Traditional Rice Varieties, Consumer Segmentation, and Preferences: A Case Study from Kerala, India

Abstract

Traditional rice varieties (TRVs), shaped by generations of adaptation to local soils and climates, are often seen as less competitive than modern rice varieties (MRVs) due to lower yields. As a result, the spread of MRVs has contributed to a global decline in TRVs. However, TRVs offer notable advantages, particularly in terms of sustainability and health benefits. In light of their gradual disappearance, this study aimed to compare the nutritional quality and consumer preferences for selected TRVs and MRVs cultivated in Kerala, India. We evaluated sensory attributes and physicochemical properties to assess their influence on consumer preference. Sensory rankings were analyzed using Kendall’s W test, while multiple linear regression was used to examine the relationship between consumer preference and various quality parameters. The study found that TRVs had significantly higher antioxidant levels, while MRVs had substantially higher protein contents. Sensory evaluations ranked TRVs more favorably, with grain appearance and taste being key drivers of preference. Physicochemical characteristics also significantly influenced consumer choice. To understand how these preferences influenced purchasing behavior, we conducted exit surveys in supermarkets and applied cluster and discriminant function analyses. The results indicated that both younger consumers and senior residents preferred TRVs in terms of purchase and consumption patterns.

1. Introduction

Rice is one of the earliest cultivated crops in human history, and numerous landraces have emerged over time through natural selection and traditional farming practices. Scientific breeding and selection introduced high-yielding varieties, significantly enhancing rice productivity, but the rate of rice yield increase based on genetic changes has declined in recent decades [1]. The widespread adoption of modern varieties also contributed to the marginalization of traditional rice varieties (TRVs) in most regions.

Rice cultivation encompasses diverse farming systems tailored to different conditions. Traditional cultivation led to the evolution of upland and lowland varieties, salt- and alkaline-tolerant varieties, and varieties valued for their aromatic or medicinal properties. These TRVs have demonstrated long-term resilience to local conditions and typically require lower inputs, making them inherently more sustainable. Given these traits, there is a growing interest in conserving and promoting TRVs for targeted crop improvement and wider cultivation [2]. Farmers preferring to cultivate these varieties are on the rise, owing to their better adaptation to local climatic and soil conditions and evolved resistance to endemic stresses leading to sustainability of the rice cultivation [1].

Although India is considered as a global hotspot of traditional rice biodiversity, modern scientific studies on the nutritional and health benefits of TRVs remain limited. This gap is primarily due to insufficient scientific data and public awareness [3]. Each agro-climatic region has developed local landraces, yet many remain undocumented. Several organizations are working to promote the conservation and cultivation of TRVs, and some studies suggest a growing interest among younger farmers [4,5]. While marketing remains challenging for TRV producers, many rice consumers seem to prefer to consume traditional varieties because of their taste, aroma, texture, and other qualities [2]. Urban consumers appear increasingly willing to pay a premium for these varieties [5]. Despite their reputed health benefits, TRVs display wide variation in biochemical composition [6,7], making scientific validation important for promoting their adoption.

Against this backdrop, our study set out to explore the following objectives:

• To assess the physicochemical and nutritional properties of selected commercially grown TRVs and MRVs.
• To evaluate the sensory (organoleptic) characteristics of these varieties as perceived by consumers.
• To determine whether sensory attributes influence overall consumer preferences.
• To examine correlations between consumer preferences and these varieties’ physicochemical and nutritional profiles.
• To identify and analyze consumer segments and their preferences for TRVs in purchasing and consumption behavior.

2. Materials and Methods

The study was conducted in the six major rice-growing districts of Kerala, India: Kasaragod, Kannur, Kozhikode, Wayanad, Malappuram, and Palakkad [7]. Five widely cultivated MRVs and five TRVs with significant local presence were selected for evaluation (

Table 1. Rice varieties selected for the study.

SI No.	Traditional Varieties	Modern Varieties
1	Valichoori	Shreyas
2	Ayirammeni	Athira
3	Thondi	Uma
4	Chembavu	Jaya
5	Kaipad	Ponmani

).

For the organoleptic evaluation, 100 evaluators, randomly selected from university staff and postgraduate students, were presented with pre-cooked rice samples without disclosing the variety names to ensure unbiased judgment. Each evaluator received adequate portions of each variety to facilitate informed evaluation across multiple sensory attributes.

An exit survey was conducted among 340 customers across 10 supermarkets in Palakkad and Malappuram districts using a semi-structured interview methodology to capture consumer preferences.

2.1. Physicochemical and Nutritional Attributes

2.1.1. Moisture

Grain moisture content was determined using the AOAC method [8]. Rice samples with known grain weights were oven-dried at 60–70 °C to a constant weight. The moisture content was calculated as the difference between the initial and final weights.

