Next Article in Journal
Clusters of Lactobacillus Strains from Vegetal Origins Are Associated with Beneficial Functions: Experimental Data and Statistical Interpretations
Previous Article in Journal
Species Identification of Red Deer (Cervus elaphus), Roe Deer (Capreolus capreolus), and Water Deer (Hydropotes inermis) Using Capillary Electrophoresis-Based Multiplex PCR
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Foods
Volume 9
Issue 8
10.3390/foods9080984
foods-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles thumb_up
...
Endorse textsms
...
Comment
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Near-Infrared Spectroscopy as a Beef Quality Tool to Predict Consumer Acceptance

by
Wilson Barragán-Hernández
1,
Liliana Mahecha-Ledesma
2,*,
Joaquín Angulo-Arizala
2 and
Martha Olivera-Angel
3
1
Centro de Investigación Turipaná, Corporación Colombiana de Investigación Agropecuaria (AGROSAVIA), Montería 230001, Colombia
2
Facultad de Ciencias Agrarias, GRICA research group, Universidad de Antioquia, Medellin 1226, Colombia
3
Facultad de Ciencias Agrarias, Biogénesis research group, Universidad de Antioquia, Medellin 1226, Colombia
*
Author to whom correspondence should be addressed.
Foods 2020, 9(8), 984; https://doi.org/10.3390/foods9080984
Submission received: 12 June 2020 / Revised: 6 July 2020 / Accepted: 10 July 2020 / Published: 24 July 2020
(This article belongs to the Section Meat)
Browse Figures
Versions Notes

Abstract

:
This study was conducted to evaluate the feasibility of using near-infrared spectroscopy (NIRS) to predict beef consumers’ perceptions. Photographs of 200 raw steaks were taken, and NIRS data were collected (transmittance and reflectance). The steak photographs were used to conduct a face-to-face survey of 400 beef consumers. Consumers rated beef color, visible fat, and overall appearance, using a 5-point Likert scale (where 1 indicated “Dislike very much” and 5 indicated “Like very much”), which later was simplified in a 3-point Likert scale. Factor analysis and structural equation modeling (SEM) were used to generate a beef consumer index. A partial least square discriminant analysis (PLS-DA) was used to predict beef consumers’ perceptions using NIRS data. SEM was used to validate the index, with root mean square errors of approximation ≤0.1 and comparative fit and Tucker–Lewis index values <0.9. PLS-DA results for the 5-point Likert scale showed low prediction (accuracy < 42%). A simplified 3-point Likert scale improved discrimination (accuracy between 52% and 55%). The PLS-DA model for purchasing decisions showed acceptable prediction results, particularly for transmittance NIRS (accuracy of 76%). Anticipating beef consumers’ willingness to purchase could allow the beef industry to improve products so that they meet consumers’ preferences.

1. Introduction

Beef is a high biological value food for human nutrition as a good source of protein, fat, vitamins, and minerals [1]. Regarding the main constituents, beef provides a high quantity, and disponibility of essential amino-acids [2], and an appropriate amount of protein per gram of meal required for a healthy diet [3]. As a source of fat, beef contributes to fatty acids called trans and trans-conjugated fatty acids [4]. Recently, these fatty acids have been related to potential human health improvement, like cancer amelioration [5,6,7]. However, despite these beef properties, consumers are interested in knowing the attributes of the quality of meat, especially intrinsic and extrinsic attributes. Intrinsic attributes refer to the inherent physical characteristics of meat, which cannot be altered without influencing the nature of the product and can be described by the color, visible fat, and appearance of the meat cut. In contrast, extrinsic cues are other factors associated with meat, including the price, brand, label, and production systems [8,9].
Consumer acceptance of beef quality is strongly influenced by visual perceptions and the intrinsic and extrinsic attributes of beef [9,10]. For example, meat color influences freshness perception [10]. A cherry-red color is considered acceptable, but a brown color is not [11], and the consumer will accept or reject a purchase decision based on the color [12]. In addition, as reported by Ardeshiri and Rose [13], consumers can form beef preferences among lighter and darker cherry-red variations.
Frank et al. [14] described the mechanisms associated with the positive effects of intramuscular fat for the beef consumer experience like making food softer, facilitating oral processing, lubricating of food particles, increasing saliva viscosity, and acting as a binder or glue assisting in the formation of a solid bolus in preparation for swallowing. Furthermore, intramuscular fat is positively correlated with tenderness, juiciness, and flavor. However, visible fat has a negative influence on consumers’ quality expectations [9,15]. Morales et al. [16] reported that consumers rated meat as being lower quality when informed of the fat contents but showed a preference for a considerable amount of fat in beef when beef was blindly rated.
Fat is considered to be unhealthy by Mexican consumers, but some fat is necessary to improve beef flavor [17], and Banović et al. [18] demonstrated that Portuguese consumers pay more attention and are more likely to choose meat with red color and lower levels of marbling, similar to the findings reported by Ardeshiri and Rose for Australian beef consumers [6]. On the contrary, Korean beef consumers’ shown preference for beef with high intramuscular fat, and this seems to be related to more juiciness and flavor [7].
Beef’s intrinsic attributes, such as color and fat contents, can be evaluated using a spectrophotometer [19], the Soxhlet method [20], or asking consumers to respond to a questionnaire using a Likert scale [21]. However, these methods could be time-consuming, pollutant, and costly, particularly those related to beef quality.
Near-infrared spectroscopy (NIRS) could be a valuable tool for the rapid quantification and classification of beef qualities for consumer acceptance. Prieto et al. [22] and Weeranantanaphan et al. [23] reported the ability of NIRS techniques to quantify beef quality characteristics. They stated that this technique involves an exact process that can be used for the quantification, classification, and authentication of tasks. However, these reviews did not report the direct application of NIRS to the prediction of beef consumer acceptance.
Over the next 10 years, the Organisation for Cooperation and Development [24] estimates that meat consumption, including beef consumption, will increase, especially in developing countries. As indicated by Realini et al. [25] and Ardeshiri and Rose [13], this situation offers the opportunity for the beef industry to customize products and benefit from beef consumer segmentation. Therefore, the use of NIRS to predict the consumer acceptance of beef intrinsic attributes could represent a valuable tool for beef production chain.
This study was conducted to determine the feasibility of using NIRS techniques to predict visual consumer acceptance and willingness to purchase beef products.

