of onion powder, 0.42 g of dill herb, and 0.27 g of black ground pepper. Green beans were steamed for five minutes and tossed with soybean oil (9 g) and salt (3 g). Seasoned green beans received 0.42 g of garlic powder, 0.27 g of onion powder, 0.42 g of dried parsley, and 0.27 g of ground black pepper. All vegetables were prepared in 2-lb (907.2 g) batches. Once prepared, the vegetables were held on a steam table until portioned into 4-oz servings (113.4 g). All vegetables were held for no longer than 15 min on the serving line.
Upon entry into the caf e, customers saw a display for the day's food selections. As part of the selections, the seasoned and unseasoned vegetable dishes were displayed side by side with their respective ingredients under their name. Vegetables were labeled "seasoned" and "steamed" (for the unseasoned option) followed by the vegetable name. Although both seasoned and unseasoned vegetables were steamed, the term "unseasoned" was not used as a label to avoid an upward bias toward the seasoned vegetable option.

Selection, liking, and intent to purchase questionnaire
The research team developed an 11-item questionnaire that was designed to assess: vegetable and seasoning status, vegetable liking, and usual intake of the vegetable, usual intake of vegetables generally at lunch, likelihood to purchase the vegetable, frequency of eating at the caf e, and demographics. If the customer refused a vegetable, the questionnaire instructed them to disregard questions on liking and likelihood to purchase the vegetable. The questionnaire was designed to be brief; thus, single-question construction was selected such that Cronbach alpha could not be calculated for reliability testing. Customers received a questionnaire upon hot entr ee purchase, regardless of their vegetable choice or refusal. Participants were not identified through the questionnaire. Completion of the questionnaire was not mandatory, and participants had the option to skip any question(s).

Vegetable service and waste data collection
Four-oz portions (113.4 g) of each vegetable were weighed on calibrated, digital scales before service. This serving size was consistent with a vegetable side on a regular production day in the caf e and with standard portion sizes (oz equivalents) for one serving of the selected vegetables. Waste was collected in aggregate but divided according to bowl content of either seasoned or unseasoned vegetable. A day dot sticker was placed on the underside of the seasoned vegetable bowls for identification purposes. The total number of bowls included in waste weight was counted. Total vegetable consumption was calculated by taking total vegetable weight distributed minus total waste weight; consumption was calculated separately for seasoned and unseasoned vegetables.

Analyses
All statistical analyses were conducted using SPSS (version 23, IBM, 2015). Descriptive statistics assessed demographic information and survey completion rate. Chi-Squared tests were used to determine differences in response distribution of intentions to purchase by vegetables and preparation methods. Correlations were completed to determine the associations between variables to find influences to seasoned vegetable selection, using Lambda for nominal to nominal evaluations and Gamma for nominal to ordinal. Logistic regression was conducted to identify predictors of and their relative influence on vegetable preparation selection.

Vegetable consumption and waste
Descriptive data for vegetable consumption and waste are presented in Table 1. All vegetables, regardless of preparation method (seasoned versus unseasoned), yielded less than or equal to than 20 g of waste per bowl (Table 1). The vegetable and preparation method that produced the least waste was seasoned broccoli, with an estimated average of 5 g of waste per bowl. The vegetable and preparation method that yielded the most waste was seasoned carrots, averaging 20 g of waste returned per bowl.

Survey results
During the study, 752 people received the 11-item questionnaire. Few surveys (n ¼ 5) were excluded from analysis as they declined a vegetable but proceeded to rate vegetable liking. After exclusion, survey response rate was 70.6% (n ¼ 536). Most customers were students (n ¼ 295, 55%), or university faculty/staff (n ¼ 222, 41.4%), the rest were not affiliated with the university or other (n ¼ 15, 2.8%, n ¼ 4 did not report occupation), with an even distribution by gender (n ¼ 265 females, 265 males, n ¼ 6 did not report gender).
There were no significant differences in vegetable liking for seasoned and unseasoned vegetables. Most respondents indicated they liked the vegetables "somewhat" or "very much" (liking of carrots ¼ 79%, broccoli ¼ 96%, green beans ¼ 93%). Furthermore, Chi-Squared tests revealed no significant differences between liking and preparation for individual vegetables (carrots p ¼ 0.965, broccoli p ¼ 0.470, green beans p ¼ 0.558).
Likelihood of purchase was dependent on vegetable type (p ¼ 0.002; Table 2); respondents indicated the highest likelihood of purchase for broccoli (84% said they were somewhat likely or very likely to purchase), followed by green beans (74%) and carrots (64%). Among seasoned vegetables, a Chi-squared test revealed there were significant differences in likelihood of purchase (p ¼ 0.003); particularly, fewer respondents indicated they were likely to purchase the seasoned carrots relative to the seasoned broccoli or green beans. However, we do not observe differences in likelihood of purchase for unseasoned vegetables.
Linear regression was used to further evaluate the likelihood to purchase. Table 3 includes predictors of the likelihood to purchase, indicating that 28% of the variance in likelihood to purchase was accounted for by which vegetable (carrot, broccoli, green beans), age, liking of the vegetable, frequency of eating a vegetable when eating lunch out, and frequency of eating that particular vegetable. Respondents expressed a higher likelihood of purchase for broccoli relative to carrots; however, there was no difference between green beans and carrots. Older consumers indicated a higher likelihood of   purchase relative to their younger counterparts. In addition, likelihood of purchase was positively related to: liking of the vegetable, frequent eating of vegetables generally with lunch, and frequency of eating the specific vegetable in question.

