Stature estimation from different combinations of foot measurements using linear and multiple regression analysis in a North Indian male population

Accepted Manuscript Stature estimation from different combinations of foot measurements using linear and multiple regression analysis in a North Indian male population Bahadur Singh, Kewal Krishan, Kawaljit Kaur, Tanuj Kanchan PII:

S1752-928X(18)30160-4

DOI:

https://doi.org/10.1016/j.jflm.2018.12.007

Reference:

YJFLM 1750

To appear in:

Journal of Forensic and Legal Medicine

Received Date: 1 April 2018 Revised Date:

18 December 2018

Accepted Date: 20 December 2018

Please cite this article as: Singh B, K, Kaur K, Kanchan T, Stature estimation from different combinations of foot measurements using linear and multiple regression analysis in a North Indian male population, Journal of Forensic and Legal Medicine, https://doi.org/10.1016/j.jflm.2018.12.007. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT Original research article Stature estimation from different combinations of foot measurements using linear and

Bahadur Singh1, MSc (PhD Research Scholar) Kewal Krishan1, PhD (Associate Professor and Chair) Kawaljit Kaur1, MSc (PhD Research Scholar)

1Department 2Department

M AN U

Tanuj Kanchan2, MD (Associate Professor)

SC

Authors with affiliations:

RI PT

multiple regression analysis in a North Indian male population

of Anthropology, Panjab University, Sector-14, Chandigarh, India of Forensic Medicine and Toxicology, All India Institute of Medical

Sciences, Jodhpur, India.

Corresponding Author

TE D

Conflict of Interest: None to declare

Dr. Kewal Krishan, PhD, FRAI

EP

Associate Professor and Chair, Department of Anthropology,

AC C

Panjab University, Sector-14, Chandigarh, India

E-mail: [email protected], [email protected] +919876048205 (Mobile)

ACCEPTED MANUSCRIPT Original research article Stature estimation from different combinations of foot measurements using linear and multiple regression analysis in a North Indian male population

RI PT

Abstract

Establishing the identity of the deceased is the most important task for forensic anthropologists in forensic case-work involving unidentified human remains.

SC

In such cases, forensic anthropologists examine the remains to derive the biological profile of the deceased i.e. estimation of age, sex, stature, and ethnicity to narrow

M AN U

down the search of the missing. Dismembered remains are recovered in mass disasters such as train mishaps, airplane crashes, earthquakes, and terrorists’ attacks or in homicidal cases where perpetrator intentionally mutilates the dead body to conceal the identity of the victim. Stature estimation is considered as one of the most important

TE D

tasks when a mutilated foot is recovered in process of narrowing down the pool of possible suspects/victims. Allometry is the underlying principle for estimation of stature from foot dimensions. It has been learnt from the published literature that

EP

multiple regression models including more than one factor enhances the estimation

AC C

accuracies. Among the various foot dimensions, foot length is the most frequent parameter used in the estimation of stature in forensic literature. In the present study, an attempt has been made to standardize the stature estimation models from various possible combinations of foot dimensions. For this purpose, 388 Jatt Sikh males aged between 18 and 30 years were recruited from various villages of Ludhiana district of Punjab State in Northern India. Stature, five foot length measurements, and two foot breadth measurements were taken on each subject. Linear and multiple regression models were derived for the estimation of stature from various foot measurements.

1

ACCEPTED MANUSCRIPT The highest coefficient of determination and estimation accuracy (the least standard error of estimation S.E.E) was observed from T1 (R2 = 0.397, S.E.E = 4.7109) when a single foot dimension was included in the model, (R2 = 0.416, S.E.E = 4.6425) from (T1, T3) when two-foot lengths were taken, (R2 = 0.418, S.E.E = 4.6426) from (T1,

RI PT

T3, T4) when three-foot lengths were included, (R2 = 0.418, S.E.E = 4.6473) from (T1, T3, T4, T5) when four-foot lengths were included, and (R2 = 0.418, S.E.E = 4.6531) when all the five foot lengths (T1, T2, T3, T4, T5) were included in the

SC

regression model. It has been concluded that multiple regression models provide more accurate results than linear regression models. However, inclusion of a factor having a

M AN U

weak correlation with stature in the regression model, decreases the accuracy of the model.

Keywords: Forensic anthropology; Foot anthropometry; Stature estimation; North

1. Introduction:

TE D

Indian population; Jatt Sikhs; Linear and multiple regression models

Establishing the personal identity of the deceased is the foremost criterion in

EP

forensic anthropological cases involving unidentified human remains. In these

AC C

circumstances, forensic anthropologists use methods and techniques of physical anthropology for identification purposes. The identification criteria are based on the principle that every individual possesses a unique biological profile constituting of age, sex, stature, and ethnicity. The identification process is much easier when the entire non-decomposed body is recovered. However, in cases of homicides, airplane crashes, train mishaps, tsunamis, flash floods or terrorist attacks, recovery of an entire body is unlikely. In such circumstances, the identification procedure depends upon the material brought in the laboratory and the state of decomposition. In disasters, the feet

2

ACCEPTED MANUSCRIPT may be recovered intact as they are usually protected in the shoes. In such situations, the foot is likely to withstand heat and other climatic factors more effectively; however, other taphonomic factors such as scavenging may change the morphology of the foot.

RI PT

Apart from age, sex, and ancestry, stature is one of the important attributes of individual’s biological profile. The estimation of stature from the recovered body parts can help in the potential search of the missing and thus, enhance the pace of

SC

identification process. Allometry is the underlying phenomenon involved in studies on stature estimation from body parts. Allometry reflects the synchronic relationship

M AN U

of growth and development between different body parts of an individual. In forensic anthropology literature, it has been comprehended that stature can be estimated with reasonable accuracy from face1, head2, upper limbs3, lower limbs4, hands5–7 and from feet8,9.

TE D

Foot being the weight bearing part of the human body evolves to provide balance, support, and propulsion instead of grasping as in other primates10. Males possess on an average larger body weight

11

and size

12

EP

the foot and other body parts shows sexual dimorphism

than females, consequently, 13,14

and variable allometric

AC C

relationship with stature in both the sexes15. It is evident from the literature that individual foot can provide reasonable estimates of stature when other parts of the body such as the long bones are not available for examination 8,9,16. It has been shown that stature can be estimated with great accuracy even from the footprints16–19. In a study conducted by Krishan20 on 1040 North Indian Gujjar males, seven-foot dimensions were recorded from the left and right foot. The study observed that all the foot measurements exhibit a statistically significant correlation with stature. Division factor and regression analysis were used to formulate stature estimation models from

3

ACCEPTED MANUSCRIPT foot dimensions. It has been observed that, multiple regression models predict stature with greater accuracy than the linear regression models, which in turns predicted the stature better than division factors4,17,20–22. In another study, Krishan et al.23 tried to understand the effect of growth factors on the stature estimation from foot

RI PT

dimensions. This study was undertaken on 154 North Indians with age ranges from 13 to 18 years. Five-foot lengths and two-foot breadths were recorded from the left and right foot. This study not only formulates the stature estimation models from all foot

SC

dimensions but also exhibited age as a critical factor in stature estimation from a growing foot.

M AN U

It is the duty of a forensic anthropologist to extract the maximum information from available material for examination to make reliable judgments about identification. Among the various foot dimensions, foot length is the most frequent parameter used in the estimation of stature in forensic literature5,24–26. In forensic

TE D

casework, when one or more than one toe is missing, then multiple regression models developed from different combinations of the foot dimensions could be of great utility in identification. In the present study, an attempt has been made to provide stature

AC C

analysis.