2. Materials and Methods

2.1. Animals and Sample Collection

The research protocol was approved by the Ethics Committee for Animal Research of Universidad de Antioquia (approval number 116/2018).
Between 15 and 20 steers aged 24 to 36 months were chosen daily, using a convenience sampling method, on nonconsecutive days, until 200 steers had been selected for the final sample. The steers came from fattening production systems characterized by grazing management, a high proportion of naturalized pasture (80% of the area), and low use of concentrate as informed by González-Quintero et al. [26]., The phenotypic characterization was undefined crossbreed between Bos indicus and with Bos Taurus. Steers were slaughtered in an inspected slaughterhouse in Medellin City, and after two days, meat products were shipped under refrigerated conditions to retail facilities. Carcasses were aged for six days, at 4 °C, and then divided into prime cuts.

2.2. Meat Samples

Samples were taken individually (600 g of muscle) from the left longissimus thoracis of each carcass, between the 12th and 13th ribs. Each sample was divided into 5 steaks (25 mm each), trimmed of all subcutaneous fat, photographed before blooming for 20 min, and stored at −80 °C in vacuum-sealed packaging for later analysis.

2.3. Beef Steak Images

Photographs were taken using a Canon PowerShot SX60HS camera, equipped with a 33-mm lens that was mounted on a photographic stand to take images of raw steaks. The camera was configured automatically, based on fixed illumination using a D65 fluorescent light [27].
Each photograph was labeled and stored, in both .tiff and RAW formats, for processing with ImageJ® software [28]. Image handling procedures included background removal, size standardization, and residual subcutaneous fat removal from the image.

2.4. Development of a Beef Consumer Acceptance Index

Each consumer received a survey, containing a picture and a Likert scale for each intrinsic attribute evaluated. The survey format showed two photographs corresponding to two individual samples, characterized as having either a low or high degree of marbling [29]. Each surveyor rated the photograph for beef color, visible fat (marbling), and overall appearance, using a modified 5-point Likert scale, based on the American Meat Science Association [21] guidelines, to evaluate consumer acceptance, where 1 indicated “Dislike very much”, 2 indicated “Dislike”, 3 indicated “Neither Like or Dislike”, 4 indicated “Like” and 5 indicated “Like very much”. Additionally, to evaluate the consumers’ willingness to purchase the raw beef steaks evaluated in the photographs, the survey asked for the consumers’ disposition to buy the beef, using a dichotomic yes or no answer. The survey also requested information regarding socioeconomic and demographic characteristics for each participant.
The survey sample was established based on population reports [30] and was applied to different socioeconomic levels. A total of 400 surveys were applied, following a proportional stratified sampling method [31,32]. Surveys were applied face-to-face on nonconsecutive days, to allow maximal beef consumer representation [33]. Surveys were performed in retail meat areas, inside of market facilities, to assure that each surveyed participant was a beef consumer [32], and only consumers over 18 years old were chosen. Each survey participant was asked about their willingness to participate in a survey, provided signed informed consent, and received a beef by-product as compensation for participation.

2.5. Perception Index: Statistical Analysis and Structural Equation Modeling

The Likert data related to the consumers’ visual perceptions of beef were subjected to polychoric correlation analysis, to ensure that the relationships among Likert variables were >0.3. The Kaiser–Meyer–Olkin measure of sample adequacy (MSA) test was used to verify that variables were pertinent. Likert responses with MSA values < 0.5 were discarded [34]. Factorial analysis was used to develop a consumer acceptance index into a single variable (“visual beef consumer perception index”) [35]. The analysis was developed to apply a maximum likelihood method to factor extraction and used a “varimax” rotation to guarantee factor independence. The construct identified was subjected to the Cronbach’s alpha coefficient, and the factor was validated to ensure it met the criterion of alpha > 0.6 [34]. The visual beef consumer perception index was validated and applied to structural equation modeling (SEM), as indicated by Beaujean [36]. The quality of the model was ensured by the absence of negative variance in the errors, correlations higher than 1, high standard errors, and factorial loadings (FL) > 1. The model was validated to have a root mean square error of approximation (RMSEA) ≤ 0.1, and comparative fit index (CFI) and Tucker–Lewis index (TLI) values ≥ 0.9 [34,36].
All data analysis was performed during index development using a psych [37] and lavaan package [38], in the software R-Project.

2.6. NIRS Collection

Each beef sample was subjected to NIRS, collected using reflectance and transmittance. All samples were thawed for 12 h at 4 °C, then subjected to re-blooming for 10 min before pre-processing for spectra collection.
The reflectance method was performed in the Animal Nutrition Laboratory, associated with the Faculty of Agricultural Sciences at the University of Antioquia, using a NIRS DS 2500F monochromator (Foss NIR system, Denmark), with a detector for silicon (800–1100 nm) and lead sulfide (1100–2500 nm). Samples were ground using a refrigerated water mill (IKA M20, IKA-Werke GmbH&Co, Staufen, Germany), to homogenize the sample to a standard particle size. An aliquot of the ground beef sample was collected using a modified syringe, to avoid bubbles [39], and placed in a circular quartz cuvette (70-mm diameter). Subsequently, samples were scanned 40 times, from 800 nm to 2500 nm, at 2 nm increments, and scans were averaged by equipment software (ISI Scan Nova® and MOSAIC®) and stored as the reciprocal of reflectance [log (1/R)].
Spectra in transmittance mode were collected at a physicochemical laboratory affiliated with the Zenú food industry research center CI + D, in Medellín (Colombia), using a FoodScan™ (FOSS North America, Eden Prairie, MN, USA). Samples were handled as described by Anderson et al. [40]. Briefly, 180 g of ground, homogenized beef (food processor HC150B 1720 +/−15% rpm), at 10 °C–20 °C, was placed into FoodScan™ sample cups (140-mm diameter × 17.5-mm height). Subsequently, samples were placed in the FoodScan™ chamber, and spectra were obtained in the near-infrared region, from 850 to 1050 nm. Each sample was scanned 20 times, and the scans were averaged by the equipment software and stored as the reciprocal of transmittance [log (1/T)].