2.1.2. Total Carbohydrate

Carbohydrate content was estimated following Sadasivam and Manickam [9]. A 50 mg sample was hydrolyzed with 5 mL of 2.5 N HCl for 3 h, cooled to room temperature, and neutralized with solid sodium carbonate until effervescence ceased. The volume was adjusted to 100 mL and centrifuged. A 0.2 mL aliquot was diluted to 1 mL before adding 4 mL of Anthrone reagent and heating for 8 min. After rapid cooling, absorbance was measured at 630 nm, and carbohydrate content was calculated using a glucose standard curve, expressed in g per 100 g sample.

2.1.3. Protein

Protein content was determined using the AOAC method [8]. Powered rice samples were digested with CuSO₄ (0.4 g), K₂SO₄ (3.5 g), and 6 mL concentrated H₂SO₄ until the sample became colorless. After dilution to 1000 mL with DI water, 10 mL aliquots were treated with 25 mL of 40% NaOH and distilled. The distillate was collected in 20% boric acid with mixed indicators and titrated with 0.2 N H₂SO₄. Nitrogen content was multiplied by 6.25 to determine protein percentage.

2.1.4. Total Fat

Following the AOAC method [2], 5 g of powdered sample was extracted with petroleum ether in a Soxhlet apparatus for 6 h without interruption. After cooling, the ether was removed by heating, with fat content calculated and expressed in g per 100 g.

2.1.5. Fiber

Fiber was estimated using acid–alkali digestion [10]. A 2 g powdered sample was boiled with 200 mL of 1.25% H₂SO₄ for 30 min, filtered through a muslin cloth, and washed sequentially with 25 mL of 1.25% H₂SO₄, 50 mL water (three times), and 25 mL alcohol. The residue was dried, cooled, and weighed before igniting at 600 °C in a muffle furnace for 30 min, cooling in a desiccator, and reweighing. Fiber content was calculated based on weight loss and expressed as a percentage.

2.1.6. Calcium

Calcium was measured using the method described by Page [11]. A 2 g powdered sample was digested with a 10 mL mixture of nitric and perchloric acids (9:4 ratio) and diluted to 100 mL. A 5 mL aliquot was treated with 10 mL water, 10 drops of hydroxylamine hydrochloride, 10 mL triethanolamine, 2.5 mL NaOH, and 10 drops of calcone and titrated with 0.02 N EDTA until the solution had a permanent blue color. Calcium content was expressed as mg per 100 g.

2.1.7. Iron

Iron content was measured colorimetrically using the method described by Raghuramulu et al. [12]. Ferric ions formed a red complex with potassium thiocyanate. A 6.5 mL aliquot of the 9:4 diacid solution was treated with 1 mL of 30% H₂SO₄, 1 mL of 7% potassium persulphate solution, and 1.5 mL of 40% potassium thiocyanate. Absorbance was measured at 540 nm within 20 min, with iron concentration determined using a standard curve and expressed as mg per 100 g.

2.1.8. Phosphorous

Phosphorus content was assessed colorimetrically using Jackson’s method [13]. A 0.2 g sample was pre-digested with 10 mL of the 9:4 diacid solution and diluted to 100 mL. A 5 mL pre-digested aliquot was reacted with nitric acid and vanadatemolybdate reagent (5 mL each), before making the volume up to 50 mL with DI water. Absorbance was read at 420 nm after 10 min, with phosphorus content determined using a standard curve and expressed in mg per 100 g.

2.1.9. Sodium

Sodium was measured using a flame photometer based on Jackson’s method [13]. The digested 9:4 diacid solution was read directly, with the result expressed as mg per 100 g.

2.1.10. Potassium

Potassium content was also measured using a flame photometer method, following Jackson’s method [13]. A 1 mL aliquot of the digested solution was diluted to 25 mL and analyzed, with potassium content expressed as mg per 100 g.

2.1.11. Anthocyanin

Anthocyanin content was determined following Neff and Chory [14]. Samples were extracted with alcoholic HCl (85%), ethanol (95%), and 15% of 1.5 N HCl and incubated at 4 °C in the dark for 48 h. After centrifugation, 1 mL of the extract was diluted to 10 mL, with absorbance measured at 535 nm. Anthocyanin content was calculated using a standard formula.

2.2. Total Antioxidant Activity

Total antioxidant activity was assessed using the phospho-molybdenum method described by Prieto et al. [15]. Rice grain extract (1 mg mL⁻¹) was combined with 3 mL reagent solution (0.6 M sulfuric acid, 28 mM sodium phosphate, and 4 mM ammonium molybdate). Ascorbic acid standards were prepared, with 1 mL of varying standard concentrations combined with 3 mL of reagent solution. Samples were incubated at 95 °C for 90 min and cooled to room temperature, and absorbance was read at 695 nm. Antioxidant activity was expressed as grams of ascorbic acid equivalent.