EP

estimation models from all possible combinations of foot dimensions using regression

In forensic casework, DNA fingerprinting has revolutionized the process of

identification27 and the field is rapidly growing. Still, forensic anthropometry28 is maintaining its way in forensic identification, especially in developing countries. Reliability and cost-effective approach of the forensic anthropometry are the key factors of its universal recognition and advancement. 2. Materials and methods 2.1 Subjects

4

ACCEPTED MANUSCRIPT A male cohort of 388 adults Jatt Sikhs was recruited for the present study. The data were collected from male participants only, as in rural households the social norms restrict carrying out research activities on females by a male investigator. The sample consisted of adult males, ranging between 18 and 30 years with the mean age

RI PT

being 21.38 (± 2.79) years. All the subjects belonged to Jatt Sikh caste, which forms a major caste group in North India. Agriculture and animal husbandry are their major occupations. The study area included different parts of district Ludhiana (Punjab State

SC

of north India) consisting eleven villages namely Sarinh, Alamgir, Dhandra, Mehmoodpur, Gill, Dhandari, Pmal, Dakha, Sadhar, Barwala, and Narangwal.

were used for data collection. 2.2 Ethical clearance

M AN U

Sampling techniques such as door-to-door survey, convenient and snowball sampling

This study is a part of an on-going doctoral research being conducted by one

TE D

of the authors (BS) in the Department of Anthropology, Panjab University, Chandigarh, India. Panjab University Institutional Ethical Committee granted the

25/10/2013.

EP

ethical approval to undertake the study vide letter no. PU/IEC/100/13/09 dated

AC C

2.3 Methodology

Only healthy subjects were enrolled in the present study. Subjects with

deformity of the foot, lower limb, spine or injury were excluded from the sample. Along with stature, seven-foot measurements i.e. five-foot lengths and two-foot breadths on the left and right foot were recorded from each subject. For all the anthropometric measurements, standard techniques and procedures were adopted following Gordon et al.29 and Krishan20. All the measurements were taken by a single investigator (BS) to minimize the inter-observer error. Prior to taking measurements,

5

ACCEPTED MANUSCRIPT the subjects were asked to remove shoes/slippers and socks. All of the measurements were taken between 11 am to 4 pm to avoid the effect of diurnal variation on different body dimensions. 2.4 Anthropometric measurements:

RI PT

2.4.1 Stature: Stature is the straight vertical distance between the point height vertex and the floor when the head kept in the FH-plane. The subjects were asked to remove shoes/slippers/socks and stand erect against the wall. The stature was measured with

SC

the help of anthropometric rod to the nearest centimeter (Figure 1).

2.4.2 Foot measurements: There are five-foot length measurements and two-foot

M AN U

breadth measurements on each side of the individual. All of the measurements were taken with the help of an anthropometric rod compass.

Five-foot length measurements are as follows (Figure 2):

1. T1 Length: Distance from pternion (pte) to the most anterior part of the first toe

TE D

(d1.t).

2. T2 Length: Distance from pternion (pte) to the most anterior part of the second toe (d2.t).

AC C

(d3.t).

EP

3. T3 Length: Distance from pternion (pte) to the most anterior part of the third toe

4. T4 Length: Distance from pternion (pte) to the most anterior part of the fourth toe (d4.t).

5. T5 Length: Distance from pternion (pte) to the most anterior part of the fifth toe (d5.t). Two-foot breadth measurements are as follows (Figure 2): 6. Foot-breadth at ball (FBB): Distance between the joint of the anterior epiphyses of the first metatarsal (mt.m), the most prominent part of the inner side of the ball of the

6

ACCEPTED MANUSCRIPT foot, and the joint of the anterior epiphyses of the fifth metatarsal (mt.l), the most prominent part of the outer side of the ball of the foot. 7. Foot-breadth at heel (FBH): Distance taken from the lateral side of the heel (ctu.l)

RI PT

to the medial side (cc.m) of the heel. For detailed procedure, landmarks and figures, please refer to the study by Krishan20. 2.5 Statistical considerations

SC

The data were first entered in MS Excel and then transported to SPPS version

M AN U

16.0 for further statistical analyses. Descriptive statistics such as range, mean and standard deviation were calculated for each dimension in left and right foot. Moments of skewness and kurtosis were calculated for the assessment of normality in the data. Bilateral asymmetry in each foot dimension was analyzed using paired t-test statistics. Karl Pearson's correlation coefficients were calculated between each foot dimension

TE D

and stature. Linear regression models were derived from each foot dimension one at a time. For multiple regression models, different combinations of foot dimensions were

EP

used. Stature was taken as dependent variable and foot dimensions as independent variables for derivation of regression models. Level of significance was set at p<0.05.

AC C

Cohen’s d statistic was used for comprehensive comparison in the discussion. 3. Results

Descriptive statistics of the sample is presented in Table 1. The average

stature of the sample is 175.32 cm with ±6.06 cm standard deviation. For foot length at first toe (T1) on left and right side, the mean value comes out to be 26.821 and 26.824, for second toe (T2) 26.584 and 26.586, for third toe (T3) 25.729 and 25.683, for fourth toe (T4) 24.27 and 24.224 and for fifth toe (T5) 22.417 and 22.378, respectively. For foot breadth measurements the mean values for right and left foot 7

ACCEPTED MANUSCRIPT breadth at ball (FBB) comes out to be 10.447 and 10.497, respectively. Similarly, for the foot breadth al heel (FBH), the mean values for right foot comes out to be 6.705 and 6.715 for the left foot. The least standard deviations have been observed in foot breadth measurements.

RI PT

Before proceeding to apply inferential statistics, the distribution of the data was checked by computing the moments of skewness and kurtosis. In the present study, the sample size is larger than 300, which according to Kim

30

falls in the

SC

category of larger sample size. Kim30 suggests that skewness and kurtosis values from a large sample size must be compared with the references values given by West et al. . All the values of skewness and kurtosis in the present investigation are shown in

M AN U

31

Table 2. The values are less than the reference values, suggesting the normality of the data, and the need to employ parametric tests for further analysis. Bilateral asymmetry in the left and right foot measurements is shown in Table

TE D

3. Differences observed in the mean values of different foot measurements were not statistically significant, except for the foot length at the third toe (T3), and foot breath at ball. In the present study, the data were collected by only one observer (BS), so it is

EP

important to check for the intra-observer error. Technical error of measurement

AC C

(TEM), relative technical error of measurement (rTEM) and coefficient of reliability (R) were calculated for all the anthropometric variables used in the present study and the corresponding values are presented in Table 4. According to Ulijaszek and Kerr 32

, a coefficient of reliability (R) greater than 0.95 is acceptable in anthropological

studies. In the present case, all the corresponding values of the coefficient of reliability (R) are greater than 0.95 suggesting intra-observer reliability. The reliability of any estimation model depends upon the strength of correlation between the variables, larger the correlation, more reliable the model will

8

ACCEPTED MANUSCRIPT be. Karl Pearson’s correlation coefficients have been calculated for each foot measurement from both the sides (left and right) with stature (Table 5). Variable correlation has been observed in different foot measurements. However, the highest correlation with stature has been observed (r = 0.632, r = 0.630) from the left foot

RI PT

length at the third toe (T3) and the right foot length at the second toe (T2), respectively. On the other hand, least association with stature has been observed (r = 0.284, r = 0.309) from the right and left foot breadth at heel (FBH) measurements,

SC

respectively.