2.7. Spectra data Analysis

Spectra data manipulation and analyses were performed in the statistical software R-Project [41].
Transmittance and reflectance NIR data were subjected to standard normal variate (SNV) transformation, followed by detrending correction (SNV&D, Barnes, Dhanoa, & Lister [42]). After SNV spectra correction, the first- or second-order derivatives, based on the Savitzky–Golay procedure [43] and using a window size of 5 data points (11 smoothing points), and a second-order polynomial equation, were applied. The spectra transformation and scatter correction were performed using a prospectr package [44].
The mdatools package [45] was used to perform a principal component analysis, using a singular value decomposition algorithm, with a cross-validated process [leave one out (LOO)], to allow two passes for the elimination of outliers, based on H and T statistics. The H statistic assesses how different a sample spectrum is from a population spectra, and the T statistic evaluates how close the predicted value is to the actual value [46].
Partial least square discriminant analysis (PLS-DA) was applied to develop a NIRS prediction model to discriminate the visual beef consumers’ perception index and consumer’s perceptions of beef color, visible fat, and overall appearance, applying a dummy variable. The PLS-DA model allows the covariance between the responses in the Likert categories (Y) and the spectra (X) data matrix to be maximized [43]. In this analysis, the spectra dataset was split into two sets: calibration (70%) and validation (30%). The test set samples were randomly selected, guaranteeing that the samples were uniformly distributed over the spectra variation, to cover the entire variation range in both datasets. The PLS-DA model was generated by the pls package [47], with a nonlinear iterative partial least squares algorithm and LOO cross-validation. The optimum numbers of components in the PLS-DA analysis were retained, based on the relationship between the explained variance among the dependent and independent variables and the calibration and cross-validation processes.
The prediction models developed during the calibration process were validated with the validation dataset. The accuracy prediction during the validation process was assessed by the percentage of samples correctly classified, based on a confusion matrix for the cross-validation process. A ± 0.5 threshold for the correct classification percentage of the predicted Likert category variables was considered [48].
All statistical processes based on the PLS-DA were performed for the original 5-point Likert scale and a simplified Likert scale using three categories. The simplification process resulted from combining Likert categories 1, “Dislike very much” and 2, “Dislike” into a “Dislike” category and combining categories 4, “Like” and 5, “Like very much” into a “Like” category.

3. Results and Discussion

3.1. Sociodemographic Characteristics

The gender proportions shown in Table 1 were comparable with the proportions reported by the National Administrative Department for Statistics in Colombia (DANE) for the census performed in 2018 [49], where the male proportion was 48.8% and the female proportion was 51.2%. Other characteristics, such as age, income, marital status, and educational status, are presented as general information, due to a lack of Colombian data sources. The proportion of housewives (19.55%) and the proportion of employed (40.58%) participants in this study were comparable with national statistics, but not the proportions of retired (11.49% vs. 3.6%), students (4.11% vs. 18%), and unemployed (0.7% vs. 5.0%) participants. The socioeconomic and demographic characteristics of the sample population were similar to those reported for the national population.

3.2. Perception Index

The polychoric correlation among the Likert scale variables revealed a strong agreement between the consumers’ responses to visible fat and color (rp = 0.819, p < 0.001), visible fat and overall appearance (rp = 0.883, p < 0.001), and color and overall appearance (rp = 0.924, p < 0.001). These correlations allowed a cohesive construct to be generated based on visual consumer perceptions. Factor analysis explained 79% of the variance for a single factor. The SEM analysis confirmed the goodness-of-fit for the indexes, CFI = 0.995, TL = 0.984, and RMSEA = 0.1 (Table 2), which met the criteria stated by Kline [50]. Correlations associated with ratings for visible fat and overall appearance could be associated with the consumers’ preferences for low-marbled beef, as indicated by Ardeshiri and Rose [13], as the beef samples used in this study had a median of 2.5% intramuscular fat [29].
To predict the consumers’ visual perceptions, SEM was performed. The results showed the significant contributions (p < 0.05) of color and overall appearance, with FL greater than 0.9, and the secondary importance of visible fat and willingness to buy (FL > 0.8). Grunert et al. [8], who used photographs to assess visual beef perception and SEM of 800 consumers in Spain, Germany, France, and the United Kingdom, found that beef color and visible fat showed high loading factors when building their index, but also included price, a parameter that we did not include. Font-i-Furnols and Guerrero [15] also stated that intrinsic cues could be associated with consumers’ expectations relative to past experiences, associating visual beef characteristics with post-purchasing experiences, such as meal preparation and eating conditions, which may motivate the future purchases of the consumer [8].
According to Ngapo et al. [17], Mexican consumers (488), who were characterized as being >37 years old (77%), female (75%), and married (60%), a proportion similar to consumers’ features in the present study, build their beef perceptions based on more than two intrinsic attributes, where color, visible fat, and freshness were the most important. This study found that visible fat and overall appearance affected willingness to purchase.
In contrast to the results of this study, Banović et al. [18] and Morales et al. [16] found that visible fat had a negative relationship with consumer perception.