2.3. Kendall’s W Test

Kendall’s coefficient of concordance (W) was used to assess the degree of agreement among evaluators based on ranked sensory attributes [16,17]. This non-parametric test measures consensus among multiple evaluators rather than individual accuracy.

2.4. Two-Sample Unpaired t-Test

A two-sample unpaired t-test was used to test the null hypothesis that the means of two independent, normally distributed populations were equal [18,19]. The t-statistic was calculated as follows:

$$t={\displaystyle \frac{{x‾}_1-{x‾}_2}{\sqrt{\frac{s^2}{n_1}+\frac{s^2}{n_2}}}}$$

(1)

where ${x‾}_1$ and ${x‾}_2$ are sample means, s² is the pooled variance, and n₁ and n₂ are sample sizes. The test followed Student’s t-distribution, with n₁ + n₂ − 2 degrees of freedom.

2.5. Multiple Linear Regression

Multiple linear regression was used to examine the relationship between dependent (consumer preference) and explanatory variables (sensory attributes and physicochemical properties) [20] of the selected rice varieties. The model is represented as follows:

Y₁ = β₀ + β₁x_1i + β₂x_2i + …… + β_px_pi + e_i

(2)

where Y₁ is the dependent variable, β₀ is the constant term (intercept), β₁ to β_p are coefficients of explanatory variables (x₁ to x_p), and e_i is the error term. The method accommodates both categorical and continuous explanatory variables.

2.6. Cluster Analysis

Cluster analysis helps to sort similar items into different groups based on their characteristics. In this study, we used K-means clustering to cluster data points that share common traits, allowing patterns or trends to emerge. This method helps reveal hidden relationships within the dataset, whether analyzing customer behavior, biological data, or images [21]. The K-means clustering algorithm generates clusters using the cluster’s object mean [22]. It begins by randomly assigning initial cluster centroids in the data space before assigning each data point to a cluster based on its distance from the centroid. After all points have been assigned, new cluster centroids are identified. This process runs iteratively until the algorithm converges and stable clusters are formed.

Here, to better understand consumer preferences for rice varieties, we performed cluster analysis followed by discriminant analysis to validate the cluster classifications.

2.7. Canonical Discriminant Analysis

Canonical discriminant analysis identifies a projection hyperplane in k-dimensional space that maximizes separation between predefined groups. It achieves this by minimizing the distances between projections of observations within the same category and maximizing the distances between them from different categories. The classification results are considered valid if the clusters identified through discriminant analysis align with those obtained from cluster analysis [23].

We used discriminant analysis to validate the results of the cluster analysis. The number of clusters was set as the dependent variable, and the independent variables listed in Table 6 were used in the analysis, carried out in SPSS version 21.

2.8. Canonical Discriminant Function Determination–Wilks’ Lambda Test

Wilks’ lambda test evaluated the significance of each variable in contributing to the discriminant function. Wilks’ lambda values close to zero indicate strong discriminatory power.

3. Results

3.1. Physicochemical and Nutritional Attributes of Selected Traditional and Modern Rice Varieties

The comparative analysis of the selected TRVs and MRVs revealed notable variations in their nutritional profiles, but in general, rice proved to be a valuable source of carbohydrates, protein, fiber, and vitamins.

Moisture content ranged from 10.50% in Sreyas to 13.10% in Athira. Mineral content also differed across the varieties. Sodium content ranged from 2.76 mg 100 g⁻¹ (Chembavu) to 7.67 mg 100 g⁻¹ (Valichoori), potassium content from 70.00 mg 100 g⁻¹ (Athira) to 127.80 mg 100 g⁻¹ (Jaya), phosphorus content from 108.60 mg 100 g⁻¹ (Chembavu) to 226.90 mg 100 g⁻¹ (Jaya), calcium content from 16.60 mg 100 g⁻¹ (Athira) to 27.87 mg 100 g⁻¹ (Kaipad), and iron content from 0.52 mg 100 g⁻¹ (Uma) to 2.80 mg 100 g⁻¹ (Kaipad).

Figure 1 shows the distribution of six selected nutritional attributes in the selected varieties. All varieties exhibited high carbohydrate content, ranging from 71.10 g 100 g⁻¹ (Thondi) to 80.40 g 100 g⁻¹ (Ayirammeni). Fiber content ranged from 1.67 g 100 g⁻¹ (Uma) to 3.88 g 100 g⁻¹ (Thondi), with no significant differences observed between Thondi, Kaipad, Sreyas, and Jaya. Among the TRVs, Thondi and Kaipad exhibited the highest fiber content. Fat content ranged from 1.08 g 100 g⁻¹ (Chembavu) to 2.67 g 100 g⁻¹ (Kaipad), with no significant differences among Kaipad, Thondi, and Jaya. Protein content ranged from 6.10 g 100 g⁻¹ (Chembavu) to 7.90 g 100 g⁻¹ (Uma), with Chembavu showing significantly lower levels than the other varieties.

Anthocyanin and antioxidant contents were generally higher in TRVs, with antioxidants being significantly more abundant. The TRVs Kaipad, Chembavu, and Ayirammeni had the highest antioxidant content among all varieties.

In overall observation, moisture content was highest in Athira (MRV); sodium content was highest in Valichoori (TRV); potassium and phosphorus contents were highest in Jaya (MRV); calcium, iron, fat, anthocyanin, and antioxidant contents were highest in Kaipad (TRV); carbohydrate content was highest in Ayirammeni (TRV); fiber content was highest in Thondi (TRV); and protein content was highest in Uma (MRV).

3.2. Organoleptic Evaluation of the Selected Rice Varieties

Table 2. Kendall’s W test results (mean ranks) for the selected rice varieties: sensory attributes.

Kendall’s W Mean Ranks
Grain Size and Length	Color	Odor	Taste	Overall Acceptance
Ayirammeni (7.02)	Valichoori (6.87)	Valichoori (6.71)	Valichoori (7.05)	Valichoori (6.90)
Valichoori (6.90)	Aayirammeni (6.82)	Aayirammeni (6.27)	Thondi (6.78)	Aayirammeni (6.83)
Thondi (6.22)	Chembavu (5.95)	Thondi (5.78)	Aayirammeni (6.64)	Thondi (6.17)
Sreyas (6.15)	Thondi (5.87)	Uma (5.74)	Uma (6.00)	Chembavu (6.04)
Athira (5.68)	Athira (5.38)	Chembavu (5.57)	Sreyas (5.69)	Sreyas (5.93)
Uma (5.54)	Sreyas (5.24)	Kaipad (5.50)	Chembavu (5.59)	Uma (5.52)
Chembavu (5.51)	Uma (5.11)	Sreyas (5.39)	Athira (5.26)	Athira (5.05)
Kaipad (4.37)	Kaipad (4.94)	Ponmani (4.99)	Ponmani (4.36)	Kaipad (4.72)
Ponmani (4.15)	Jaya (4.51)	Athira (4.76)	Kaipad (4.10)	Ponmani (4.37)
Jaya (3.46)	Ponmani (4.32)	Jaya (4.30)	Jaya (3.52)	Jaya (3.47)

Red: Ayirammeni, Light blue: Valichoori, Orange: Thondi, Peach: Chembavu, Yellow: Sreyas, Pista green: Uma, Green: Athira, Grass green: Kaipad, Mustard: Ponmani and Blue: Jaya.

presents the mean ranks, as determined by Kendall’s W test, for grain size and length, color, aroma, taste, and overall acceptance across the ten rice varieties.

Three TRVs—Ayirammeni, Valichoori, and Thondi—consistently ranked highest across all sensory attributes, while the MRVs generally received lower preference scores from evaluators. The TRV Valichoori held the highest rank in four of the five sensory attributes, namely color, odor, taste, and overall acceptance. The TRV Ayirammeni was the next most preferred, with the second highest rank in three attributes: color, odor, and overall acceptance. Ayirammeni held the highest rank regarding the grain shape and size attribute. TRV Thondi was the third most preferred, having the third highest rank in three attributes: grain shape and size, odor, and overall acceptance. Thondi also had the second highest rank in taste and fourth rank in color. The most widely cultivated MRV in Kerala, Uma, had the fourth highest rank in odor and taste. Another widely cultivated MRV Jaya, was ranked lowest in four attributes, grain shape and size, odor, taste and overall acceptance.

The strong performance of TRVs in the sensory evaluation prompted a closer comparison of their nutritional profiles with MRVs, carried out using a two-sample unpaired t-test (

Table 3. Two-sample unpaired t-test results comparing selected traditional and modern rice varieties.

Attribute	Mean		t-Value	p-Value
	Traditional Varieties	Modern Varieties
Carbohydrate	76.04	75.81	0.115	0.911 NS
Fiber	2.718	2.640	0.142	0.891 NS
Fat	1.732	1.752	−0.05	0.961 NS
Protein	7.014	7.420	−1.479	0.178 **
Sodium	5.358	4.166	1.026	0.335 NS
Potassium	96.040	100.980	−0.388	0.708 NS
Phosphorus	144.72	171.520	−0.929	0.380 NS
Calcium	19.686	21.252	−0.554	0.595 NS
Iron	1.830	1.502	0.581	0.577 NS
Anthocyanin	23.176	24.358	−0.155	0.881 NS
Antioxidants	1.106	0.740	4.083	0.004 *

* significant at the 0.01 level; ** significant at the 0.2 level; NS, not significant.

).

Only protein and anthocyanin featured a significant difference among TRVs and MRVs. Protein content was significantly higher in MRVs, and antioxidant content was significantly higher in TRVs.

3.3. Do Sensory Attributes Influence Consumer Preference?

Table 4. Influence of sensory attributes on overall preference for the selected rice varieties: Multiple linear regression analysis (n = 100).

SI No.	Coefficients	Estimate	Standard Error	t-Value	p-Value
Valichoori
1	Grain shape and size	0.481	0.222	2.162	0.039 *
2	Color	0.012	0.142	0.088	0.931
3	Aroma	0.277	0.105	2.638	0.013 *
4	Taste	0.006	0.121	0.046	0.963
Ayirammeni
1	Grain shape and size	0.277	0.115	2.405	0.022 *
2	Color	0.326	0.109	2.986	0.005 **
3	Aroma	−0.146	0.091	−1.609	0.117 ***
4	Taste	0.539	0.104	5.160	0.000 **
Thondi
1	Grain shape and size	0.351	0.141	2.486	0.019 *
2	Color	−0.052	0.174	−0.296	0.769
3	Aroma	0.320	0.102	3.136	0.004 **
4	Taste	0.052	0.102	0.506	0.617
Chembavu
1	Grain shape and size	0.172	0.127	1.350	0.187 ***
2	Color	0.184	0.139	1.326	0.195 ***
3	Aroma	0.061	0.110	0.553	0.584
4	Taste	0.438	0.113	3.891	0.001 **
Kaipad
1	Grain shape and size	0.883	0.227	3.886	0.001 **
2	Color	−0.277	0.210	−1.315	0.199 ***
3	Aroma	−0.083	0.162	−0.513	0.612
4	Taste	0.514	0.142	3.630	0.001 **
Sreyas
1	Grain shape and size	−0.084	0.261	−0.322	0.750
2	Color	0.423	0.269	1.577	0.126 ***
3	Aroma	−0.076	0.152	−0.503	0.619
4	Taste	0.520	0.162	3.206	0.003 **
Athira
1	Grain shape and size	−0.109	0.214	−0.508	0.615
2	Color	0.029	0.205	0.142	0.888
3	Aroma	0.346	0.142	2.439	0.021 *
4	Taste	0.435	0.197	2.206	0.035 *
Uma
1	Grain shape and size	0.148	0.149	0.997	0.327
2	Color	0.186	0.126	1.474	0.151 ***
3	Aroma	0.088	0.142	0.619	0.541
4	Taste	0.468	0.160	2.931	0.006 **
Jaya
1	Grain shape and size	0.211	0.146	1.447	0.158 ***
2	Color	0.424	0.150	2.833	0.008 **
3	Aroma	0.099	0.112	0.889	0.381
4	Taste	0.048	0.041	1.161	0.255
Ponmani
1	Grain shape and size	0.330	0.139	2.381	0.024 *
2	Color	0.159	0.182	0.874	0.389
3	Aroma	0.228	0.195	1.168	0.252
4	Taste	0.353	0.155	2.273	0.030 *

* significant at the 0.05 level; ** significant at the 0.01 level; and *** significant at the 0.2 level.

presents the multiple linear regression results examining the relationship between individual sensory attributes and overall preference for the selected rice varieties.

Significant predictors of overall preference included grain shape and size and taste for seven varieties, color for six varieties, and aroma for four varieties, with taste being the most influential, followed by grain shape and size. In TRVs, it was the grain shape and size attribute that was found to influence consumer preference most significantly (significant in all five varieties), and in MRVs, taste played the more influential role (significant in four varieties).

3.4. Relationship Between Physicochemical and Nutritional Attributes and Consumer Preference for Selected Rice Varieties

Table 5. Relationship between varietal attributes and consumer preference for selected rice varieties.

SI No.	Variables	Estimate	Standard Error	t-Value	p-Value
1	Carbohydrate	−0.168	0.008	−22.150	0.029 **
2	Fiber	0.450	0.028	16.307	0.039 **
3	Fat	−0.472	0.040	−11.726	0.054 *
4	Protein	0.382	0.038	9.965	0.064 *
5	Sodium	0.108	0.014	7.688	0.082 *
6	Potassium	0.034	0.003	11.027	0.058 *
7	Phosphorus	−0.030	0.002	−14.499	0.044 **
8	Calcium	−0.005	0.008	−0.662	0.628

** significant at the 0.05 level; and * significant at the 0.1 level.

presents the regression analysis results exploring the role of physicochemical and nutritional properties in shaping consumer preferences.

The results indicate that carbohydrate, fiber, fat, protein, sodium, potassium, and phosphorus contents significantly influenced consumer preferences, with higher statistical significance for carbohydrate, fiber, and phosphorus contents.

3.5. Clustering of Consumers Based on Their Preferences

The resulting clusters revealed groups of individuals with similar characteristics, which differed substantially from those in other clusters. Each cluster was interpreted and named according to key demographic and behavioral attributes (

Table 7. Distribution of consumer profile characteristics in the identified clusters.

Attributes		Cluster 1	Cluster 2	Cluster 3	Cluster 4
Age	<30 years	41.18	44.12	8.82	5.88
	30–50 years	30	35.71	17.14	17.14
	51–70 years	24.62	6.15	41.54	27.69
	>70 years	0	33.33	33.33	33.33
Gender	Men	25.33	34.67	22.67	17.33
	Women	32.99	19.59	26.80	20.62
Marital status	Married	26.52	22.73	28.79	21.97
	Unmarried	40	37.5	12.5	10
Education	Secondary school	0	20	20	60
	High school	0	16.67	33.33	50
	Graduate	35.56	30	22.22	12.22
	Above graduation	29.23	23.08	27.69	20
Family size	3–4 members	51.22	7.32	23.17	18.29
	5–6 members	15.25	32.20	32.20	20.34
	>6 members	0	64.52	16.13	19.35
Location	Rural	34.78	21.74	26.09	17.39
	Small town	26.53	42.86	16.33	14.29
	Semi-urban	22.22	24.44	28.89	24.44
	Urban	36.36	4.55	36.36	22.73
	Non-resident Keralite	37.50	25	25	12.50
	Non-resident Indian	50	16.33	12.17	21.50
Occupation	Student	70.59	29.41	0	0
	Government sector	73.53	26.47	0	0
	Private sector	0	100	0	0
	Abroad	31.82	68.18	0	0
	Teacher/ scientist	0	0	61.90	38.10
	Farmer	0	0	66.67	33.33
	Business	0	0	47.06	52.94
	Unemployed	0	0	53.33	46.67
	Others	0	0	100	0
Information about traditional rice varieties	Poor	54.55	13.64	13.64	18.18
	Moderate	20.75	45.28	18.87	15.09
	Good	28.87	18.56	30.93	21.65
Information sources accessed on rice types	Low	43.55	9.68	32.26	14.52
	Medium	27.12	30.51	23.73	18.64
	High	15.69	41.18	17.65	25.49
Criteria for rice variety preference	Low	33.77	27.27	23.38	15.58
	Medium	23.61	23.61	29.17	23.61
	High	34.78	30.43	17.39	17.39
Main purchase source	Public distribution system	30.33	33.33	19.67	16.67
	Local retail shop	28.79	34.85	22.73	13.64
	Supermarket	31.46	22.47	23.60	22.47
	Online	23.25	26.53	21.37	28.85
	Farmer outlet/farmer direct	0	22.22	33.33	44.44
Reason for not buying traditional rice	Higher price	15.22	39.13	45.65	0
	Unavailable	38.10	19.05	42.86	0
	Lack of knowledge	30.43	39.13	17.39	13.04
	Not applicable	34.43	16.39	0	49.18
Willingness to pay a higher price for traditional rice	Not willing	10	52.5	30	7.5
	Up to 5% extra	27.78	30.56	27.78	13.89
	Up to 10% extra	30	15	20	35
	Up to 20% extra	42.55	14.89	21.28	21.28
	Up to 50% extra	50	0	33.33	16.67
	>50% extra	66.67	0	33.33	0
Type of rice bought	Traditional red rice	32.26	16.13	0	51.61
	Traditional white rice	54.55	18.18	22.73	4.55
	Modern red rice	17.14	37.14	45.71	0
	Modern white rice	24.53	33.96	41.51	0

). The initial cluster centroids obtained under each cluster are shown in Table 6, along with their Euclidean distance.

Table 6. Initial cluster centroids.

Attributes	Cluster
	1	2	3	4
Age	2	1	2	2
Gender	2	1	1	2
Marital status	1	2	1	1
Education	4	2	3	4
Family_size	3	2	4	2
Location	5	3	1	2
Occupation	2	1	8	9
Knowledge	1	2	2	3
Information_sources	2	3	1	3
Buying_decision	2	1	1	2
Source_of_purchase	3	2	5	5
Reason_not_buying_TR	2	4	1	4
Willingness_to_pay more_for_TR	5	1	4	2
Type_of_rice	2	3	3	1

Table 6. Initial cluster centroids.

Attributes	Cluster
	1	2	3	4
Age	2	1	2	2
Gender	2	1	1	2
Marital status	1	2	1	1
Education	4	2	3	4
Family_size	3	2	4	2
Location	5	3	1	2
Occupation	2	1	8	9
Knowledge	1	2	2	3
Information_sources	2	3	1	3
Buying_decision	2	1	1	2
Source_of_purchase	3	2	5	5
Reason_not_buying_TR	2	4	1	4
Willingness_to_pay more_for_TR	5	1	4	2
Type_of_rice	2	3	3	1

Table 7 comprehensively summarizes demographic, socioeconomic, and preference-related attributes across the identified clusters. It highlights differences in age, gender, education, occupation, rice purchasing behavior, knowledge levels, and willingness to pay more for traditional rice.

• Cluster 1: Young Cosmopolites

Cluster 1 consisted primarily of young consumers (41% under 30 years; 30% between 30 and 50 years) who were mostly educated and mainly single women with a non-resident Indian or non-resident Keralite background. They typically consumed traditional white rice purchased from supermarkets. Professionally, they were predominantly students and public sector employees. They had minimal knowledge of TRVs and relied on only a few sources of information. The main barrier to purchasing authentic traditional rice was its limited availability. Despite this, they were willing to pay a premium—up to twice the price—for traditional rice.

• Cluster 2: Young Locals

Comprising mainly young consumers (44% under 30 years; 36% between 30 and 50 years), this cluster included educated single men working in the private sector with small-town backgrounds. They preferred modern red rice purchased from local retailers. While they had moderate awareness of TRVs and access to various sources of information, they were deterred by the high cost and the lack of credible information. Given their typically large family sizes (more than six members), they were generally unwilling to pay extra for traditional rice.

• Cluster 3: Senior Urbanites

This group included consumers aged 50 to 70 years from urban areas, many of whom were retired professionals. They preferred modern red rice purchased directly from farmers. They had excellent knowledge of traditional rice, though they consulted only a limited number of information sources. Price and the unavailability of authentic TRVs were key barriers. Despite these issues, and with an average family size of five to six members, many in this cluster were willing to pay more than 50% extra for traditional rice.

• Cluster 4: Senior Semi-Urbanites

Comprising mostly elderly (over 70 years old), less-educated, married women from semi-urban areas with business backgrounds, this cluster favored traditional red rice purchased directly from farmers. Though they had moderate knowledge of TRVs, they accessed various information sources. They were willing to pay extra for traditional rice.

3.6. Confirmation of Cluster Analysis Using Discriminant Analysis

A larger eigenvalue indicates a stronger linear discriminant function [24].

Table 8. Discriminant function results: eigenvalues, canonical correlations, and Wilks’ lambda.

Function	Eigen Value	Percentage of Variance	Canonical Coefficient	Wilks’ Lambda	Chi-Square	Sig
1	5.78	74.4	0.92	0.038	528.57	0.000
2	1.36	17.5	0.76	0.26	218.50	0.000

presents the eigenvalues, canonical correlation coefficients, and Wilks’ lambda values for the discriminant functions.

Function 1 and Function 2 had eigenvalues of 5.78 and 1.36, respectively, suggesting that Function 1 explains most of the variance in the linear combination of variables. The percentage of variance explained by Function 1 was 74.4%, while Function 2 explained 17.5% of the variance. The canonical correlations for these two functions were 0.92 and 0.76, respectively, with a combined variance explanation of 91.9%.

Wilks’ lambda assesses how well each function separates the groups; the closer the value is to 0, the better the discriminatory power. For Function 1, the Wilks’ lambda was 0.038, indicating a strong ability to distinguish between clusters. The chi-square values for both functions were statistically significant (p < 0.001), confirming the robustness of the discriminant functions.

Table 9. Classification results of predicted group membership.

		Cluster	Predicted Group Membership				Total
			1	2	3	4
Original	Count	1	49	2	0	0	51
		2	1	44	0	0	45
		3	0	0	43	0	43
		4	0	0	0	33	33
	Percent	1	96.1	3.9	0.0	0.0	100.0
		2	2.2	97.8	0.0	0.0	100.0
		3	0.0	0.0	100.0	0.0	100.0
		4	0.0	0.0	0.0	100.0	100.0
98.3% of original grouped cases were correctly classified.

shows the classification results for predicted group membership. The analysis correctly classified 98.3% of the original cases, confirming the accuracy of the clustering solution. Each cluster exhibited high classification accuracy, ranging from 96.1% to 100%, further validating the cluster analysis results.

3.7. Cluster Centroids Based on Canonical Discriminant Functions

The results presented in Figure 2 indicate that all group centroids are distinctly separated, suggesting significant differences among the clusters. Among these, Clusters 3 and 4 were the most homogenous internally and appeared less similar to the other clusters.

4. Discussion

The highest fiber content was found in TRVs (Thondi and Kaipad), and the same is true for antioxidant content (Kaipad, Chembavu, and Ayirammeni). This suggests that TRVs may offer subtle health advantages, contributing to their long-standing preference across generations. No clear distinction was evident between TRVs and MRVs for other nutritional parameters.

Mineral content varied across varieties, but no significant differences were observed between TRVs and MRVs—except in the case of sodium. Two TRVs (Ayirammeni and Valichoori) had significantly higher sodium content than MRVs. Conversely, two MRVs (Ponmani and Jaya) had the highest potassium and phosphorus contents.

Overall, TRVs had significantly higher antioxidant levels than MRVs. While MRVs, developed through scientific breeding, had superior protein content, no significant differences occurred between TRVs and MRVs for the other assessed nutrients. Therefore, it cannot be stated conclusively that TRVs are nutritionally superior to MRVs. This finding is consistent with studies by Pillai et al. [5] and Longvah et al. [6], which also reported that no variety exhibited comprehensive nutritional superiority, although some varieties were richer in specific nutrients. Another study found that cargo and polished rice had relatively high average protein contents—8.50% and 7.78%, respectively [25]. Carbohydrates were the dominant macronutrient in all five tested varieties, ranging from 70.94 to 73.57% in cargo/brown rice and 73.85 to 77.11% in polished rice. Moreover, the degree of milling affected the nutritional profile of the grain; unpolished/brown/cargo rice had higher nutrient levels (except carbohydrates) than polished/white rice [25].

Consumers tend to favor TRVs in many regions, often citing superior taste as the reason. Africa is a relevant comparison source due to its widespread consumption of TRVs [26]. As noted by Rutsaert et al. [27], locally produced rice in Africa is generally perceived to be of higher quality than imported rice, although it tends to be less accessible in urban areas (e.g., in Burkina Faso) and is often sold at premium prices (e.g., in Gambia). Local rice producers typically sell their harvests in nearby markets, lacking the market access or logistical infrastructure to supply urban areas. Certain local varieties, such as Gambiaka in Mali and others in Guinea, are popular and command premium prices. In rice-growing regions like Senegal, local rice remains the preferred choice, likely due to better taste and higher trust in its quality.

Studies have identified specific consumer segments that prefer traditional local rice. For instance, in Senegal, older and less-educated consumers—making up approximately 14% of the market—favor traditional local rice [28]. Similarly, in Ghana, a niche market segment of about 14% of consumers preferred traditional varieties. Notably, traditional rice grains were the most accepted among all consumer groups, although cooking methods and eating quality preferences varied [29].

In our study, two prominent consumer segments preferred traditional rice varieties. These segments included a young, cosmopolitan group—primarily financially independent women—and an older, predominantly female, semi-urban group, indicating a promising shift toward increased preference for TRVs. In both rural and semi-urban communities, women are typically the primary decision makers regarding household food choices and also the primary buyers. The emerging trend of young, informed, and financially independent women favoring TRVs suggests a meaningful change in consumer behavior. Both these groups were also willing to pay a premium for traditional rice. In contrast, middle-aged to older consumers from urban and local backgrounds tended to favor MRVs, likely due to economic constraints and lower awareness levels. Many of these individuals were employed in the private sector, which may further influence their purchasing behavior. Increasing consumer awareness about food safety and nutrition security [30,31,32] influences these preferences. These findings could inform adjustments in market structures to accommodate diverse consumer needs and offer appropriate rice varieties.

Grain shape, size, and taste emerged as the key factors influencing consumer preferences. These preferences vary significantly across regions. In West Bengal, India, traditional or local rice varieties were preferred mainly for their distinct taste [33]. Across other global regions, consumer priorities often include taste, color, grain size, cooking quality, cooking time, aroma, cleanness, and origin [34,35,36]. For instance, Costa Rican consumers valued aroma and the water-to-rice ratio during cooking [37]. In Southeast Asia, nutrition, softness, and aroma were prioritized, while South Asia consumers valued grain appearance, satiety, and aroma [38]. In Cambodia, softness ranked highest, followed by taste and aroma [39]. In Congo, flavor, aroma, purity, swelling capacity, breakage rate, and whiteness were the main preferences [40].

5. Conclusions

The comparison between TRVs and MRVs revealed a wide distribution of nutritional attributes across both types of rice. While TRVs generally exhibited higher fiber and antioxidant contents, MRVs had significantly higher protein content. However, beyond these distinctions, there was no conclusive evidence to support the overall nutritional superiority of traditional varieties.

Organoleptic evaluations consistently favored three TRVs, highlighting their superior taste—a key factor influencing consumer preference. Consumer choices were shaped primarily by grain shape, size, and taste. However, physicochemical characteristics such as carbohydrate, fiber, fat, protein, and mineral contents also significantly influenced preferences for cooked rice. Among consumer segments, both younger consumers and older semi-urban residents preferred TRVs and were willing to pay more for them.

Our findings underscore the importance of promoting TRVs demonstrating nutritional value and sensory appeal, though this study has inherent limitations regarding sample size in terms of both varieties and respondents. Targeted extension strategies should be developed to reach specific consumer segments, enabling faster and more effective positive outcomes.

Government food support programs should also consider incorporating TRVs, given their multifaceted benefits beyond nutrition and economics. These varieties offer added value through medicinal and cosmetic applications and have potential suitability for infant nutrition—areas that warrant further investigation. Finally, broader consumer-oriented traits such as taste and grain appearance should guide varietal choices among large-scale rice producers. This holistic approach is essential for preserving the genetic diversity of TRVs and enhancing sustainability and resilience in rice cultivation.