Linear regression models were developed using a one-foot dimension at a time

M AN U

(Table 6). The stature estimation model using the right foot length at third toe (T3) explained the largest variance (R2 = 0.399) with the least standard error of estimation (S.E.E. = 4.7040), while the model developed for right foot breadth at heel (FBH) yields with least explained variance (R2 = 0.081) and highest standard error of

TE D

estimate (S.E.E. = 5.8182). All the other models show variable accuracies for stature estimation within these limits.

Multiple regression models were developed using all possible combinations of

EP

two-foot dimensions at a time. Total ten combinations of two different foot lengths from left and right side are presented in Table 7 along with the estimation models,

AC C

coefficients of determination (R2) and standard error of estimation (S.E.E). Using two-foot dimensions at a time, the highest estimation accuracy (R2 = 0.416, S.E.E = 4.6425) had been observed, when right foot length at first toe (T1) and third two (T3) were used. The least estimation accuracy (R2 = 0.356, S.E.E = 4.8762) was observed with the combination of right foot length at the fourth toe (T4) and fifth (T5). Similarly, three-foot measurements were taken at a time to construct stature estimation models and the same are presented in Table 8. The highest accuracy for

9

ACCEPTED MANUSCRIPT stature estimation (R2 = 0.418, S.E.E. = 4.6426) was shown for the combination of right foot lengths at first toe (T1), third toe (T3), and fourth toe (T4). The least accuracy (R2 = 0.390, S.E.E. = 4.7540) was observed from the combination of three left foot lengths at first toe (T1), fourth toe (T4), and the fifth toe (T5).

RI PT

Four-foot lengths were taken together to formulate stature estimation models (Table 9). The highest and the least estimation accuracies were observed from the same combination of four-foot measurements namely foot length at first toe (T1),

SC

third toe (T3), fourth toe (T4) and fifth toe (T5) on the right and left side respectively. The highest estimation accuracy was shown for the right foot length combinations (R2

of variables from the left side.

M AN U

= 0.418, S.E.E = 4.6473), and the least (R2 = 0.404, S.E.E = 4.7027) for the same set

Table 10 shows models for stature estimation from all the five length dimensions of the foot and those using all seven measurements on foot that included

TE D

all length and breadth dimensions of the foot. The coefficient of determination and standard error of estimation for all left and right foot length dimensions came out to be (R2 = 0.410, S.E.E = 4.6851) and (R2 = 0.418, S.E.E = 4.6531), respectively. In

EP

this case, the foot lengths on the right side showed greater estimation accuracy. When all foot measurements were considered for model approximation, right foot

AC C

measurements show greater accuracy (R2= 0.422, S.E.E = 4.6521) in stature estimation than the left foot measurements (R2= 0.410, S.E.E = 4.6962). 4. Discussion

Parts of the body are generally available for identification in mass disasters or in homicides, where dead bodies are purposely disfigured to conceal the identity of the victim33. The state of decomposition is another critical aspect in examination of human remains, which must be kept in mind to apply the available formulae available

10

ACCEPTED MANUSCRIPT for identification. It is well established fact in the forensic literature that the estimation models developed from one ethnic populations should be used on a similar one for the better accuracy17. In the present study, stature estimation models are derived from different combinations of foot dimensions in the Jatt Sikh male

RI PT

population of Northern part of India.

The human foot attains its maximum size up to the age of 16 years and adult stature is attained by the age of 18 years in males following which, minor incremental

SC

changes in foot size and stature may be noted. Later part of the adulthood also realizes a decline in stature of an individual. Therefore, 18 to 30 years is the only age group

M AN U

which ensures minimum age-related effects on the body size of an individual. The present study, thus was planned to include participants aged between 18 and 30 years. In the present study, the mean foot lengths (T1,T2) and foot breadth (FBB, FBH), were marginally larger on the right side, while mean foot lengths (T3, T4, T5),

TE D

were larger on the left side. All the mean foot dimensions were larger than that reported in 18-year male sub-sample of an earlier reported study sample from another North Indian state of Haryana23. The larger mean values from the present study can be

EP

attributed to the fact that males continue to grow in stature even after 18 years34 and

AC C

that all the participants in the present investigation were aged ≥18 years. The mean values from both left and right foot are also larger as compared to North Indian Gujjar population20, which may be attributed to genetic and nutritional factors. Gujjars, Rajputs and Jatt Sikhs make three large ethnic communities of north India. A metaanalysis has been performed using Cohen’s d method to comprehensively elucidate the differences in Jatt Sikhs and Gujjars (Table 11). Cohen (1988) provided reference values (in standard deviation) as 0.1, 0.5 and 0.8 for small, medium and large differences, respectively, to estimate the effect size in two groups. Further, these

11

ACCEPTED MANUSCRIPT values were modified by Sawilowasky 35 as 0.01, 0.2, 0.5, 1.20 and 2.0 for very small, small, medium, large and very large, respectively. For stature, the Cohen's d value came out to be -0.39 SD, which according to reference values is medium. The Cohen's d values for all other foot dimensions are presenting medium to large

RI PT

difference in these two populations. In Table 12, the comparison of the present study with other national and international studies has also been presented using Cohen’s d method. Only stature and foot length (T1) from the present study were used for

SC

comparison with earlier studies. For stature, Turkish population showed the largest differences (-2.73 S.D.) while Sudanese Arab showed the least difference (-0.10 S.D.)

M AN U

from the present population. In terms of the foot length, the highest (-3. 72-R, -3. 74L S.D.) and the least (-0.30-L S.D.) differences were indicated by Malaysian Chinese and Sudanese Arab, respectively. All other studies indicated medium to very large differences from the present study in stature as well as foot length. The comparison

TE D

with other studies also strengthens the fact that the present population is reasonably different from others, and thus, justifies the need for deriving models for stature estimation specific to this population.

EP

It is crucial to judge whether the parametric or non-parametric test should be incorporated for further analysis, for this reason, skewness and kurtosis were

AC C

computed to test the normality in the data. According to Kim30 a sample of 300 subjects or more falls under the range of large sample size and the values of skewness and kurtosis must be compared with the reference values provided by West31 , which are 2 for skewness and 7 for kurtosis, it has also been mentioned that values of kurtosis obtained using SPSS must be first added with a factor of 3, as SPSS provides it by subtracting that. It has been observed that all the values of skewness and kurtosis

12

ACCEPTED MANUSCRIPT are less than the reference values, confirming the normality in the data and support to use parametric tests for further analysis. The bilateral difference in body parts is an important consideration while modeling for stature estimation36. In general, a human body is symmetrical about its

RI PT

axis, meaning that the left side of the human body is the mirror image of the right side. However, in practical situations, because of right or left side domination or sometimes because of occupational pressure, one side tends to be larger than the

SC

other36. The bilateral asymmetry in left and right foot dimensions was tested using the paired t-test. Left foot dominant asymmetry was found in T3, while right foot

M AN U

domination was found only in FBB. The results are in contradiction with the earlier published study by Krishan20, in which the bilateral differences were found in foot length at first toe (T1), second toe (T2) and fifth toe (T5). Considering these differences, side check was incorporated in model formulations.