3.3. Estimating Consumers’ Perceptions with NIRS

Raw and preprocessed spectra for the first derivative of reflectance and the SNV&D for transmittance are shown in Figure 1. Both spectroscopy methods presented characteristic absorbance spectra for common organic bonds and near-infrared regions.
The reflectance spectra showed absorbance peaks near 900 nm, related to -OH and -CH bonds, at 1250 nm, associated with the second overtone of -CH, at 1451 nm, related to the first overtone of -OH bonds, and, between 1695 and 1725 nm, linked to the second overtone of -CH bonds. These spectra features are similar to those reported by Cozzolino and Murray [51], Mamani-Linares et al. [52], Prieto et al. [53], and Morsy and Dan-Wen [54]. Similarly, the transmittance spectrum in the present research showed a wide peak between 950 and 1000 nm, related to moisture (934–960 nm and 984–996 nm) and fat (962–968 nm), as indicated by Sierra et al. [55] and De Marchi et al. [56].
The PLS-DA analysis in the visual beef consumer perception assessed by a 5-point Likert scale is shown in Table 3. The calibration process in reflectance spectra to visible fat, overall appearance, and the visual perception index achieved a low percentage of correct classification that ranged from 0% to 68.2%. The Likert scale in consumers’ beef color reached 80% of correct classification in the category “Neither like or dislike”. However, when the consumers’ perception models were externally validated, the maximum rate of correct classification was 77.8%, with accuracy ranging from 29% to 42%. The discrimination of Likert scale in transmittance mode improves the correct classification estimate. The calibration process achieved 70.4%, 82.4%, and 94.1% in the “Neither like or dislike” category in visible fat, overall appearance, and the visual beef perception index, respectively. Nonetheless, the accuracy models ranged between 39.4% and 42.4% for the correct classification during the external validation process.
The simplification of the Likert scale (3-point—Table 4) in visual beef consumers’ perception on the PLS-DA analysis, resulted in a high percentage of correct classification, compared to the 5-point Likert scale. Calibration models for Likert scale prediction in reflectance spectra showed an accuracy of 61.7% in overall appearance and 60.2% in the visible fat. In these models, the Likert categories “Neither like or dislike” and “Like” achieved correct classification percentages of 92.0% and 56.8% in the evaluation of beef visible fat, and 73.3%, and 68% for overall appearance, respectively. Similarly, consumers’ beef color evaluation in the “Neither like or dislike” category reached 80% correct classification. However, the calibration color model obtained a low model accuracy (30.4%).
Furthermore, the visual beef consumers’ perception index followed a similar behavior to that observed in the visual color calibration model. The performance of the transmittance model on a 3-point Likert scale reached the highest accuracy in the calibration and external validation, with 73.2% and 54.6% in the visible fat. Besides, external validation models achieved an accuracy lower than 53%.
The prediction of consumers’ perceptions of visible fat by NIRS in transmittance mode achieved the highest correct classification rate among the intrinsic attributes evaluated, with accuracy values of 73.2% and 54.6% during the calibration and external validation models, respectively, when using the simplified 3-point Likert scale. This model showed a high rate of error (ranging from 33.4% to 83.3%). However, for the prediction of willingness to purchase, the accuracy during the external validation process increased to 76% (Table 4). No other studies have been performed examining this outcome for comparisons.
The abilities of reflectance and transmittance NIRS to predict the intrinsic attributes of meat [22], and meat classifications during the adulteration/authentication process [57,58], have been widely documented. In comparison, the ability of NIRS to predict consumer sensory parameters related to beef has been less well documented, and contradictory results have been reported [59]. Until now, no other studies have examined the application of NIRS for the prediction of consumers’ visual perceptions of beef attributes.
The research reported by Liu et al. [60] regarding NIRS for the prediction of beef sensory attributes used a trained sensory panel (n = 7) to evaluate how chewiness and juiciness correlated with reflectance data and reported R-squared (R2) calibration values of 0.58 and 0.5, respectively, concluding that NIRS data had minimal predictive abilities. Furthermore, Yancy et al. [61] applied reflectance NIRS to predict beef consumers’ (n = 240) perceptions in a visual- and taste-based experiment, examining the overall impression and tenderness of beef, and reported a predictive model with R2 values during the full cross-validation (R2cv) process of 0.79 and 0.74, respectively.
Recently, Magalhães et al. [62] applied NIRS reflectance data, obtained from intact longissimus thoracic muscle (n = 644) from Nelore steers, to predict meat quality attributes closely associated with beef consumers’ acceptance, such as share force, marbling, and CIELab color. They reported a low predictive ability (R2 prediction values from 0 to 0.45), associated with high beef heterogeneity, especially for the marbling score. Likewise, Cafferky et al. [63] reported the low accuracy of NIRS reflectance predictive models for beef Warner–Bratzler shear force (R2cv = 0.14–0.22), sensory texture (R2cv = 0.06–0.13), sensory juiciness (R2cv = 0.04–0.09), and sensory flavor (R2cv = 0.04–0.28), as assessed by a trained sensory panel (n = 12). On the contrary to the above-cited negative reports, Prieto et al. [64], showed successful application of NIRS (1100–2500 nm) to the prediction of beef’s physical parameters using ground meat, arguing that correlation between color measurement (L*, a* and b*) and intramuscular fat, allowed to the developed of the NIRS prediction model.
The limited ability of NIRS to predict meat quality attributes preferred by beef consumers has been associated with the narrow scale used to evaluate sensory characteristics, and high heterogeneity among the meat traits within muscle contributes to the generation of significant bias within NIRS models [59]. In addition, Liu et al. [60] stated that handling processes, such as freeze-thaw, could contribute to samples scanned by NIRS having different attributes from those tested by consumers and trained panels or analyzed in laboratories. Moreover, Cozzolino and Murray [65] and Prieto et al. [48] affirmed that the absence of concordance between NIRS spectra and physical beef attributes could be related to the grinding process. Grinding affects muscle integrity by altering the form, which alters how light is trapped and scattered in the muscle and disrupts the randomized muscle fiber arrangement, which alters light scattering. On the other hand, Ripoll et al. [66] reported accurate NIRS prediction models for instrumental beef characteristics like Warner–Bratzler (R2p = 0.743) and water holding capacity (R2p = 0.829), and the sensory attribute tenderness (R2p = 0.981), using ground meat. These authors state that indirect spectral information, retrieved from beef moisture and fat and their relationship with physical beef parameters, allowed the development of the prediction models for instrumental and sensory beef parameters.
The PLS-DA used to predict consumers’ willingness to purchase beef, as evaluated by assessments of beef photographs (Table 4), showed accuracy values of 69.6% and 75.6% during calibration and 70% and 75.8% during external validation for the reflectance and transmittance models, respectively. Both models showed a high percentage of correct classification for positive disposition to buy, ranging from 93% to 100%. However, the correct classification of negative decisions only reached 33% for the transmittance model.
The average NIRS spectra of both reflectance and transmittance methods for willingness to buy are shown in Figure 2. Both spectra modes showed that the spectra of samples that were negatively rated to buy reached the highest absorbance at 950 nm, 1250 nm, and between 1695 and 1725 nm for reflectance, and from 962 to 968 nm in transmittance. These wavelengths were related to -CH molecular bonds [43,44,45,46,47,48]. Similar to the statement by Ripoll et al. [66], the spectra fingerprint related to visible fat in the beef sample used for this study could have contributed to separating the unfavorable decision to buy associated with more visible fat, from lean beef samples.
Ardeshiri and Rose [13] affirmed that fat and marbling played important roles in purchasing decisions. Consumers use these beef attributes to create a perception of beef quality, integrating past experiences and future expectations [9,15]. Therefore, the relationship between beef visual attributes and consumer’s purchasing decisions in the present study could explain the better performance of the beef purchasing discrimination model, especially for transmittance data, where spectra separation in the near-infrared transmittance region associated with fat could be observed [55].

4. Conclusions

We proposed a beef consumers’ perception index, based on the evaluation of beef’s intrinsic attributes, supported by a high correlation among beef color, visible fat, and overall appearance, and generated using an SEM model.
The transmittance spectral data could be used to develop an acceptable discrimination model for consumers’ purchasing decisions using ground meat. Our results reported an accuracy of 76%, but these models could be improved by the use of NIRS equipment robust enough to read intact samples.
The prediction of consumer’s willingness to purchase could improve the marketing dynamic in the beef production chain, allowing beef producers and animal science technicians to develop a production strategy to produce beef with a high probability of purchase.
In the meat industry, the NIRS prediction model of consumers’ willingness to buy beef would allow the segmentation of meat with potential to be purchased, from meat that would not be purchased and would be assigned for the production of meat by-products.

Author Contributions

Conceptualization, L.M.-L. and W.B.-H.; design of consumer’s survey, M.O.-A. and J.A.-A.; training personnel and collecting field data, W.B.-H. and J.A.-A.; formal analysis, L.M.-L. and W.B.-H.; writing—original draft preparation, W.B.-H.; writing—review and editing, L.M.-L., M.O.-A. and J.A.-A.; visualization, W.B.-H.; funding acquisition, L.M.-L., J.A.A. and M.O.-A. All authors have read and agreed to the published version of the manuscript.