Influences on vegetable preparation choice
Binary logistic regression was conducted with vegetable intake during lunch, along with gender and age as they were also significantly associated with seasoned vegetable selection. Table 4 includes likelihood of consuming a vegetable with lunch when eating out, gender, and age as predictors for seasoned vegetable selection. All three variables were significant predictors of seasoned vegetable selection. Participants who often consumed a vegetable with their lunch were less likely to consume the seasoned vegetable (OR ¼ 0.630, p ¼ 0.007). After reverse coding, those who reported low vegetable intake during lunch were 1.59 times more likely to select the seasoned vegetable. Male gender was a significant predictor of seasoned vegetable selection (OR ¼ 1.177, p ¼ 0.006). Finally, age under 50 years was a significant predictor of seasoned vegetable selection (OR ¼ 0.580, p ¼ 0.026). The Hosmer and Lemeshow Test shows a good fitting model (p ¼ 0.607). The Cox and Snell test provided a pseudo R 2 that suggested our model accounted for 5% of the variance in vegetable selection.

Discussion
More vegetables were selected in this study than in the previous year's purchase data. Most likely this is related to cost factors, as this study offered vegetables at no cost and data from the previous year was for purchased vegetables. Studies using vouchers for produce or rebates have found increased fruit and vegetable consumption or purchase (Cohen et al., 2017;Phipps et al., 2015).
Negligible waste supports that both preparation styles were well consumed, without offering dip (Fisher et al., 2012;Savage et al., 2013) or changing the nutrient profile of the dish as in previous studies (Ghawi et al., 2014;Peters et al., 2014). Although these previous studies were conducted with children, studies with adults' vegetable intake with dips or changed nutrient profile of the dish have not been published to the authors' knowledge.
An interesting finding was that those consumers who did not habitually purchase a vegetable with their lunch were more likely to select the seasoned vegetables. Seasoning of vegetables may be instrumental in encouraging those consumers who do not often consume vegetables to select and ingest vegetables more regularly. This consumer preference for seasoned dishes may be related to consumer perception of additional flavor associated with seasoned food (Peter, 2006). When several factors of food intake behaviors of Americans were examined, taste was reported as the most important deciding factor (Glanz, Basil, Maibach, Goldberg, & Snyder, 1998;Kourouniotis et al., 2016). While consumers on Day 2 of the current study may have also tasted the vegetable of choice during the previous week of the study, the inclusion of all choices regardless of repeat customers has been well accepted in cafeteriarelated literature, although study duration has varied (Arsenault, Singleton, & Funderburk, 2014;Wansink, Cao, Saini, Shimizu, & Just, 2012).
Research has shown that older adults and women are more likely to consume vegetables (Hiza, Casavale, Guenther, & Davis, 2013). Interestingly, women and older adults weigh health as a more important factor influencing food choice (Glanz et al., 1998). We found that male gender and consumers under the age of 50 years were more likely to select seasoned vegetables compared to their demographic counterparts. This suggests that consumers who do not regularly consume vegetables may be responsive to seasoned vegetable products and behavioral strategies promoting seasoned vegetables. In regards to FV consumption, both health and taste have been found as predictors to consumption (Glanz et al., 1998;Krebs-Smith et al., 1995). Groups who perceive nutrition as a more influential driving force for vegetable consumption may not benefit from seasoning changing the perception of flavor as it is not the primary driving force for consumption.
Habitual Intake of the Vegetable-(0 ¼ never, 1 ¼ less than once per week, 2 ¼ at least once but less than twice per week, 3 ¼ more than twice per week). Note: Effects of Broccoli and Green Beans should be interpreted relative to the carrot vegetable type.  , 2015). Aside from age and gender, race has demonstrated an influence on frequency of seasoning use on vegetables as well as seasoning type used, with Asians and Pacific Islanders using seasoning more often than other races (Nikolaus, Ellison, Heinrichs, Nickols-Richardson, & Chapman-Novakofski, 2017). Unfortunately, we did not include race in our analysis. The type of vegetable (broccoli, green beans, or carrots) had an impact on the likelihood to purchase, possibly because of inclusion of carrots. Although annual intake of carrots is increasing and green beans declining, ranking of carrots nationally usually includes vegetables eaten "as is" or raw (Produce for Better Health Foundation, 2015). Furthermore, seasoned carrots yielded twice as much waste as seasoned green beans and three times as much waste as seasoned broccoli, despite seasoned carrots being selected slightly less often than the other seasoned vegetables. This suggests that the seasoning used for carrots was not as well accepted by consumers. Green beans and broccoli outrank carrots in national data of side dishes eaten at home (Produce for Better Health Foundation, 2015), and broccoli was cited as a favorite vegetable in focus groups (Heinrichs, Nikolaus, Ellison, Nickols-Richardson, & Chapman-Novakofski, 2016). This is also reflected in our data including frequency of eating this vegetable in the overall regression of likelihood to purchase. Vegetables are most often eaten at dinner rather than lunch and at home rather than out. Indeed, vegetables at restaurants account for only 10% of vegetables eaten, offering opportunity to expand vegetable intake growth (Produce for Better Health Foundation, 2015). Studies have reported that adults consider vegetable offerings in restaurants to be minimal (Heinrichs et al., 2016;Lucan, Barg, & Long, 2010).
It is not surprising that vegetable liking was a significant predictor in the model for likelihood to purchase. However, seasoned or unseasoned was not, although more consumers selected the seasoned option. In focus groups probing both vegetables and seasoning, participants did not suggest seasoning vegetables as a strategy to increase vegetable intake, although discussions had recently included seasoning of foods (Heinrichs et al., 2016). It is possible that seasoning or preparation methods are latent variables for liking of the vegetable, but that the vegetable itself should be liked first. While it is possible that seasoning could move neutral or dislikers to likers of a vegetable type, this study did not explore this possibility.
The strength of this study is that it was conducted in a freeliving setting, similar to what might be encountered in any worksite or commercial cafeteria. In addition, the serving bowls, environment, lighting, and service, which may all affect selection (Hadi & Block, 2014), were the same across vegetable choices.
However, there were a few limitations to the current study. Vegetables were selected for this study based on high consumer intake. However, only three vegetables were tested, limiting generalizability. Moreover, potatoes and tomatoes were not tested in this study, although they are consumed at higher rates than our study vegetables (Produce for Better Health Foundation, 2015).
A statistical limitation was the lack of power to distinguish between vegetable preparation liking distributions. Regardless, both the seasoned and unseasoned vegetables were well liked with the majority of responses in the "somewhat like" and "very much like" category. Although no measures were taken to control for social desirable answers for liking, negligible waste supports the high rankings on this measure. In addition, anonymous self-report and placing surveys in a response box as in this study, provides neutrality and detachment. Unfortunately, power was not achieved for waste collection. Vegetable waste was collected in aggregate form at the end of service each day. This led to only two entries for each vegetable and preparation method, and thus, lack of variability. One concern was that because the vegetables were provided free of charge, customers would select a vegetable and not consume it. However, all vegetables regardless of preparation method yielded less than or equal to 20 g of waste per bowl. In addition, most plates were returned without any waste. Few plates may be responsible for positively skewing the waste collection quantities. In future studies, each plate should be weighed individually.

Conclusions
Unlike intervention studies that involve testing the effect of campaigns or goal setting strategies, this was a realistic observational study. This study did not provide an incentive for participating; however, the cost barrier to vegetable accessibility and affordability was removed. Results demonstrate that when given a choice, consumers will more likely select seasoned vegetables over their unseasoned counterparts in a typical cafeteria setting. Liking and intent to purchase responses for both seasoned and unseasoned vegetables were high. Finally, predictors for seasoned vegetable choice were male gender, age less than 50 years, and low vegetable consumption and for overall intention to purchase, type of vegetable (carrot, broccoli, green beans), age, liking of the vegetable, frequency of eating a vegetable when eating lunch out, and frequency of eating that particular vegetable.

Funding
Funded by an unrestricted gift from the McCormick Science Institute.