TE D

In the present study, all the foot measurements exhibited a statistically significant correlation with stature, which advocates their use for stature estimation. However, the extent of correlation is lower than other two reported studies from

EP

North India20,23. In the present study, the highest correlation was observed in the right

AC C

T3 (r = 0.632) while in Gujjars20 the highest correlation was reported for T1 (r = 0.86). The inherent default in the definition of foot length, which usually refers to longest foot length measurement as the one at first toe or at times at second toe, restricts the possible comparison of present work with other studies. For the linear regression models derived in the study, the highest accuracy i.e. the least S.E.E. (standard error of estimate) was observed from right foot dimension at T3. In this case, the model explained almost 40% (R2 = 0.399) of the variability in the data. The least accuracy was observed in the right FBH. As compared to other

13

ACCEPTED MANUSCRIPT published studies from North India, the highest accuracy was reported from the foot length at the first toe (T1)20,23. In Table 13, comparison of the coefficient of correlation (r) and standard error of estimation (S.E.E) from different studies has been presented.

RI PT

In the present study, multiple regression models were also developed for stature estimation using all possible combinations of foot lengths from both left and right foot. The estimation accuracies of these models could not be compared with

SC

other studies; as to the best of our knowledge, no similar studies where estimation accuracies have been calculated are available for comparison. Hence, the comparisons

M AN U

were only made between accuracies of models with different combinations of foot lengths in the present investigation. In estimation accuracy in multiple regression models was higher than that observed in linear regression models. Greater estimation accuracy (R2 = 0.418, S.E.E =4.6426) and the least estimation accuracy (R2 = 0.390,

TE D

S.E.E = 4.7540) were achieved when three-foot lengths were used together, from right (T1, T3, T4) and left (T1, T4, T5) foot respectively (Table 8). These results are better in comparison with linear models but the upper limit of the accuracy declined by a

EP

factor of 0.0001 when compared with the upper limit accuracy when two-foot lengths

AC C

were used at a time (Table 7). The factor is very negligible; however, it shows that the highest accuracy did not improve. The lowest and the overall accuracy enhanced considerably as compared to Table 7. In Table 9, ten stature estimation models using four-foot lengths at a time from both the right and left side of the foot are presented along with their stature estimation accuracies. The highest (R2 = 0.418, S.E.E = 4.6473) and the least (R2 = 0.404, S.E.E = 4.7027) estimation accuracies came from the same combination (T1, T3, T4, T5) of foot lengths from right and left foot sides, respectively. In comparison to the combination right (T1, T3, T4), the coefficient of

14

ACCEPTED MANUSCRIPT determination (R2 = 0.418) remains the same but S.E.E decreased by a factor of (0.0047). However, the lower limit of accuracy improved by a factor of (0.0513), as compared to the S.E.E from left (T1, T4, T5). In Table 10, stature estimation models are presented, first by considering all but only foot lengths and secondly by

RI PT

considering all foot lengths along with foot breadths from left and right foot sides. In the first case, when all but only foot lengths were considered then the coefficient of determination remains the same (R2 = 0.418), but the estimation accuracy decreased

SC

by a factor of 0.0058 when compared to highest accuracy obtained by right (T1, T3, T4, T5) (Table 9).

The comparison of the lowest and the highest estimation

M AN U

accuracies from different foot length combinations has been presented in Table 14. 5. Conclusion

In the present study, linear and multiple regression models were derived for the estimation of stature from various foot measurements. It has been observed that

TE D

multiple regression models provide better stature estimation than the linear regression models. The strength of the correlation tends to be enhanced when more than one associative factor was introduced in the derivation of regression models. This study

EP

was performed to standardize a different set of foot length combinations to help in

AC C

situations when all the foot lengths of an unidentified individual foot are not available for examination due to different taphonomic factors. In light of the results, it can be concluded that stature could be estimated with reasonable accuracy from different combinations of foot lengths. The variability within the estimation accuracies from a different set of foot length combinations indicates that the investigation officer/forensic scientist must be cautious while applying these models in stature approximation. Sex and age are the limiting factors of the present study. Similar studies are suggested among females and in different age groups, so that sex and age-

15

ACCEPTED MANUSCRIPT specific models can be formulated for the estimation of stature. Other caste groups from the same region should also be studied to observe inter-caste variations, if any. References Krishan K. Estimation of stature from cephalo-facial anthropometry in north Indian population. Forensic Sci Int. 2008;181(1-3):1-6. doi:10.1016/j.forsciint.2008.08.001. 2.

Chourasia R, Bamne A, Gohia V, Pandey R. Estimation of stature from head

SC

length. Int J Med Sci Res Pract. 2014;1(2):40-43. 3.

RI PT

1.

Mahakizadeh S, Moghani F, Moshkdanian G, Mokhtari T, Hassanzadeh G. The

M AN U

determination of correlation between stature and upper limb and hand measurements in Iranian adults. Forensic Sci Int. 2016;260:27-30. doi:10.1016/j.forsciint.2015.12.005. 4.

Ahmed AA. Estimation of stature using lower limb measurements in Sudanese

TE D

Arabs. J Forensic Leg Med. 2013;20(5):483-488. doi:10.1016/j.jflm.2013.03.019. 5.

Krishan K, Sharma A. Estimation of stature from dimensions of hands and feet

EP

in a North Indian population. J Forensic Leg Med. 2007;14(6):327-332.

6.

AC C

doi:http://dx.doi.org/10.1016/j.jcfm.2006.10.008. Ishak NI, Hemy N, Franklin D. Estimation of stature from hand and handprint dimensions in a Western Australian population. Forensic Sci Int. 2012;216(13):199.e1-199.e7. doi:10.1016/j.forsciint.2011.09.010.

7.

Jee S, Yun MH. Estimation of stature from diversified hand anthropometric dimensions from Korean population. J Forensic Leg Med. 2015;35:9-14. doi:10.1016/j.jflm.2015.06.014.

8.

Agnihotri AK, Purwar B, Googoolye K, Agnihotri S, Jeebun N. Estimation of

16

ACCEPTED MANUSCRIPT stature by foot length. J Forensic Leg Med. 2007;14(5):279-283. doi:10.1016/j.jcfm.2006.10.014. 9.

Davies CM, Hackman L, Black SM. The foot in forensic human identification A review. Foot. 2014;24(1):31-36. doi:10.1016/j.foot.2013.12.001. Harcourt-Smith WEH, Aiello LC. Fossils , feet and the evolution of human bipedal locomotion. J Anat. 2004;204(5):403-416.

11.

RI PT

10.

Ogden CL, Fryar CD, Carroll MD, Flegal KM. Mean Body Weight, Height,

SC

and Body Mass Index: United States 1960-2002. Department of Health and Human Services, Centers for Disease Control and …; 2004.

Gustafsson A, Lindenfors P. Human size evolution: No evolutionary allometric

M AN U

12.

relationship between male and female stature. J Hum Evol. 2004;47(4):253266. doi:10.1016/j.jhevol.2004.07.004. 13.

Atamturk D. Estimation of sex from the dimensions of foot, footprints, and

14.

TE D

shoe. Anthropol Anz. 2010;68(1):21-29.

Hemy N, Flavel A, Ishak N-I, Franklin D. Sex estimation using anthropometry of feet and footprints in a Western Australian population. Forensic Sci Int.

Zeybek G, Ergur I, Demiroglu Z. Stature and gender estimation using foot

AC C

15.

EP

2013;231(1-3):402-e1.

measurements. Forensic Sci Int. 2008;181(1-3):54.e1-5. doi:10.1016/j.forsciint.2008.08.003.

16.

Robbins LM. Estimating height and weight from size of footprints. J Forensic

Sci. 1986;31(1):143-152. http://www.ncbi.nlm.nih.gov/pubmed/16151444. 17.