Funding

This research was funded by Colombian Ministry of Science and Technology through doctoral scholarship 727 of 2015.

Acknowledgments

We thank the research groups GRICA and Biogénesis for the student’s support in this study. We also thank the Corporación Colombiana de investigación Agropecuaria (Agrosavia) for granting time permission for education to the first author and for financial aid to publish, and the beef brand “Quality Beef,” belonging to EURO Supermercados, for supporting the sample collection and allowing the survey.

Conflicts of Interest

The authors declare no conflict of interest.

References

  1. Scollan, N.; Hocquette, J.F.; Nuernberg, K.; Dannenberger, D.; Richardson, I.; Moloney, A. Innovations in beef production systems that enhance the nutritional and health value of beef lipids and their relationship with meat quality. Meat Sci. 2006, 74, 17–33. [Google Scholar] [CrossRef] [PubMed]
  2. Pencharz, P.B.; Elango, R.; Wolfe, R.R. Recent developments in understanding protein needs—How much and what kind should we eat? Appl. Physiol. Nutr. Metab. 2016, 41, 577–580. [Google Scholar] [CrossRef] [Green Version]
  3. Bender, A.E. Meat and Meat Products in Human Nutrition in Developing Countries; FAO: Rome, Italy, 1992; ISBN 9251031460. [Google Scholar]
  4. Polidori, P.; Vincenzetti, S.; Pucciarelli, S.; Polzonetti, V. CLAs in animal source foods: Healthy benefits for consumers. In Bioactive Molecules in Food. Reference Series in Phytochemistry; Mérillon, J.M., Ramawat, K., Eds.; Springer: Cham, Switzerland, 2018; ISBN 978-3-319-54528-8. [Google Scholar]
  5. Visvalingam, J.; Vahmani, P.; Rolland, D.C.; Dugan, M.E.R.; Yang, X. Anti-mutagenic Properties of Mono- and Dienoic Acid Biohydrogenation Products from Beef Fat. Lipids 2017, 52, 651–655. [Google Scholar] [CrossRef]
  6. Vahmani, P.; Meadus, W.J.; Rolland, D.C.; Duff, P.; Dugan, M.E.R. Trans10,cis15 18:2 Isolated from Beef Fat Does Not Have the Same Anti-Adipogenic Properties as Trans10,cis12–18:2 in 3T3-L1 Adipocytes. Lipids 2016, 51, 1231–1239. [Google Scholar] [CrossRef]
  7. Vahmani, P.; Meadus, W.J.; Da Silva, M.L.P.; Mitchell, A.D.; Mapiye, C.; Duff, P.; Rolland, D.C.; Dugan, M.E.R. A trans10-18:1 enriched fraction from beef fed a barley grain-based diet induces lipogenic gene expression and reduces viability of HepG2 cells. Biochem. Biophys. Rep. 2016, 7, 84–90. [Google Scholar] [CrossRef] [Green Version]
  8. Grunert, K.G.; Bredahl, L.; Brunsø, K. Consumer perception of meat quality and implications for product development in the meat sector—A review. Meat Sci. 2004, 66, 259–272. [Google Scholar] [CrossRef]
  9. Henchion, M.M.; McCarthy, M.; Resconi, V.C. Beef quality attributes: A systematic review of consumer perspectives. Meat Sci. 2017, 128, 1–7. [Google Scholar] [CrossRef] [Green Version]
  10. Troy, D.J.; Kerry, J.P. Consumer perception and the role of science in the meat industry. Meat Sci. 2010, 86, 214–226. [Google Scholar] [CrossRef]
  11. Mancini, R.A. Meat color. In Improving the Sensory and Nutritional Quality of Fresh Meat; Woodhead Publishing Limited: Cambridge, UK, 2009; pp. 89–110. ISBN 9781845693435. [Google Scholar]
  12. Carpenter, C.E.; Cornforth, D.P.; Whittier, D. Consumer preferences for beef color and packaging did not affect eating satisfaction. Meat Sci. 2001, 57, 359–363. [Google Scholar] [CrossRef]
  13. Ardeshiri, A.; Rose, J.M. How Australian consumers value intrinsic and extrinsic attributes of beef products. Food Qual. Prefer. 2018, 65, 146–163. [Google Scholar] [CrossRef]
  14. Frank, D.; Joo, S.T.; Warner, R. Consumer acceptability of intramuscular fat. Korean J. Food Sci. Anim. Resour. 2016, 36, 699–708. [Google Scholar] [CrossRef] [PubMed]
  15. Font-i-Furnols, M.; Guerrero, L. Consumer preference, behavior and perception about meat and meat products: An overview. Meat Sci. 2014, 98, 361–371. [Google Scholar] [CrossRef] [PubMed]
  16. Morales, R.; Aguiar, A.P.S.; Subiabre, I.; Realini, C.E. Beef acceptability and consumer expectations associated with production systems and marbling. Food Qual. Prefer. 2013, 29, 166–173. [Google Scholar] [CrossRef]
  17. Ngapo, T.M.; Braña Varela, D.; Rubio Lozano, M.S. Mexican consumers at the point of meat purchase. Beef choice. Meat Sci. 2017, 134, 34–43. [Google Scholar] [CrossRef]
  18. Banović, M.; Chrysochou, P.; Grunert, K.G.; Rosa, P.J.; Gamito, P. The effect of fat content on visual attention and choice of red meat and differences across gender. Food Qual. Prefer. 2016, 52, 42–51. [Google Scholar] [CrossRef]
  19. Mancini, R.A.; Hunt, M.C. Current research in meat color. Meat Sci. 2005, 71, 100–121. [Google Scholar] [CrossRef]
  20. Font-i-Furnols, M.; Fulladosa, E.; Prevolnik Povše, M.; Čandek-Potokar, M. Future trends in non-invasive technologies suitable for quality determinations. In A Handbook of Reference Methods for Meat Quality Assessment (M Font-i-Furnols); Čandek-Potokar, M., Maltin, C., Povše, M.P., Eds.; SRUC: Edinburgh, Scotland, 2015; pp. 90–103. [Google Scholar]
  21. AMSA. Research Guidelines for Cookery, Sensory Evaluation, and Instrumental Tenderness Measurements of Meat; American Meat Science Association: Champaign, IL, USA, 2016; ISBN 8005172672. [Google Scholar]
  22. Prieto, N.; Pawluczyk, O.; Dugan, M.E.R.; Aalhus, J.L. A review of the principles and applications of near-infrared spectroscopy to characterize meat, fat, and meat products. Appl. Spectrosc. 2017, 71, 1403–1426. [Google Scholar] [CrossRef] [Green Version]
