Factors Associated with Vitamin D Deficiency and Their Relative Importance among Indian Adolescents: An Application of Dominance Analysis

Vitamin D deficiency is a serious issue in developing nations, including India. This study investigates the determinants of vitamin D deficiency among Indian adolescents and assesses their relative importance using dominance analysis. Data from the Comprehensive National Nutrition Survey (CNNS) conducted between 2016 and 2018 were utilized in this study. Vitamin D levels were assessed based on serum 25-hydroxyvitamin D concentration, with a sample size encompassing 13,065 adolescents aged between 10 and 19 years. Backward stepwise multivariate logistic regression was used to identify the correlates of vitamin D deficiency, and the relative importance of these factors was assessed using dominance analysis. The study identified nine predictors that were significantly associated with vitamin D deficiency at a 1% level of significance (α = 0.001). Among these factors, sex was found to be the most significant predictor, with female adolescents being 2.66 (95% CI: 95% CI: 2.39–2.96) times more likely to be vitamin D deficient compared to male adolescents. Lifestyle and behavioral factors, such as “sex,” “wealth index,” and “place of residence,” were more dominant in predicting vitamin D deficiency than biological indicators like “BMI” and “serum creatinine.” This underscores the vital role of sunlight exposure in maintaining sufficient vitamin D levels. In summary, this study sheds light on the multifaceted factors contributing to vitamin D deficiency among Indian adolescents, emphasizing the significance of targeted interventions and public health awareness campaigns to mitigate this pressing issue.


Introduction
Vitamin D defciency has emerged as a signifcant global public health concern, impacting more than one billion individuals worldwide.Insufcient serum 25hydroxyvitamin D levels have been associated not only with musculoskeletal health issues like osteoporosis, rheumatoid arthritis, fractures, and bone metabolism but also with an elevated risk of diverse nonskeletal disorders, including type-1 diabetes mellitus, cardiovascular diseases, infectious diseases, multiple sclerosis, autoimmune diseases, depression, schizophrenia, obesity, and chronic obstructive pulmonary diseases (COPD) [1][2][3][4].However, the exact nature of the relationship between vitamin D and some of these diseases remains incompletely understood.It remains uncertain whether vitamin D defciency contributes to the development of these disorders or if the disorders themselves lead to vitamin D defciency [5].Nonetheless, the accumulating body of evidence linking vitamin D defciency to a range of health conditions underscores the signifcance of maintaining sufcient vitamin D levels for overall health and well-being [5].
Vitamin D serves a multitude of critical functions within the body.Its classical role involves the regulation of calcium absorption and the maintenance of mineral homeostasis.Beyond this, it has been found that vitamin D is essential for the absorption of other essential minerals such as iron, zinc, magnesium, and phosphorus [6].In its active form, vitamin D aids in controlling the release of parathyroid hormone (PTH).Excessive PTH levels can result in bone loss and fragility, elevating the risk of osteoporosis [7].In addition, emerging evidence suggests that vitamin D may play a role in regulating insulin secretion and glucose levels in the body [8][9][10].Furthermore, vitamin D contributes to the generation of antimicrobial peptides, bolstering cellular immunity and aiding in the defense against infections [11].Recent research has uncovered that vitamin D plays a role in both innate and adaptive immunity, showcasing its potential as an immunomodulator for autoimmune conditions and cancer.Multiple studies have established a connection between a defciency in vitamin D and the occurrence of psoriasis, autoimmune thyroid disorders, thyroid cancer, and other autoimmune diseases [12][13][14][15].Moreover, vitamin D infuences the proliferation and diferentiation of various cell types and modulates cell growth [16].Te presence of vitamin D receptors in numerous tissues throughout the body indicates a broader spectrum of clinical and physiological functions for this vital micronutrient [17].Given the diverse array of functions it fulflls in the human body, it is evident that vitamin D is an essential nutrient for promoting overall health and well-being.
Vitamin D defciency is an escalating global concern, often receiving insufcient attention.While certain gaps in our understanding of its prevalence persist, research consistently indicates a rising incidence of vitamin D defciency worldwide.For instance, when considering the threshold of 25(OH)D < 20 ng/ml, it is estimated that 24% of Americans, 37% of Canadians, and 40% of Europeans have insufcient levels of vitamin D [18].However, in regions such as Africa, the Middle East, and Asia, vitamin D defciency appears to be even more prevalent [19].In India, despite the abundance of sunlight in this tropical country, multiple small scale studies have reported a high prevalence of vitamin D defciency.Estimates of vitamin D defciency in various population groups in India range from 34% to a staggering 94% [20].Nevertheless, it is worth noting that these studies have been conducted on relatively small sample sizes and may not be representative of the entire nation.
Te adolescent years mark a period of signifcant physical and mental development, underscoring the importance of maintaining adequate vitamin D levels for overall health and well-being.While previous studies have underscored the prevalent vitamin D defciency among Indian adolescents, none of these studies have achieved a large-scale, nationally representative scope.In addition, none have undertaken a comprehensive analysis of the anthropometric and biological factors associated with vitamin D defciency within this population [21].To bridge these gaps in knowledge, the present study seeks to investigate the demographic, socioeconomic, and anthropometric factors associated with vitamin D levels among Indian adolescents.Furthermore, the study employs a novel dominance analysis approach to evaluate the relative signifcance of these factors in predicting vitamin D defciency.To the best of our knowledge, this study represents the world's frst use of this approach for analyzing the determinants of vitamin D defciency.
Te aims of this study are twofold.First, it seeks to explore the relationship between vitamin D defciency and a range of demographic, socioeconomic, anthropometric, and biological factors among adolescents in India.Second, it employs dominance analysis to assess the relative importance of the most signifcant predictor variables.By achieving these objectives, this study aims to contribute to the understanding of the factors that infuence vitamin D status among Indian adolescents, which can inform the development of efective strategies for preventing and addressing vitamin D defciency in this population.