Hemy N, Flavel A, Ishak NI, Franklin D. Estimation of stature using anthropometry of feet and footprints in a Western Australian population. J Forensic Leg Med. 2013;20(5):435-441. doi:10.1016/j.jflm.2012.12.008.

17

ACCEPTED MANUSCRIPT 18.

Reel S, Rouse S, Vernon OBE W, Doherty P. Estimation of stature from static and dynamic footprints. Forensic Sci Int. 2012;219(1-3):283.e1-283.e5. doi:10.1016/j.forsciint.2011.11.018.

19.

Krishan K. Estimation of stature from footprint and foot outline dimensions in

RI PT

Gujjars of North India. Forensic Sci Int. 2008;175(2-3):93-101. doi:10.1016/j.forsciint.2007.05.014. 20.

Krishan K. Determination of Stature From Foot and Its Segments in a North

doi:10.1097/PAF.0b013e3181847dd3.

Nor FM, Abdullah N, Mustapa A-M, Faisal NA, Nazari DAAA. Estimation of

M AN U

21.

SC

Indian Population. Am J Forensic Med Pathol. 2008;29(4):297-303.

stature by using lower limb dimensions in the Malaysian population. J Forensic Leg Med. 2013;20(8):947-952. doi:10.1016/j.jflm.2013.09.006. 22.

Kim W, Kim YM, Yun MH. Estimation of stature from hand and foot

TE D

dimensions in a Korean population. J Forensic Leg Med. 2018;55(January):8792. doi:10.1016/j.jflm.2018.02.011. 23.

Krishan K, Kanchan T, Passi N, DiMaggio JA. Stature estimation from the

EP

lengths of the growing foot-A study on North Indian adolescents. Foot.

AC C

2012;22(4):287-293. doi:10.1016/j.foot.2012.08.005. 24.

Hisham S, Rozid Mamat C, Ibrahim MA. Regression analysis for stature estimation from foot anthropometry in Malaysian Chinese. Aust J Forensic Sci. 2012;0618(December 2013):1-9. doi:10.1080/00450618.2012.673637.

25.

Uhrová P, Beňuš R, Masnicová S, et al. Estimation of stature using hand and foot dimensions in Slovak adults. Leg Med. 2015;17(2):92-97. doi:10.1016/j.legalmed.2014.10.005.

26.

Sen J, Ghosh S. Estimation of stature from foot length and foot breadth among

18

ACCEPTED MANUSCRIPT the Rajbanshi: An indigenous population of North Bengal. Forensic Sci Int. 2008;181(1–3):55.e1-55.e6. doi:http://dx.doi.org/10.1016/j.forsciint.2008.08.009. 27.

Gill P, Jeffreys AJ, Werrett DJ. Forensic application of DNA ‘fingerprints.’

28.

RI PT

Nature. 1985;318:577. http://dx.doi.org/10.1038/318577a0.

Krishan K. Anthropometry in Forensic Medicine and Forensic Science’Forensic Anthropometry’. Internet J Forensic Sci. 2007;2(1):1-7.

29.

SC

doi:10.5580/1dce.

Gordon CC, Chumlea WC, Roche AF. Stature, recumbent length, and weight.

30.

M AN U

Anthr Stand Ref manual Champaign Hum Kinet Books. 1988:3-8. Kim HY. Statistical notes for clinical researchers: assessing normal distribution (2) using skewness and kurtosis. Restor Dent Endod. 2013;38(1):52. doi:10.5395/rde.2013.38.1.52.

West SG, Finch JF, Curran PJ. Structural equation models with nonnormal

TE D

31.

variables: Problems and remedies. 1995. 32.

Ulijaszek SJ, Kerr DA. Anthropometric measurement error and the assessment

EP

of nutritional status. Br J Nutr. 1999;82(3):165-177.

AC C

doi:10.1017/S0007114599001348. 33.

Krishan K, Kanchan T, Asha N, Kaur S, Chatterjee PM, Singh B. Estimation of sex from index and ring finger in a North Indian population. J Forensic Leg Med. 2013;20(5):471-479. doi:10.1016/j.jflm.2013.03.004.

34.

Roche AF, Sun SS. Human Growth: Assessment and Interpretation. Cambridge University Press; 2005.

35.

Sawilowsky SS. New effect size rules of thumb. J Mod Appl Stat Methods. 2009;8(2):597-599. http://digitalcommons.wayne.edu/coe_tbf/4.

19

ACCEPTED MANUSCRIPT 36.

Krishan K, Kanchan T, DiMaggio JA. A study of limb asymmetry and its effect on estimation of stature in forensic case work. Forensic Sci Int.

Legend for figures: Figure 1: The investigator measuring the stature of a subject

RI PT

2010;200(1-3):181. doi:10.1016/j.forsciint.2010.04.015.

SC

Figure 2: Length and breadth measurements taken on the human foot; Foot length

measurements as taken from pternion (pte.) to the most anterior part of the five

AC C

EP

TE D

M AN U

toes and foot breadth measurements at ball (FBB) and heel (FBH)

20

ACCEPTED MANUSCRIPT Acknowledgements: This study is a part of an on-going doctoral research being conducted by one of the authors (BS) in the Department of Anthropology, Panjab University, Chandigarh, India. KK is supported by a PURSE GRANT and the UGC Centre for

RI PT

Advanced Study (CAS) awarded to the Department of Anthropology, Panjab University, Chandigarh, India. BS is thankful to University Grants Commission for providing BSR fellowship for his PhD research. Thanks are also due to the subjects

AC C

EP

TE D

M AN U

SC

who voluntarily participated in the study.

ACCEPTED MANUSCRIPT Table1. Descriptive statistics of stature along with left and right foot dimensions (cm)

T4 T5 FBB FBH

Left Right Left Right Left Right Left Right Left Right Left Right Left Right

Max. 191.5 30.6 30.5 30.5 30.3 29.5 29.4 27.0 27.4 25.5 25.8 12.1 12.4 8.1 8.1

Mean 175.322 26.821 26.824 26.584 26.586 25.729 25.683 24.27 24.224 22.417 22.378 10.447 10.497 6.705 6.715

S.D. 6.060 1.330 1.303 1.328 1.322 1.293 1.286 1.261 1.273 1.169 1.175 0.588 0.573 0.394 0.420

RI PT

T3

Min. 158.8 22.4 22.7 22.5 22.6 22.0 21.8 21.0 20.4 18.8 18.7 8.7 8.9 5.3 5.4

SC

T2

Side

M AN U

Measurements Stature T1

T1- Foot length at first toe (pte- d1.t), T2- Foot length at second toe (pte- d2.t), T3- Foot length at third toe (pted3.t), T4- Foot length at fourth toe (pte- d4.t), T5- Foot length at fifth toe (pte- d5.t), FBB- Foot breadth at ball (mt.m- m.tl), FBH- Foot breadth at heel (cc.m- ctu.l).

T2

AC C

T3 T4 T5

FBB FBH

Side

Left Right Left Right Left Right Left Right Left Right Left Right Left Right

EP

Measurements Stature T1

TE D

Table 2. Test of normality by skewness and kurtosis of left and right feet and stature Skewness -0.132 -0.124 -0.109 -0.168 -0.128 -0.238 -0.176 -0.181 -0.156 -0.208 -0.257 -0.019 0.163 -0.076 .092

Kurtosis -0.112 0.141 0.014 -0.038 0.021 -0.024 0.030 -0.020 0.035 0.010 0.120 -0.155 -0.031 0.344 0.103

T1- Foot length at first toe (pte- d1.t), T2- Foot length at second toe (pte- d2.t), T3- Foot length at third toe (pted3.t), T4- Foot length at fourth toe (pte- d4.t), T5- Foot length at fifth toe (pte- d5.t), FBB- Foot breadth at ball (mt.m- m.tl), FBH- Foot breadth at heel (cc.m- ctu.l).

ACCEPTED MANUSCRIPT Table 3. Bilateral asymmetry in the foot dimensions t-values -0.161 -0.103 2.132 1.933 1.788 -3.466 -0.845

p- values 0.872 0.918 0.034 0.054 0.074 0.001 0.399

RI PT

Measurements T1L – T1R T2L – T2R T3L – T3R T4L – T4R T5L – T5R FBBL - FBBR FBHL - FBHR

M AN U

SC

L- left, R- right, T1- Foot length at first toe (pte- d1.t), T2- Foot length at second toe (pte- d2.t), T3- Foot length at third toe (pte- d3.t), T4- Foot length at fourth toe (pte- d4.t), T5- Foot length at fifth toe (pte- d5.t), FBB- Foot breadth at ball (mt.m- m.tl), FBH- Foot breadth at heel (cc.m- ctu.l).

AC C

EP

TE D

Table 4. TEM (Technical error of measurement), rTEM and R values of anthropometric variables. Variable Side TEM rTEM R Stature 0.542 0.320 0.988 Left T1 0.166 0.728 0.967 Right 0.191 0.826 0.964 Left T2 0.226 1.009 0.946 Right 0.231 1.023 0.952 Left T3 0.220 1.020 0.960 Right 0.213 0.992 0.962 Left T4 0.202 0.999 0.959 Right 0.199 0.988 0.962 Left T5 0.185 1.002 0.951 Right 0.161 0.874 0.973 Left FBB 0.104 1.422 0.958 Right 0.132 1.770 0.957 Left FBH 0.093 2.511 0.951 Right 0.079 2.132 0.962 TEM- technical error of measurements, rTEM- relative technical error of measurements, R- coefficient of reliability

ACCEPTED MANUSCRIPT Table 5. Karl Pearson’s correlation (r) of all the foot dimensions with stature

T2 T3 T4 T5 FBB FBH

r 0.607** 0.621** 0.630** 0.628** 0.617** 0.632** 0.582** 0.591** 0.594** 0.583** 0.364** 0.327** 0.309** 0.284**

RI PT

Side Left Right Left Right Left Right Left Right Left Right Left Right Left Right

SC

Measurements T1

M AN U

*. p< 0.05, **. p< 0.01, T1- Foot length at first toe (pte- d1.t), T2- Foot length at second toe (pte- d2.t), T3Foot length at third toe (pte- d3.t), T4- Foot length at fourth toe (pte- d4.t), T5- Foot length at fifth toe (pte- d5.t), FBB- Foot breadth at ball (mt.m- m.tl), FBH- Foot breadth at heel (cc.m- ctu.l).

Table 6. Stature estimation regression models when only one-foot measurement was taken at a time (linear regression models)

T3 T4

AC C

T5

FBB

FBH

Models 101.213 + 2.763 (T1-L)** 97.888 + 2.887 (T1-R)** 98.838 + 2.877 (T2-L)** 98.817 + 2.878 (T2-R)** 100.871 + 2.894 (T3-L)** 98.874 + 2.977 (T3-R)** 107.392 + 2.799 (T4-L)** 107.250 + 2.810 (T5-R)** 106.306 + 3.079 (T5-L)** 108.092 + 3.004 (T5-R)** 136.147 + 3.750 (FBB-L)** 139.050 + 3.456 (FBB-R)** 143.547 + 4.739 (FBH-L)** 147.802 + 4.098 (FBH-R)**

TE D

T2

Side Left Right Left Right Left Right Left Right Left Right Left Right Left Right

EP

Measurements T1

R2 0.368 0.385 0.397 0.394 0.381 0.399 0.339 0.349 0.353 0.339 0.132 0.107 0.095 0.081

S.E.E. 4.824 4.758 4.710 4.724 4.773 4.704 4.933 4.897 4.881 4.932 5.652 5.735 5.772 5.818

*. p< 0.05, **. p< 0.01, T1- Foot length at first toe (pte- d1.t), T2- Foot length at second toe (pte- d2.t), T3- Foot length at third toe (pte- d3.t), T4- Foot length at fourth toe (pte- d4.t), T5- Foot length at fifth toe (pte- d5.t), FBBFoot breadth at ball (mt.m- m.tl), FBH- Foot breadth at heel (cc.m- ctu.l).

ACCEPTED MANUSCRIPT Table 7. Stature estimation regression models when two-foot measurements were taken at a time (Multiple regression models)

T1, T5 T2, T3 T2, T4 T2, T5 T3, T4 T3, T5 T4, T5

R2 0.402 0.404 0.399 0.416 0.384 0.400 0.389 0.396 0.399 0.405 0.398 0.397 0.403 0.398 0.382 0.400 0.385 0.400 0.357 0.356

S.E.E. 4.698 4.693 4.712 4.642 4.767 4.708 4.749 4.721 4.710 4.686 4.716 4.720 4.696 4.713 4.775 4.706 4.764 4.707 4.871 4.876

RI PT

T1, T4

Models 96.987 + 0.794 (T1-L) + 2.145 (T2-L)** 95.811 + 1.257 (T1-R)** + 1.723 (T2-R)** 96.432 + 1.273 (T1-L)**+ 1.739 (T3-L)** 93.917 + 1.314 (T1-R)** + 1.797 (T3-R)** 98.052 + 1.828 (T1-L)** + 1.164 (T4-L)** 95.779 + 1.987 (T1-R)** + 1.084 (T4-R)** 97.934 + 1.671 (T1-L)** + 1.452 (T5-L)** 95.991 + 2.098 (T1-R)** + 1.030 (T5-R)** 98.253 + 2.222 (T2-L)** + 0.700 (T3-L) 97.260 + 1.249 (T2-R)* + 1.747 (T3-R)** 98.712 + 2.777 (T2-L)** + 0.115 (T4-L) 98.261 + 2.383 (T2-R)** + 0.566 (T4-R) 97.515 + 2.229 (T2-L)** + 0.828 (T5-L) 97.682 + 2.329 (T2-R)** + 0.703 (T5-R) 100.993 + 3.434 (T3-L)** - 0.578 (T4-L) 98.901 + 3.438 (T3-R)** - 0.490 (T4-R) 100.059 + 2.184 (T3-L)** + 0.851 (T5-L) 98.548 + 2.735 (T3-R)** + 0.291 (T5-R) 104.908 + 0.958 (T4-L) + 2.104 (T5-L)** 105.416 + 1.737 (T4-R)** + 1.244 (T5-R)*

SC

T1, T3

Side Left Right Left Right Left Right Left Right Left Right Left Right Left Right Left Right Left Right Left Right

M AN U

Variables T1, T2

AC C

EP

TE D

*. p< 0.05, **. p< 0.01, T1- Foot length at first toe (pte- d1.t), T2- Foot length at second toe (pte- d2.t), T3- Foot length at third toe (pte- d3.t), T4- Foot length at fourth toe (pte- d4.t), T5- Foot length at fifth toe (pte- d5.t), FBBFoot breadth at ball (mt.m- m.tl), FBH- Foot breadth at heel (cc.m- ctu.l).

ACCEPTED MANUSCRIPT Table 8. Stature estimation regression models when three-foot measurements were taken at a time (Multiple regression models)

T1, T2, T4

Left Right

T1, T2, T5

Left Right

T1, T3, T4

Left Right

T1, T3, T5

Left Right Left

TE D

T1, T4, T5

Right T2, T3, T4

Left

Left

AC C

T2, T3, T5

EP

Right

Right

T2, T4, T5

Left

Right

T3, T4, T5

Left Right

R2 0.404

S.E.E. 4.697

0.416

4.648

RI PT

Right

Models 96.350 + 0.806 (T1-L) + 1.452 (T2-L) + 0.728 (T3-L) 93.859 + 1.360 (T1-R)** - 0.117 (T2R) + 1.871 (T3-R)** 96.889 + 0.792 (T1-L) + 2.066 (T2L)** + 0.094 (T4-L) 95.312 + 1.243 (T1-R)** + 1.263 (T2R)* + 0.541 (T4-R) 96.220 + 0.646 (T1-L) + 1.737 (T2L)** + 0.696 (T5-L) 95.204 + 1.470 (T1-R)* + 1.403 (T2R)** + 0.538 (T5-R) 96.540 + 1.280 (T1-L)** +2.313 (T3L)** -0.621 (T4-L) 93.876 + 1.334 (T1-R)** 2.336 (T3R)** - 0.592 (T4-R) 96.256 + 1.196 (T1-L)** + 1.419 (T3L)** + 0.467 (T5-L) 93.591 + 1.331 (T1-R)** + 1.853 (T3R)** - 0.086 (T5-R) 97.700 + 1.626 (T1-L)** + 0.318 (T4L) + 1.174 (T5-L) 95.633 + 1.931 (T1-R)** + 0.858 (T4R) + 0.318 (T5-R) 98.378 + 2.198 (T2-L)** + 1.151 (T3L) -0.458 (T4-L) 97.307 + 1.230 (T2-R)* + 2.183 (T3R)** - 0.444 (T4-R) 97.504 + 2.200 (T2-L)** + 0.44 (T3L) + 0.812 (T5-L) 97.052 + 1.225 (T2-R)* + 1.595 (T3R)* + 0.212 (T5-R) 97.565 + 2.585 (T2-L)** – 0.996 (T45) + 1.482 (T5-L)** 97.690 + 2.288 (T2-R)** + 0.112 (T4R) + 0.629 (T5-R) 99.739 + 3.122 (T3-L) – 1.629 (T4-L) + 1.547 (T5-L) 98.091 + 3.317 (T3-R)** - 1.020 (T4R) + 0.749 (T5-R)

0.402

4.704

0.406

4.690

0.406

4.690

SC

Side Left

M AN U

Measurements T1, T2, T3

0.406

4.689

0.400

4.713

0.418

4.642

0.400

4.718

0.416

4.648

0.390

4.754

0.400

4.713

0.400

4.713

0.406

4.689

0.403

4.702

0.406

4.691

0.406

4.688

0.398

4.719

0.392

4.745

0.403

4.702

*. p< 0.05, **. p< 0.01, T1- Foot length at first toe (pte- d1.t), T2- Foot length at second toe (pte- d2.t), T3- Foot length at third toe (pte- d3.t), T4- Foot length at fourth toe (pte- d4.t), T5- Foot length at fifth toe (pte- d5.t), FBBFoot breadth at ball (mt.m- m.tl), FBH- Foot breadth at heel (cc.m- ctu.l).

ACCEPTED MANUSCRIPT Table 9. Stature estimation regression models when four-foot measurements were taken at a time (Multiple regression models)

T1, T2, T3, T5

Left Right

T1, T2, T4, T5

Left Right

T1, T3, T4, T5

Left Right

T2, T3, T4, T5

Left Right

R2 0.405

S.E.E. 4.699

0.418

4.648

RI PT

Right

Models 96.444 + 0.827 (T1-L) + 1.404 (T2-L) + 1.251 (T3-L) – 0.529 (T4-L) 93.839 + 1.409 (T1-R)** - 0.191 (T2-R) + 2.468 (T3-R)** - 0.604 (T4-R) 96.113 + 0.668 (T1-L) + 1.568 (T2-L)* + 0.232 (T3-L) + 0.609 (T5-L) 93.929 + 1.383 (T1-R)** - 0.129 (T2-R) + 1.939 (T3-R)** - 0.092 (T5-R) 96.472 + 0.542 (T1-L) + 2.124 (T2-L)** 0.861 (T4-L) + 1.283 (T5-L)* 95.155 + 1.180 (T1-R)* + 1.268 (T2-R)* + 0.299 (T4-R) + 0.337 (T5-R) 96.262 + 1.111 (T1-L)** + 2.249 (T3-L)** - 1.341 (T4-L) + 1.067 (T5-L) 95.753 + 1.286 (T1-R)** + 2.330 (T3R)** - 0.785 (T4-R) + 0.279 (T5-R) 97.329 + 2.108 (T2-L)** + 0.961 (T3-L) – 1.431 (T4-L) + 1.425 (T5-L)* 96.757 + 1.144 (T2-R) + 2.172 (T3-R)** 0.881 (T4-R) + 0.612 (T5-R)

SC

Side Left

M AN U

Measurements T1, T2, T3, T4

0.406

4.696

0.416

4.654

0.408

4.686

0.406

4.694

0.404

4.702

0.418

4.647

0.408

4.688

0.408

4.688

TE D

*. p< 0.05, **. p< 0.01, T1- Foot length at first toe (pte- d1.t), T2- Foot length at second toe (pte- d2.t), T3- Foot length at third toe (pte- d3.t), T4- Foot length at fourth toe (pte- d4.t), T5- Foot length at fifth toe (pte- d5.t), FBBFoot breadth at ball (mt.m- m.tl), FBH- Foot breadth at heel (cc.m- ctu.l).

AC C

EP

Table 10. Stature estimation regression models when all the foot measurements were taken at a time (Multiple regression models) Measurements Side Models R2 S.E.E. T1, T2, T3, T4, Left 96.120 + 0.587 (T1-L) + 1.559 (T2-L)* + 0.410 4.685 T5 1.062 (T3-L) – 1.330 (T4-L) + 1.203 (T5-L) Right 93.721 + 1.357 (T1-R)** - 0.177 (T2-R) 0.418 4.653 +2.453 (T3-R)** - 0.794 (T4-R) +0.274 (T5-R) T1, T2, T3, T4, Left 96.658 + 0.602 (T1-L) + 1.567 (T2-L)* + 0.410 4.696 T5, FBB, FBH 1.069 (T3-L) – 1.320 (T4-L) + 1.220 (T5-L) – 2.37 (FBB-L) + 0.082 (FBH-L) Right 96.750 + 1.426 (T1-R)** - 0.047 (T2-R) 0.422 4.652 + 2.414 (T3-R)** - 0.788 (T4-R) + 0.331 (T5-R) – 0.523 (FBB-R) -0.492 (FBH-R) *. p< 0.05, **. p< 0.01, T1- Foot length at first toe (pte- d1.t), T2- Foot length at second toe (pte- d2.t), T3- Foot length at third toe (pte- d3.t), T4- Foot length at fourth toe (pte- d4.t), T5- Foot length at fifth toe (pte- d5.t), FBBFoot breadth at ball (mt.m- m.tl), FBH- Foot breadth at heel (cc.m- ctu.l).

ACCEPTED MANUSCRIPT

Table 11. Comparison of the present study (Jatt Sikhs) with previously published study (Gujjars) on North Indians.

T3 T4 T5 FBB FBH

Cohen’s d (SD) -0.39 -0.31 -0.54 -0.35 -0.51 -0.65 -0.60 -0.81 -0.96 -0.46 -0.60 -0.85 -0.67 -0.92 -0.63

RI PT

T2

Left Right Left Right Left Right Left Right Left Right Left Right Left Right

Present Study (2018) North Indian Jatt Sikhs Mean (cm) SD (cm) 175.32 6.06 26.82 1.33 26.82 1.30 26.58 1.32 26.58 1.32 25.72 1.29 25.68 1.28 24.27 1.26 24.22 1.27 22/41 1.16 22.37 1.17 10.44 0.58 10.49 0.57 6.70 0.39 6.71 0.42

SC

Stature T1

Krishan (2008) North Indian Gujjars Mean (cm) SD (cm) 172.68 6.84 25.91 3.26 25.28 3.19 25.57 3.25 25.14 3.19 23.93 3.12 24.05 3.06 22.18 2.89 21.89 2.73 21.43 2.38 21.08 2.39 8.91 2.06 9.26 2.10 5.52 1.48 5.89 1.49

M AN U

Measurements Side

AC C

EP

TE D

T1- Foot length at first toe (pte- d1.t), T2- Foot length at second toe (pte- d2.t), T3- Foot length at third toe (pted3.t), T4- Foot length at fourth toe (pte- d4.t), T5- Foot length at fifth toe (pte- d5.t), FBB- Foot breadth at ball (mt.m- m.tl), FBH- Foot breadth at heel (cc.m- ctu.l).

ACCEPTED MANUSCRIPT Table 12. Comparison of present study (Jatt Sikhs) with previously published national and international studies Age (yr.)

Sample Size

Mean Stature (cm.)

Cohen’s d (S.D.)

(Krishan and Sharma, 2007)1, N. Indian Rajputs (Uhrová et al., 2015)2 Slovak (Ahmed, 2013)3 Sudanese Arab (Nor et al., 2013)4, Malaysian (Hisham et al., 2012)5, Malaysian Chinese (Kim et al., 2018)6, Korean (Zeybek et al., 2008)7 Turk (Agnihotri et al., 2007)8, Indian (Present Study, 2018), Jatt Sikh

17-20

123

168.24

-1.14

18-24

120

179.50

25-30

80

174.71

20-90

69

164.84

20-64

107

166.07

-1.53

20-59

2750

170.69

-0.72

18-30 18-30

-1.67

159.0

-2.73

125

173.99

0.21

388

175.32

0

TE D EP

-0.10

136

L- left, R- right, Yr. - Year, S.D. – Standard deviation, cm. – Centimeter

AC C

0.67

SC

M AN U

18-44

Mean Foot Length (cm.) R – 24.72 L – 24.70 R – 26.26 L – 26.25 R- ---L – 26.43 R – ---L – 24.01 R – 22.20 L – 22.29 R – 25.12 L – ---R – 23.3 L – 23.3 R – 26.12 L – 26.09 R – 26.82 L – 26.82

RI PT

Study & Population

Cohen’s d (S.D.)

-1.61 -1. 65 -0.42 -0.90 ----0.30 ----2.06 -3.72 -3.74 -1.48 ----2.76 -2.81 -0.54 -0.58 0 0

ACCEPTED MANUSCRIPT Table 13: Comparison of coefficient of correlation (r) and standard error of estimation (S.E.E) in different male population studies.

0.71 0.39

0.71 0.41 0.60 0.29

0.703 0.423 0.412

0.697 0.401 0.395

EP

5.83 6.96

5.065 6.453 6.488

SC

FL FBB FBH FL FBB T1 T2 T3 T4 T5 T1 T2 T3 T4 T5 FBB FBH T1 T2 T3 T4 T5 FBB FBH

Comb.

Standard error of estimation (S.E.E) Left Right Comb . 4.55 4.56 5.98 5.91

RI PT

Right

FL FBB

AC C

Present Study, (2018) 18-30 Years

FL FBB

Left

5.105 6.624 6.543

0.77 0.44 0.82 0.76 0.76 0.74 0.76

0.86 0.85 0.81 0.81 0.79 0.71 0.55 0.60 0.63 0.61 0.58 0.59 0.36 0.30

0.86 0.84 0.84 0.83 0.80 0.72 0.52 0.62 0.62 0.63 0.59 0.58 0.32 0.28

TE D

Uhrova et al.2 (2015) 18-24 Years Nor et al.4 (2013) 20-90 years Hemy et al.9 (2013) 19-68 Years Ahmed3 (2013) 25-30 Years Krishan et al. 10 (2012) 18 years

Krishan 11 (2008) 18-30 Years

Correlation (r)

Parameters

M AN U

Study (Age group)

2.16 2.24 2.28 2.21 2.25 2.93 3.57 4.82 4.71 4.77 4.93 4.88 5.65 5.77

3.66 5.15 3.908 4.443 4.461 4.646 4.460 2.15 2.27 2.26 2.20 2.24 2.96 3.61 4.75 4.22 4.70 4.89 4.93 5.73 5.81

L- left, R- right, T1- Foot length at first toe (pte- d1.t), T2- Foot length at second toe (pte- d2.t), T3- Foot length at third toe (pte- d3.t), T4- Foot length at fourth toe (pte- d4.t), T5- Foot length at fifth toe (pte- d5.t), FBB- Foot length at ball (mt.m- m.tl), FBH- Foot length at heel (cc.m- ctu.l), Comb. – Combined.

ACCEPTED MANUSCRIPT

Table 14. Comparison of accuracies acquired by different set of foot length combinations in the present study No of foot Highest estimation accuracy lengths 2 S.E.E taken at a Side Comb. R time One L T2 0.397 4.710

Lowest estimation accuracy

Two

R

T1, T3

0.416

Three

R

Four

R

Five

R

T1, T3, 0.418 T4 T1, T3, 0.418 T4, T5 T1, T2, 0.418 T3, T4, T5

Comb.

L

T4

4.642

R

4.642

L

4.647

L

6.653

L

R2

S.E.E

RI PT

Side

4.933

T4, T5

0.336

4.876

T1, T4, T5 T1, T3, T4, T5 T1, T2, T3, T4, T5

0.390

4.754

M AN U

SC

0.339

0.404

4.702

0.410

6.685

AC C

EP

TE D

L- left, R- right, T1- Foot length at first toe (pte- d1.t), T2- Foot length at second toe (pte- d2.t), T3- Foot length at third toe (pte- d3.t), T4- Foot length at fourth toe (pte- d4.t), T5- Foot length at fifth toe (pte- d5.t), FBB- Foot length at ball (mt.m- m.tl), FBH- Foot length at heel (cc.m- ctu.l).

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

ACCEPTED MANUSCRIPT Highlights: 1. Possibility of stature estimation from different foot dimensions is explored. 2. Linear and multiple regression models were derived for stature estimation from foot

RI PT

dimensions. 3. Multiple regression models provide better estimates of stature than the linear regression models.

4. Right foot measurements show greater accuracy in stature estimation than the left foot

SC

measurements.

introduced in the regression.

M AN U

5. Strength of correlation tends to enhance when more than one associative factor is

6. Inclusion of any factor having weak correlation with stature in the regression model,

AC C

EP

TE D

decreases the accuracy of the model.

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

The authors declare that there is no conflict of interest arising out of this manuscript.