  23. Weeranantanaphan, J.; Downey, G.; Allen, P.; Sun, D.W. A review of near infrared spectroscopy in muscle food analysis: 2005–2010. J. Near Infrared Spectrosc. 2011, 19, 61–104. [Google Scholar] [CrossRef]
  24. OCDE. OCDE-FAO Perspectivas Agrícolas 2019-2028; OCDE-FAO Perspectivas Agrícolas; OECD: Roma, Italy, 2019; ISBN 978-92-64-18272-1. [Google Scholar]
  25. Realini, C.E.; Kallas, Z.; Pérez-Juan, M.; Gómez, I.; Olleta, J.L.; Beriain, M.J.; Albertí, P.; Sañudo, C.; Kallas, Z. Relative importance of cues underlying Spanish consumers’ beef choice and segmentation, and consumer liking of beef enriched with n-3 and CLA fatty acids. Food Qual. Prefer. 2014, 33, 74–85. [Google Scholar] [CrossRef]
  26. González-Quintero, R.; Sánchez-Pinzón, M.S.; Bolívar-Vergara, D.M.; Chirinda, N.; Arango, J.; Pantévez, H.A.; Correa-Londoño, G.; Barahona-Rosales, R. Technical and environmental characterization of Colombian beef cattle-fattening farms, with a focus on farm size and ways of improving production. Outlook Agric. 2019. [Google Scholar] [CrossRef]
  27. Cortez Passetti, R.A.; Akamine, J.; Garcia, M.; Mottin, C.; Lopes De Olivera, C.A.; Guerrero, A.; Campo, M.; Sañudo, C.; Nunes, I. Validation of photographs usage to evaluate meat visual acceptability of young bulls fi nished in feedlot fed with or without essential oils. Meat Sci. 2017, 123, 105–111. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  28. Schneider, C.A.; Rasband, W.S.; Eliceiri, K.W. NIH Image to ImageJ: 25 years of image analysis. Nat. Methods 2012, 9, 671–675. [Google Scholar] [CrossRef] [PubMed]
  29. Barragán-Hernández, W.Á.; Mahecha-Ledesma, L.; Angulo-Arizala, J.; Olivera-Angel, M. Clasificación de la calidad de la carne bovina mediante el uso de infrarrojo por transmitancia y técnicas multivariadas. XV Encuentro Nacional y VIII Internacional de los Investigadores de las Ciencias Pecuarias (ENICIP). Rev. Colomb. Cienc. Pecu. 2019, 32, 50. [Google Scholar]
  30. Mejía-Arango, L.J.; Estrada, C.M.; Salcedo-Amaya, G.J.; Higuita, J.M.; Granados, J.D. Informe de Calidad de Vida Medellìn 2016; Pregón SAS: Medellín, Colombia, 2016. [Google Scholar]
  31. Alinda, F.; Kavoi, M.; Migisha, J. Consumer willingness to pay for quality beef in Kampala, Uganda. J. Agric. Sci. Technol. 2016, 17, 59–77. [Google Scholar]
  32. Enciso, K.; Burkart, S.; Charrya, A.; Rodriguez, C.D.P.; Quiceno, J.J.M.; Ruiz, L.R.; Solis, J.F.G.; Quilac, N.J.V.; Lopez, N.A.; Velasco, S.M.; et al. Consumer Preferences and Market Segmentation for Differentiated Beef with Less Environmental Impact; CGSpace: Montpellier, France, 2016. [Google Scholar]
  33. Xue, H.; Mainville, D.; You, W.; Nayga, R.M. Consumer preferences and willingness to pay for grass-fed beef: Empirical evidence from in-store experiments. Food Qual. Prefer. 2010, 21, 857–866. [Google Scholar] [CrossRef]
  34. Hair, J.; Black, W.; Babin, B.; Anderson, R. Multivariate Data Analysis; Pearson, Education: Edinburgh, UK, 2017; ISBN 9781118895238. [Google Scholar]
  35. Bishu, K.G.; O’Reilly, S.; Lahiff, E.; Steiner, B. Cattle farmers’ perceptions of risk and risk management strategies: Evidence from Northern Ethiopia. J. Risk Res. 2016, 21, 579–598. [Google Scholar] [CrossRef] [Green Version]
  36. Beaujean, A.A. Latent Variable Modeling Using R: A Step-by-Step Guide; Routledge, Taylor & Francis Group: East Sussex, NY, USA, 2014; ISBN 9781315869780. [Google Scholar]
  37. Revelle, W. Package ‘psych’. Available online: https://personality-project.org/r/psych-manual.pdf (accessed on 8 January 2020).
  38. Rosseel, Y. Lavaan: An R package for structural equation modeling and more. Version 0.5–12 (BETA). J. Stat. Softw. 2012, 48, 1–36. [Google Scholar] [CrossRef] [Green Version]
  39. Prieto, N.; López-Campos, Ó.; Aalhus, J.L.; Dugan, M.E.R.R.; Juárez, M.; Uttaro, B. Use of near infrared spectroscopy for estimating meat chemical composition, quality traits and fatty acid content from cattle fed sunflower or flaxseed. Meat Sci. 2014, 98, 279–288. [Google Scholar] [CrossRef] [PubMed]
  40. Anderson, S. Determination of fat, moisture, and protein in meat and meat products by using the FOSS FoodScanTM near-infrared spectrophotometer with FOSS artificial neural network calibration model and associated database: Collaborative study. J. AOAC Int. 2007, 90, 1073–1083. [Google Scholar]
  41. R Development Core Team. R: A Language and Environment for Statistical Computing; R Foundation for Statistical Computing: Vienna, Austria, 2009. [Google Scholar]
  42. Barnes, R.J.; Dhanoa, M.S.; Lister, S.J. Standard normal variate transformation and de-trending of near-infrared diffuse reflectance spectra. Appl. Spectrosc. 1989, 43, 772–777. [Google Scholar] [CrossRef]
  43. Naes, T.; Isaksson, T.; Fearn, T.; Davies, T. Introduction A User-Friendly Guide to Multivariate Calibration and Classification; NIR publications: Chichester, UK, 2002; ISBN 0952866625. [Google Scholar]
  44. Stevens, A.; Ramirez Lopez, L. An introduction to the prospectr package. R Packag. Vignette Rep. No. R Packag. Version 0.1 2014, 3, 1–22. [Google Scholar]
  45. Kucheryavskiy, S. Mdatools: Multivariate Data Analysis for Chemometrics, R Package Version 0.7.0. 2015. Available online: https://CRAN.R--project.org/package=mdatools (accessed on 8 January 2020).
  46. Williams, P.; Norris, K. Near-Infrared Technology in the Agricultural and Food Industries; American Association of Cereal Chemists, Inc.: Saint Paul, Minnesota, 1997; Volume 36, ISBN 091325049X. [Google Scholar]