Methodology
Tis study draws upon data obtained from the Comprehensive National Nutrition Survey (CNNS), conducted between 2016 and 2018.Tis comprehensive survey was undertaken by the Indian Ministry of Health and Family Welfare, with technical support from UNICEF, and data management services were furnished by the Population Council.Te CNNS survey encompasses a wealth of information on the nutritional status, body measurements, dietary habits, and micronutrient levels of individuals aged 0-9 years across 30 states in India.Te survey adopted a multistage sampling approach to gather data, with a specifc focus on three age categories: preschoolers, school-age children, and adolescents.Tis study, in particular, concentrates on adolescents aged 10-19 years.Te original planned national sample size for the survey encompassed 122,100 children and adolescents from 2,035 primary sampling units.Tis comprised a sample size of 40,700 individuals for household interviews and anthropometric measurements, as well as 20,350 individuals for biological sample collection.It is worth noting that the individual interview response rate was notably high at 95%.However, the response rate for the collection of biological samples was relatively lower, ranging between 63% and 64%.For further details regarding the survey's methodology and data management, readers are encouraged to refer to the CNNS (2016-18) report [22].
2.1.Collected Parameters.Te information collected in the CNNS can be broadly categorized into three main groups: household interviews, anthropometric measurements, and biological sample collection.In household interviews, data on general socioeconomic and demographic characteristics of the household and information regarding feeding practices and dietary diversity within the household were collected.Anthropometric measurements encompassed the assessment of various characteristics, including height/ length, weight, mid-upper arm circumference (MUAC), triceps skinfold thickness (TSFT), and subscapular skinfold thickness (SSFT).Te survey also encompassed the collection of data related to a wide range of biological indicators, such as hemoglobin levels, C-reactive protein, serum protein and albumin levels, serum ferritin, serum zinc, serum B12, serum 25(OH)D, urinary iodine levels, blood pressure, HbA1C, serum cholesterol, LDL (low-density lipoprotein), HDL (high-density lipoprotein), triglycerides, and serum creatinine.A list of all the anthropometric and biological 2 International Journal of Endocrinology predictor variables included in the analysis is shown in Table 1.Detailed information regarding the equipment used and the method of assessment of the abovementioned parameters can be found in the CNNS report.

Sample Size.
Te planned sample size in the survey for collecting biological samples from adolescents aged 10-19 years was 20,350.However, data on serum vitamin D (the primary outcome variable of this study) were available for only 13,065 adolescents.Te reduction in the sample size could be attributed to factors such as nonresponse, subpar data quality, and invalid observations.Consequently, this study relies on a sample size of 13,065 adolescents.To ensure that the fndings and estimates accurately represent the entire national population, we applied sample weights from the CNNS dataset.

Blood Collection and Estimation of Serum 25(OH)D.
During the data collection process, the study followed established protocols for the collection and estimation of serum 25(OH)D levels.Participants were advised to fast for 8-10 hours prior to blood sample collection, and trained phlebotomists collected 10 ml of venous blood to assess micronutrient status.Te samples were transported to the nearest collection center in cooling bags and stored at −20 °C until analysis.An antibody competitive immunoassay method with a chemiluminescence (Siemens Centaur) method was used to determine the concentration of serum 25(OH)D in the samples.Standard and extensive measures of internal and external quality control were implemented throughout the entire procedure to ensure the validity and reliability of the data.Both parents and children participated in the study.More detailed information about the collection process can be found in the CNNS (2016-18) report [22].To ensure ethical standards, informed written consent was secured from parents or caregivers of children under 10 years old.For adolescents aged 11-17 years, written informed consent was obtained from both parents or caregivers as well as from the adolescents themselves.Respondents aged above 17 years provided their own written informed consent.Tese consent procedures were reported by CNN and are elaborated upon in the CNNS report [22].

Outcome
Variable.Tis study focuses on the level of serum 25-hydroxyvitamin D (25(OH)D) as the primary outcome variable, which is considered the most reliable indicator of vitamin D levels in the body [1].Te 25(OH)D concentration is an indicator not only of the amount of vitamin D produced in the skin but also of the amount absorbed through diet.In this study, 25(OH)D concentration was measured using an antibody competitive immunoassay based on the chemiluminescence (Siemens Centaur) method.To defne vitamin D defciency, this study adhered to the criteria set by the Institute of Medicine (IOM), USA, which designates a serum 25(OH)D level of less than 12 ng/ ml as indicative of defciency [22,23].For the sake of facilitating statistical analysis, the study established a binary variable, where "defcient" corresponds to 25(OH)D levels less than 12 ng/ml (coded as 1) and "adequate" signifes 25(OH)D levels of 12 ng/ml or higher (coded as 0).Te "Region" variable not only served as a control for state-level fxed efects but also functioned as a proxy indicator for altitude, thereby accounting for the potential infuence of altitude in the analysis.

Anthropometric and Biological
Variables.Almost all the anthropometric and biological variables provided in the CNNS dataset were included in the regression analyses as independent variables.

Statistical Analysis.
Statistical analyses in this study were performed using STATA-16 software [24].Backward stepwise multivariate logistic regression was utilized to examine the associations between various factors and vitamin D defciency.Te alpha level for the backward stepwise estimation was set at 1% (0.01), indicating that only predictor variables with strong statistical evidence of an association with vitamin D defciency were considered.In the regression model, odds ratios were reported solely for predictor variables with a p value less than 0.01.It is important to note that the data used in this study had a hierarchical structure, with observations clustered within primary sampling units (PSUs).To account for any potential intracluster correlation, clustered robust standard errors were applied in the regression analysis.
We employed dominance analysis to assess the relative importance of the signifcant predictors of vitamin D defciency (correlations identifed through regression analysis).Tis technique is an extension of multiple regression and measures the importance of predictor variables for an outcome variable [25].Te analysis involves creating all possible combinations of predictor variables and running regression against the outcome variable to calculate the R 2 values for pairwise comparison.Te "domin" and "moremata" packages in STATA-16 were used to conduct dominance analysis.Suppose there are "x" predictor variables in the analysis, then the technique will make 2x − 1 combinations of predictor variables.For instance, if there are fve predictor variables, the technique will make 31 International Journal of Endocrinology Te "domin" package in STATA calculates three statistics: general dominance statistics, conditional dominance statistics, and complete dominance.Te general dominance statistics represent the predictor's additional/marginal contribution in the overall ft statistics across all models in which it was included.If a variable X has a general dominance statistics greater than variable Y, it implies that variable X dominates variable Y. Conditional dominance statistics are the average incremental contribution to the nth order overall model ft statistics.Te nth order refers to the estimation model which has exactly "n" predictor variables.If a predictor variable X has greater conditional dominance statistics than variable Y across all n orders, then it indicates that variable X conditionally dominates variable Y. Te general dominance statistics of a predictor variable are the arithmetic mean of all conditional dominance statistics for that variable.A predictor variable X completely dominates variable Y when the additional contribution of X in each subset model ft is greater than the contribution of Y.In other words, X is more associated with the outcome variable than Y in both pairwise comparison and comparison in the presence of all possible combinations of predictor variables.Complete dominance provides the strongest evidence, as it is purely unaveraged and puts each predictor variable against one another in every possible comparison.More detailed information about the technique can be found in STATA's "domin" module [25,26].

Results
Mean levels of anthropometric and biological parameters are shown in Table 1.Te adolescents had a mean BMI of 18.2 (SD � 4.0), a mean hemoglobin level of 12.9 ng/ml (SD � 1.7), an average serum cholesterol level of 140.6 mg/dl (SD � 32.9), and a mean serum 25(OH)D level of 17.1 (SD � 8.2).
Table 2 presents the sociodemographic characteristics of the study participants and the prevalence of vitamin D defciency across various background variables.Among the participants, 48.6% were female, 56.87% resided in rural areas, and 72.64% identifed as Hindu.In terms of socioeconomic status, 35.05% belonged to the Other Backward Classes, 32.09% came from the ffth quantile (richest) of the wealth index, and 23.77% were from the North-Eastern region of India.Te prevalence of vitamin D defciency exhibited signifcant variations across diferent subgroups of the study population.Among females, the prevalence of vitamin D defciency was higher at 35.47%, while urban residents had a prevalence rate of 35.35%.Sikh adolescents showed the highest prevalence rate at 68.91%, whereas adolescents from afuent households had a prevalence rate of 35.91%.Adolescents from the northern region of the country had the highest prevalence rate at 43.93%.Overall, the prevalence of vitamin D defciency in the Indian adolescent population stood at 24.97%, considering the serum 25(OH)D level cut-of of <12 ng/ml.Table 3 displays the results of stepwise logistic regression that only included predictor variables with a p value less than 0.001, indicating signifcance at a 99% confdence level.Te analysis revealed that the odds of being vitamin D defcient decreased by 5% with each year increase in age (OR: 0.95; 95% CI: 0.93-0.97).Gender disparities were evident, as female adolescents had 2.66 times higher odds of vitamin D defciency compared to their male counterparts (OR: 2.66; 95% CI: 2.39-2.96).Adolescents residing in urban areas were 1.63 times more likely to be vitamin D defcient than their rural counterparts (OR: 1.63; 95% CI: 1.43-1.87).Te analysis also found a positive association between wealth index and vitamin D defciency (OR: 1.17; 95% CI: 1.11-1.23).Tis implies that adolescents from afuent households faced a higher risk of vitamin D defciency compared to those from less afuent backgrounds.Body mass index (BMI) was also found to be signifcant, with a 5% increase in the odds of vitamin D defciency for every unit increase in BMI (OR: 1.05; 95% CI: 1.024-1.074).Conversely, there were negative associations observed between vitamin D defciency and mid-upper arm circumference (MUAC) (OR: 0.96; 95% CI: 0.93-0.98),serum creatinine (OR: 0.28; 95% CI: 0.20-0.98),and glycosylated hemoglobin levels (OR: 0.83; 95% CI: 0.73-0.94).Te odds of vitamin D defciency were found to increase by 4% with each unit increase in TSFT (OR: 1.04; 95% CI: 1.02-1.06).Triceps skinfold thickness (TSFT) displayed a distinct pattern, with a 4% increase in the odds of vitamin D defciency for each unit increase in TSFT (OR: 1.04; 95% CI: 1.02-1.06).Notably, all other predictor variables, including UV index, did not demonstrate signifcant associations with vitamin D defciency at the 99% confdence level (1% level of signifcance).
As there were 9 predictor variables, a total of 511 (2 9 − 1) regressions were run by the software to conduct dominance analysis.Summary of general dominance statistics is presented in Table 4. Column 1 of the table shows general dominance statistics; column 2 shows standardized dominance statistics which is general dominance statistic vector normed or standardized to be out of 100%.From the table, it becomes evident that the "sex" of individuals emerged as the most infuential predictor of vitamin D defciency.Remarkably, "sex" alone accounted for the explanation of 41% of the explained variance in the dependent variable.Following closely, serum creatinine assumed the second-most dominant position, contributing to 16.35% of the explained variance in vitamin D defciency.Te wealth index claimed the third most vital position among the variables, while triceps skinfold thickness (TSFT) and place of residence occupied the 4th and 5th ranks in terms of dominance.Glycosylated hemoglobin was found to be the least dominant factor in explaining vitamin D defciency.
Table 5 shows the conditional dominance among the predictor variables.Te frst column of the table displays the average marginal contribution to the overall model ft statistic with 1 independent variable in the model; similarly, the second column shows the average marginal contribution to the overall model ft statistic with 2 independent variables in the model, and so on.Here also, sex was conditionally International Journal of Endocrinology dominant on all the predictor variables.TSFT was dominant over serum creatinine at 1 st and 2 nd orders, but at higher orders, serum creatinine was signifcantly dominant on TSFT.Glycosylated hemoglobin was found to have the least importance in predicting vitamin D defciency.
Complete dominance designations are presented in Table 6.Te rows of the table correspond to the dominance of the independent variable in that row over the independent variable in each column.Te value "1" indicates that the independent variable associated with the row completely dominates the independent variable associated with the column.On the other hand, the value "−1" indicates that the independent variable associated with the row is completely dominated by the independent variable associated with the column.Te value "0" indicates that there is no complete dominance relationship between the independent variable associated with the row and the independent variable associated with the column.From the frst row, we can conclude that "Age" is completely dominated by "sex," "wealth index," and "serum creatinine."Te second row shows that "sex" is completely dominant over all the eight predictor variables.Te third row of the table depicts that "area" completely dominates "BMI" and "glycosylated hemoglobin" and is completely dominated by "sex" and "wealth index."

Discussion
Tis study aims to evaluate the key predictors that signifcantly infuence vitamin D defciency among adolescents in India.Te results indicate that several variables, including "sex," "serum creatinine," "wealth index," and "place of residence," emerge as the most infuential factors linked to vitamin D defciency among adolescents.In addition, the study underscores the signifcance of BMI, TSFT, and age as other important determinants of vitamin D defciency in this demographic.
Among the selected predictors, "sex" emerged as the most dominant factor in predicting vitamin D defciency.Te study fndings revealed signifcantly higher odds of vitamin D defciency among female adolescents in comparison to their male counterparts, a trend consistent with prior research [27][28][29][30][31]. Te body's production of vitamin D3 relies on exposure to ultraviolet B (UVB) radiation from sunlight [1,32].However, various individual factors, such as skin pigmentation, clothing habits, and the use of sunscreen creams, as well as sun-avoidant lifestyles, can infuence the extent of cutaneous UVB exposure.Te divergence in vitamin D levels between males and females can be attributed to diferences in lifestyle and behavior.Males typically engage more in outdoor activities, which may account for their higher serum 25(OH)D concentration.In addition, the use of sunscreen products and sun-avoidance behaviors tend to be more prevalent among females [27].It is important to note that gender-based discrimination and a cultural preference for sons are widespread in India, factors that could also contribute to the lower levels of vitamin D observed among female adolescents compared to males [33].
Creatinine is a byproduct generated when muscles metabolize creatine, a compound found in the body.It circulates in the bloodstream and is fltered by the kidneys for elimination through urine.Serum creatinine levels serve as indicators of both kidney function and muscle mass.Elevated levels of serum creatinine are indicative of kidney disease, as the kidneys' ability to flter it decreases with declining kidney function [34].Interestingly, this study identifed a negative association between vitamin D defciency and serum creatinine levels, implying that adolescents with higher serum creatinine levels exhibited lower rates of vitamin D defciency.Dominance analysis further underscored the signifcance of serum creatinine as the second most important predictor of vitamin D defciency.A study conducted in the United States also reported a similar association, linking vitamin D activation to increased serum creatinine levels and decreased estimated glomerular fltration rates [35].However, further research is necessary to gain a deeper understanding of this complex relationship.
As per the study results, the "Wealth Index" emerged as the third most dominant predictor vitamin D defciency.In line with prior research fndings, the wealth index displayed a positive association with vitamin D defciency, indicating that individuals with greater wealth were more likely to experience vitamin D defciency compared to those with lower socioeconomic status [36][37][38].Tis observation can be attributed to lifestyle disparities between afuent and less afuent individuals.Individuals with higher socioeconomic status often lead more sedentary lives and may engage less in outdoor activities that expose them to sunlight.Conversely, adolescents from lower socioeconomic backgrounds tend to participate in outdoor activities and have prolonged exposure to sunlight, possibly due to their involvement in work alongside their parents [36][37][38][39].Tese lifestyle distinctions can contribute to the varying prevalence of vitamin D defciency among wealth-related categories.
Triceps skinfold thickness (TSFT) and body mass index (BMI) are commonly used indicators to assess the nutritional status and fat reserves of the body in public health research.Te study fnds a positive association between both BMI and TSFT and vitamin D defciency among adolescents in India, meaning that those with higher BMI and TSFT levels have lower serum 25(OH)D concentration compared to those with lower BMI and TSFT.Tis aligns with previous research that has also established a negative association between BMI and 25(OH)D [37,[40][41][42].Tis can be explained by the fact that high body fat content, which is associated with high BMI and TSFT levels, acts as a reservoir of vitamin D, leading to its sequestration in adipose tissues, thus reducing the amount of vitamin D available in the bloodstream.Te release of vitamin D from fat is slow, resulting in a decreased concentration of serum 25(OH)D in individuals with a high percentage of body fat [43,44].In addition, volumetric dilution may also account for the variability in serum 25(OH)D concentration due to overweight and obesity [45].
As outlined in the study, adolescents residing in urban areas exhibit a higher likelihood of vitamin D defciency in comparison to their rural counterparts.Tis trend is likely attributable to several factors.Rural residents often beneft from increased sunlight exposure due to agricultural work and engagement in outdoor activities.Conversely, urban residents tend to limit their sunlight exposure, and they also face elevated levels of environmental pollution.Tis environmental pollution reduces the amount of UV radiation that reaches their skin, further exacerbating the risk of vitamin D defciency.Furthermore, the study did fnd associations between both mid-upper arm circumference (MUAC) and glycosylated hemoglobin and vitamin D defciency.However, these associations were not as robust as other factors.In fact, these predictors were either completely or conditionally dominated by other covariates, indicating their minimal contribution to explaining the variability in vitamin D defciency.To highlight their limited impact, the conditional dominance statistic at the 9th order was zero for glycosylated hemoglobin and 0.0001 for MUAC.Tis underscores their relatively small role in infuencing vitamin D defciency.
Age was identifed as the 6th most signifcant factor associated with vitamin D defciency in Indian adolescents.However, upon examining the complete dominance table, it was found that the impact of age on vitamin D defciency is entirely overshadowed by "sex," "serum creatinine," and "wealth index."Tis means that regardless of age, females and adolescents with higher socioeconomic status are more likely to be at risk for vitamin D defciency.
Tis study provides valuable insights into the prevalence of vitamin D defciency among Indian adolescents.It underscores the role of lifestyle and individual-level factors in contributing to this issue, ofering a comprehensive understanding of high-risk groups susceptible to vitamin D We believe that this study represents a novel approach to understanding vitamin D defciency and can serve as a cornerstone for future research endeavors on this topic within India.

Limitations
Te current study is not without limitations, and these limitations should be addressed in future data collection for the CNNS.One of the major limitations is the lack of data on important aspects related to vitamin D. For instance, the amount of time spent in sunlight is a critical factor that afects vitamin D levels, but the CNNS does not provide this information.Te absence of this information makes it challenging to adequately control for sunlight exposure in regression analyses, potentially introducing omitted variable bias.In addition, calcium intake is strongly associated with vitamin D levels, but it could not be included in the analysis due to the unavailability of data.Tis omission could limit the study's ability to draw conclusive insights regarding the signifcance of vitamin D predictors.To enhance the reliability and validity of conclusions about vitamin D defciency predictors, it is essential that future studies, including those conducted within the CNNS, collect comprehensive data encompassing all relevant factors related to vitamin D levels.Tis approach will provide a more accurate understanding of the determinants of vitamin D defciency.
International Journal of Endocrinology

Table 1 :
Selected anthropometric and biological variables included in the regression analysis as independent variables.

Table 2 :
Distribution of sample population with respect to demographic and socioeconomic categories and prevalence of vitamin D defciency in respective categories.

Table 3 :
Results of backward stepwise logistic regression model assessing adjusted odds ratios of vitamin D defciency among Indian adolescents.

Table 5 :
Conditional dominance table.Te bold values are those values which are found substantially dominant in the conditional dominance.For example sex, the conditional dominance statistics are highest for sex at each order.