  47. Mevik, B.-H.; Wehrens, R.; Liland, K.H.; Hiemstra, P. Package ‘pls’. Available online: https://cran.rediris.es/web/packages/pls/pls.pdf (accessed on 8 January 2020).
  48. Prieto, N.; López-Campos, Ó.; Zijlstra, R.T.; Uttaro, B.; Aalhus, J.L. Discrimination of beef dark cutters using visible and near infrared reflectance spectroscopy. Can. J. Anim. Sci. 2014, 94, 445–454. [Google Scholar] [CrossRef] [Green Version]
  49. Censo Nacional de Población y Vivienda (CNPV)-2018; DANE: Bogotá, Colombia, 2019.
  50. Kline, R.B. Principles and Practice of Structural Equation Modeling, 3rd ed.; Guiford: New York, NY, USA, 2011; ISBN 978-1-60623-876-9. [Google Scholar]
  51. Cozzolino, D.; Murray, I. Identification of animal meat muscles by visible and near infrared reflectance spectroscopy. LWT Food Sci. Technol. 2004, 37, 447–452. [Google Scholar] [CrossRef]
  52. Mamani-Linares, L.W.; Gallo, C.; Alomar, D. Identification of cattle, llama and horse meat by near infrared reflectance or transflectance spectroscopy. Meat Sci. 2012, 90, 378–385. [Google Scholar] [CrossRef] [PubMed]
  53. Prieto, N.; Roehe, R.; Lavín, P.; Batten, G.; Andrés, S. Application of near infrared reflectance spectroscopy to predict meat and meat products quality: A review. Meat Sci. 2009, 83, 175–186. [Google Scholar] [CrossRef]
  54. Morsy, N.; Sun, D.W. Robust linear and non-linear models of NIR spectroscopy for detection and quantification of adulterants in fresh and frozen-thawed minced beef. Meat Sci. 2013, 93, 292–302. [Google Scholar] [CrossRef] [PubMed]
  55. Sierra, V.; Aldai, N.; Castro, P.; Osoro, K.; Coto-Montes, A.; Oliván, M. Prediction of the fatty acid composition of beef by near infrared transmittance spectroscopy. Meat Sci. 2008, 78, 248–255. [Google Scholar] [CrossRef]
  56. De Marchi, M.; Penasa, M.; Cecchinato, A.; Bittante, G. The relevance of different near infrared technologies and sample treatments for predicting meat quality traits in commercial beef cuts. Meat Sci. 2013, 93, 329–335. [Google Scholar] [CrossRef]
  57. Moran, L.; Andres, S.; Allen, P.; Moloney, A.P. Visible and near infrared spectroscopy as an authentication tool: Preliminary investigation of the prediction of the ageing time of beef steaks. Meat Sci. 2018, 142, 52–58. [Google Scholar] [CrossRef]
  58. López-Maestresalas, A.; Insausti, K.; Jarén, C.; Pérez-Roncal, C.; Urrutia, O.; Beriain, M.J.; Arazuri, S. Detection of minced lamb and beef fraud using NIR spectroscopy. Food Control 2019, 98, 465–473. [Google Scholar] [CrossRef]
  59. Berri, C.; Picard, B.; Lebret, B.; Andueza, D.; Lefèvre, F.; Le Bihan-Duval, E.; Beauclercq, S.; Chartrin, P.; Vautier, A.; Legrand, I.; et al. Predicting the quality of meat: Myth or reality? Foods 2019, 8, 436. [Google Scholar] [CrossRef] [Green Version]
  60. Liu, Y.; Lyon, B.G.; Windham, W.R.; Realini, C.E.; Pringle, T.D.D.; Duckett, S. Prediction of color, texture, and sensory characteristics of beef steaks by visible and near infrared reflectance spectroscopy. A feasibility study. Meat Sci. 2003, 65, 1107–1115. [Google Scholar] [CrossRef]
  61. Yancey, J.W.S.; Apple, J.K.; Meullenet, J.F.; Sawyer, J.T. Consumer responses for tenderness and overall impression can be predicted by visible and near-infrared spectroscopy, Meullenet-Owens razor shear, and Warner-Bratzler shear force. Meat Sci. 2010, 85, 487–492. [Google Scholar] [CrossRef]
  62. Magalhães, A.F.B.; de Almeida Teixeira, G.H.; Ríos, A.C.H.; dos Santos Silva, D.B.; Mota, L.F.M.; Muniz, M.M.M.; de Morais, C.D.L.M.; de Lima, K.M.G.; Júnior, L.C.C.; Baldi, F.; et al. Prediction of meat quality traits in Nelore cattle by near-infrared reflectance spectroscopy. J. Anim. Sci. 2018, 96, 4229–4237. [Google Scholar] [CrossRef]
  63. Cafferky, J.; Sweeney, T.; Allen, P.; Sahar, A.; Downey, G.; Cromie, A.R.; Hamill, R.M. Investigating the use of visible and near infrared spectroscopy to predict sensory and texture attributes of beef M. longissimus thoracis et lumborum. Meat Sci. 2020, 159, 107915. [Google Scholar] [CrossRef] [PubMed]
  64. Prieto, N.; Andrés, S.; Giráldez, F.J.; Mantecón, A.R.; Lavín, P. Ability of near infrared reflectance spectroscopy (NIRS) to estimate physical parameters of adult steers (oxen) and young cattle meat samples. Meat Sci. 2008, 79, 692–699. [Google Scholar] [CrossRef] [PubMed]
  65. Cozzolino, D.; Murray, I. Effect of sample presentation and animal muscle species on the analysis of meat by near infrared reflectance spectroscopy. J. Near Infrared Spectrosc. 2002, 10, 37–44. [Google Scholar] [CrossRef]
  66. Ripoll, G.; Albertí, P.; Panea, B.; Olleta, J.L.; Sañudo, C. Near-infrared reflectance spectroscopy for predicting chemical, instrumental and sensory quality of beef. Meat Sci. 2008, 80, 697–702. [Google Scholar] [CrossRef]
Foods 09 00984 g001 550
Figure 1. Near-infrared reflectance (A) and transmittance (B) in the raw spectrum and after mathematical preprocessing for the first derivative (C) and SNV&D (D) in beef samples.
Figure 1. Near-infrared reflectance (A) and transmittance (B) in the raw spectrum and after mathematical preprocessing for the first derivative (C) and SNV&D (D) in beef samples.
Foods 09 00984 g001
Foods 09 00984 g002 550
Figure 2. Near-Infrared reflectance (A) raw spectra and transmittance (B) standard normal variate (SNV) spectra for consumer’s purchasing decisions of beef samples.
Figure 2. Near-Infrared reflectance (A) raw spectra and transmittance (B) standard normal variate (SNV) spectra for consumer’s purchasing decisions of beef samples.
Foods 09 00984 g002
Table
Table 1. Socioeconomic and Demographic Characteristics of the Survey Participants.
Table 1. Socioeconomic and Demographic Characteristics of the Survey Participants.
Sample (n = 400)
GenderMale41.32%
Female58.67%
Age (years)<3731.78%
37–5538.14%
55–7428.11%
>741.95%
IncomeUndeclared16.38%
(times the statutory minimum wage)<111.49%
1–241.07%
3–421.02%
5–67.33%
>62.68%
Marital statusMarried43.27%
Separated6.35%
Single30.56%
Widow3.66%
Consensual union15.64%
Other0.4%
Educational statusPrimary school8.55%
Secondary school35.94%
Technological level21.51%
Professional22.00%
Postgraduate11.00%
Other0.90%
EmploymentHousewife19.55%
Unemployed0.7%
Employed40.58%
Retired11.49%
Student4.15%
Self-employed23.47%
Table
Table 2. Factor analysis (varimax rotation) and structural equation model for visual beef consumer perception.
Table 2. Factor analysis (varimax rotation) and structural equation model for visual beef consumer perception.
Likert Variable 1CommunalityCronbach’s AlphaKMO 2Beef Consumers Perception 3Variance
Color0.900.8940.850.8879%
Visible fat0.840.850.83
Overall appearance0.980.760.99
Willingness to buy0.840.890.84
Perception IndexIndependent VariableFactor Loading (FL)p-value
Visual beef consumers’ perceptionColor0.902<0.001
Visual fat0.841<0.001
Overall appearance0.975<0.001
Willingness to buy0.843<0.001
1 Likert scale response (1. “Dislike very much” and 5. “Like very much”). 2 The Kaiser–Meyer–Olkin measure of sampling adequacy. 3 Factor loadings.
Table
Table 3. Partial least square discriminant analysis for visual beef consumer perception (5-point Likert scale) based on transmittance and reflectance near-infrared spectra.
Table 3. Partial least square discriminant analysis for visual beef consumer perception (5-point Likert scale) based on transmittance and reflectance near-infrared spectra.
ReflectanceTransmittance
Mat. Process. 1C 2Calibration 3 (%)External Validation 3 (%)Mat. Process.CCalibration 3 (%)External Validation 3 (%)
Color1. Dislike very muchRaw30.0000.000SNV&D78.300.000
2. Dislike14.3016.7045.500.000
3. Neither like or dislike80.0075.0064.7033.30
4. Like29.7018.8046.9076.90
5. Like very much0.0000.0000.0000.000
Model accuracy28.4032.5038.1040.00
Visible fat1. Dislike very muchFirst derivative30.0000.000Second derivative20.0000.000
2. Dislike25.000.00014.300.000
3. Neither like or dislike60.0028.6060.9077.80
4. Like67.9070.0070.4054.50
5. Like very much25.000.0000.0000.000
Model accuracy47.6029.0041.5039.40
Overall appearance1. Dislike very muchSNV&D510.000.000SNV&D70.0000.000
2. Dislike14.300.00025.0033.30
3. Neither like or dislike60.0077.8082.4057.10
4. Like62.5060.0054.3053.30
5. Like very much0.0000.0000.00020.00
Model accuracy43.9039.4042.7042.40
Visual Perception index1. Dislike very muchRaw70.0000.000SNV&D40.0000.000
2. Dislike22.200.0000.0000.000
3. Neither like or dislike68.2066.7094.1057.10
4. Like56.3055.0056.4056.30
5. Like very much0.0000.0000.0000.000
Model accuracy43.1042.5045.8040.60
1 Mathematically preprocessed. 2 Number of components in the PLS-DA model. 3 Percentage of correct classification.
Table
Table 4. Partial least square discriminant analysis for visual beef consumer perception index (3-point Likert scale) and consumers’ willingness to purchase based on transmittance and reflectance near-infrared spectra.
Table 4. Partial least square discriminant analysis for visual beef consumer perception index (3-point Likert scale) and consumers’ willingness to purchase based on transmittance and reflectance near-infrared spectra.
ReflectanceTransmittance
Mat. Process. 1C 2Calibration 3 (%)External Validation3 (%)Mat. Process.CCalibration 3 (%)External Validation 3 (%)
Color1. DislikeSNV30.000.00SNV717.4012.50
2. Neither like or dislike80.0083.3076.5066.70
3. Like28.8027.8050.0064.70
Model accuracy30.4037.5046.4051.60
Visible fat1. DislikeFirst derivative713.3020.00SNV&D1035.7016.70
2. Neither like or dislike92.0057.1082.6055.60
3. Like56.8057.9080.0066.70
Model accuracy60.2051.6073.2054.60
Overall appearance1. DislikeSecond derivative529.4014.30First derivative716.7016.70
2. Neither like or dislike73.3066.7076.5071.40
3. Like68.0041.2063.8055.00
Model accuracy61.7043.8056.1051.50
Visual perception index1. DislikeFirst derivative60.0014.30SNV711.1016.70
2. Neither like or dislike72.7077.8076.5071.40
3. Like57.6050.0052.1057.90
Model accuracy49.050.0048.2053.10
Willingness to buyNoRaw10.000.00SNV433.3033.30
Yes10010093.1091.70
Model accuracy69.6070.0075.6075.80
1 Mathematically preprocessed. 2 Number of components in the PLS-DA model. 3 Percentage of correct classification.

Share and Cite

MDPI and ACS Style

Barragán-Hernández, W.; Mahecha-Ledesma, L.; Angulo-Arizala, J.; Olivera-Angel, M. Near-Infrared Spectroscopy as a Beef Quality Tool to Predict Consumer Acceptance. Foods 2020, 9, 984. https://doi.org/10.3390/foods9080984

AMA Style

Barragán-Hernández W, Mahecha-Ledesma L, Angulo-Arizala J, Olivera-Angel M. Near-Infrared Spectroscopy as a Beef Quality Tool to Predict Consumer Acceptance. Foods. 2020; 9(8):984. https://doi.org/10.3390/foods9080984

Chicago/Turabian Style

Barragán-Hernández, Wilson, Liliana Mahecha-Ledesma, Joaquín Angulo-Arizala, and Martha Olivera-Angel. 2020. "Near-Infrared Spectroscopy as a Beef Quality Tool to Predict Consumer Acceptance" Foods 9, no. 8: 984. https://doi.org/10.3390/foods9080984

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop