Forensic podiatry

(1)

i

FORENSIC PODIATRY

Investigation of the variability of bare footprints over time and the

variability of measurements taken by different practitioners

(2)

ii

Forensic podiatry

Investigation of the variability of bare footprints over time and the variability of measurements taken by different practitioners

Student: Lesley Amanda Heuser E-mail: laheuser@student.avans.nl Phone number: +44 7926596036 Period: 01/02/2012 – 01/07/2013 Version: 2 Word count:

Supervisor Staffordshire University: Dr Claire Gwinnett E-mail: c.gwinnett@staffs.ac.uk

Supervisor forensic podiatry: Prof Wesley Vernon OBE Wesley.vernon@nhs.net

Supervisor Avans: Lute-Harm Zwiers E-mail: lh.zwiers@avans.nl

Location: Staffordshire University

Department of Forensic Science and Crime Science Leek Road, Science Centre

ST4 2DF , Stoke-on-Trent, United Kingdom

Sheffield podiatry centre 722 Prince of Wales Road Sheffield

S9 4EU

Avans University

School of life sciences and environment technologies Postbus 90.116

(3)

iii

Acknowledgement

First and foremost, I would like to thank my supervisor Dr Claire Gwinnett, senior lecturer in forensic science at Staffordshire University, for the valuable guidance and advice. Her willingness to motivate me contributed tremendously to this project. Besides, I would like to thank Dr Laura Walton, senior lecturer in forensic science at Staffordshire University, for her additional guidance where necessary. I would like to express my special gratitude to Professor Wesley Vernon, Jeremy Walker and Sarah Reel for giving me the opportunity to participate in this important project and for provide me with valuable information as the guidance of this project. I also would like to thank Staffordshire University for providing me with a good environment and facilities to complete this project. I am grateful to Miss Charlotte-Maria Orphanou for her valuable time to help me improve my research and report writing skills. I am also grateful to Paul Bailey, technician in forensic science at

Staffordshire University, for taking care of all the equipment I needed. Finally, yet importantly, I also would like to thank all those who were prepared to participate in this study. Without help of the particular that mentioned above, I would face many difficulties while performing this project.

(4)

iv

Abstract

Although the ‘inkless’ shoe print kit is an approved forensic methodology for the collection of bare footprints, no previous research has considered whether bare footprints collected with the inkless shoe print kit are reliable for re-examination after a certain amount of time. In order to reach the aim of this study, multiple research questions were developed to be examined. In order to answer these research questions, multiple experiments were carried out. Bare footprints from the previous study treated paper versus thermal fax paper (appendix VIII) were used for re-examination by a second time scanning of the original bare footprints with a time range from 11-14 weeks. As a

complemented study, the reliability of measuring bare footprints by the same participant and different participants on the same samples has been assessed.

Bare footprints were used from the previous treated paper versus thermal fax paper study. The images were scanned for the second time by the use of an Epson perfection 3490 PHOTO scanner (greyscale, 150 dpi). The images were stored as a Tiff image on a Packerd bell Easynote LV laptop with a backup on a verbatim external hard drive. The software GNU Image Manipulation Program (GIMP 2) was used to measure the bare footprints to establish the lengths of each toe from the furthest point of the heel, the width of de ball of the foot and the width of the heel of the foot. IBM SPSS Statistics 19 for windows was used to apply the statistical tests to the data.

The examination of dynamic changes in bare footprints during time showed a variety of increase and decrease of the bare footprint samples. Bare footprints scanned after 11 weeks overall showed less dynamic changes over time. Bare footprints scanned after 14 weeks showed high dynamic changes over time. The investigation of the reliability of measuring bare footprints suggest a high intra-rater reliability for bare footprints measurement by the same participant with an intraclass correlation coefficient (ICC) of at least 0,98 for all the measurements. Measurements taken by different participants using the same samples suggest a negative value for the ICC and Cronbach’s Alpha which can be due to negatively correlated data.

(5)

v

Content

List of tables ... 1 List of figures ... 2 1. Introduction ... 3 1.1. Aim... 3 1.2. Objectives ... 4 1.3. Report structure ... 5 2. Theoretical background ... 6 2.1. Forensic podiatry ... 6

2.1.1. Bare footprints in a forensic context ... 6

2.1.2. Forensic podiatry and the role in forensic research ... 7

2.2. Statistical analysis of bare footprints ... 8

2.2.1. Descriptive statistics ... 8

2.2.2. Analysis of the normal distribution ... 8

2.2.3. Parametric and non-parametric tests ... 10

2.2.4. Reliability analysis... 11

3. Methodology ... 12

3.1. Research design ... 12

3.2. Collection of bare footprints ... 12

3.3. Dynamic changes of bare footprints over time ... 13

3.3.1. Measuring bare footprints using GNU Image Manipulation Program ... 13

3.3.2. Statistical analysis: reliability of measuring bare footprints ... 13

3.4. Reliability of measuring bare footprints ... 14

3.4.1. Measuring bare footprints using GNU Image Manipulation Program ... 14

3.4.2. Statistical analysis: intra-rater reliability test ... 14

4. Results and discussion ... 15

4.1. Dynamic changes of bare footprints over time ... 15

4.1.1. Static bare footprints on treated paper ... 16

4.1.2. Dynamic bare footprints on treated paper ... 20

4.1.3. Static bare footprints on thermal fax paper ... 24

4.1.4. Dynamic bare footprints on thermal fax paper ... 28

4.2. Reliability of measuring bare footprints ... 32

4.2.1. Measurements taken by one practitioner... 32

(6)

vi 5. Conclusion ... 36 6. Recommendation ... 37 7. References ... 38 8. Glossary ... 40 Appendices ... I Appendix I: Measurements ... I Appendix II: Representative bare footprints ... XIV Appendix III: Ethical approval ... XV Appendix IV: PRA form approved ... XVI Appendix V: Consent form ... XIX Appendix VI: Information sheet ... XXI Appendix VII: Material Safety Data Sheet ... XXIII Appendix VIII: Abstract previous study ... XXIV Appendix IX: Guide for statistical analysis in SPSS IBM 19 ... XXV

(7)

1

List of tables

Section Page Table 1: Descriptive statistics and Kolmogorov-Smirnov test static bare footprints on

treated paper over time (n=150) 16

Table 2: Paired sample t-test static bare footprints on treated paper over time (n=150) 18 Table 3: Descriptive statistics and Kolmogorov-Smirnov test dynamic bare footprints on

treated paper over time (n=150) 20

Table 4: Paired sample t-test dynamic bare footprints on treated paper over time (n=150) 22 Table 5: Descriptive statistics and Kolmogorov-Smirnov test static bare footprints on thermal

fax paper over time (n=150) 24

Table 6: Paired sample t-test static bare footprints on thermal fax paper over time (n=150) 26 Table 7: Descriptive statistics and Kolmogorov-Smirnov test dynamic bare footprints on

thermal fax paper over time (n=150) 28

Table 8: Paired sample t-test dynamic bare footprints on thermal fax paper over time (n=150) 30 Table 9: Descriptive statistics and Kolmogorov-Smirnov test for the measurements repeatedly

performed by the same person (n=5) 32

Table 10: Intra-rater reliability test including the intraclass correlation coefficient ICC and

Cronbach's alpha as a coefficient for reliability and consistency 33

Table 11: Descriptive statistics and Kolmogorov-Smirnov test for measurements performed by different persons using the same samples (n=1) 34

Table 12: intra-rater reliability test for measurements performed by different persons using

(8)

2

List of figures

Section Page

Fig. 1: Method to collect static bare footprints 7

Fig. 2: Method to collect dynamic bare footprints 7

Fig. 3: The normal distribution curve 9

Fig. 4: Negatively skewed and positively skewed curve 9

Fig. 5: How to choose the right statistical test 9

Fig. 6: Mean differences for static bare footprints on treated paper plot into a histogram (n=150) 17

Fig. 7: p-value for static bare footprints on treated paper plot into a graph (n=150) 18

Fig. 8: Mean differences for dynamic bare footprints on treated paper plot into a histogram (n=150) 21

Fig. 9: p-value for dynamic bare footprints on treated paper plot into a graph (n=150) 22

Fig. 10: Mean differences for static bare footprints on thermal fax paper plot into a histogram (n=150) 25

Fig.11: p-value for static bare footprints on thermal fax paper plot into a graph (n=150) 26

Fig.12: Mean differences for dynamic bare footprints on thermal fax paper plot into a histogram (n=150) 29

(9)

3

1. Introduction

When terms have been formatted in bold, refer to the glossary (Chapter 8, page 41) According to DiMaggio et al. in 2011(1) ‘In the context of forensic practice, bare footprint

examinations are usually conducted to determine whether a person’s feet have made bare footprints and/or impressions associated with a crime scene’. Current techniques in forensic podiatry are able to compare unknown bare footprints from a crime scene with known bare footprints from a suspect or a victim(2). The comparison of the shape and outlines of bare footprints has been examined in multiple criminal investigations(3). A previous research project proposed by Reel et al. in 2010(4) established a reliable two-dimensional measuring technique that could be used for the comparison of bare footprints in criminal investigations and forensic podiatry related research. However, new problems such as time constraints, intelligence probing and relevance of the results to the case have meant that further technology is needed. A recent study conducted by the author has examined whether thermal fax paper could be a reliable substitution for the treated paper (appendix VIII) to be considered as a cost effective improvement for the collection of bare footprints.

With the progression received in the previous years, forensic podiatry will become a full profession and discipline in forensic science. However, more research is necessary to establish bare footprint examination. As an addition to the previous study of the comparison between treated paper and thermal fax paper, a time study on the same samples has been carried out. No previous research has considered whether bare footprints collected with the inkless shoe print kit are reliable for re-examination after a certain amount of time. The two-dimensional measuring technique has shown to be reliable for measuring bare footprints. No previous research has considered whether the measuring method for bare footprints could be personal sensitive or reliable to be performed by any person.

1.1. Aim

This report is a part of a study developed by Professor Wesley Vernon, Jeremy Walker and Sarah Reel from Sheffield podiatry centre and is supervised by Dr Claire Gwinnett from Staffordshire University. This research project is divided by three different aims:

1. to investigate the reliability of bare footprints over time, by a comparison of the original measurements with the measurements obtained after a second time scanning.

2. Assess the reliability of the measuring method for bare footprints with the measurements taken by one practitioner.

3. Assess the reliability of the measuring method for bare footprints with the measurements taken by several practitioners.

(10)

4

1.2. Objectives

In order to reach the aim of this research, different objectives were addressed.

1. Bare footprints from the previous study (appendix VIII) were used for re-examination. A second time scanning was performed with a time range of 11-14 weeks.

2. An intra-rater reliability study was conducted by the analysis of bare footprint measurements taken by one practitioner using Cronbach’s Alpha and the interclass correlation coefficient. 3. An intra-rater reliability study was conducted by the analysis of bare footprint measurements

taken by several practitioners using Cronbach’s Alpha and the interclass correlation coefficient.

Research questions and hypotheses Research question 1:

Is there a significant difference in the size of the bare footprint samples from the previous thermal fax paper versus treated paper study (appendix VIII) compared to the original samples after a second time scanning, with a time range of 11- 14 weeks.

Null hypothesis states:

There will be no significant difference in the size of the bare footprint samples from the previous thermal fax paper versus treated paper study (appendix VIII) compared to the original samples after a second time scanning, with a time range of 11- 14 weeks. The hardcopy bare footprint samples will not fade away when stored after the samples have been collected.

Alternative hypothesis states:

There will be a significant difference in the size of the bare footprint samples from the previous thermal fax paper versus treated paper study (appendix VIII) compared to the original samples after a second time scanning, with a time range of 11- 14 weeks. The hardcopy bare footprint samples will slowly fade away when stored after the samples have been collected.

Research question 2:

Are the measured bare footprints taken by the same practitioner reliable so no significant difference between the measurement can be observed?

The measurements taken by the same practitioner will show no significant difference (p>0,05) after performing an intra-rater reliability test using Cronbach’s Alpha and the interclass correlation coefficient. The measuring method will be reliable to be used repeatedly by the same practitioner. Alternative hypothesis states:

The measurements taken by the same practitioner will show significant difference (p<0,05) after performing an intra-rater reliability test using Cronbach’s Alpha and the interclass correlation coefficient. The measuring method will not be reliable to be used repeatedly by the same practitioner.

(11)

5 Research question 3:

Is the measuring method personal sensitive or can this method be performed by any practitioner without showing significant difference between the same measurements obtained by several practitioner.

The measuring method is not personal sensitive and can be performed by any participant without showing significant difference between the same measurements obtained by several practitioners, after performing an intra-rater reliability test using Cronbach’s Alpha and the interclass correlation coefficient.

Alternative hypothesis states:

The measuring method is personal sensitive and cannot be performed by any participant without showing significant difference between the same measurements obtained by several practitioners, after performing an intra-rater reliability test using Cronbach’s Alpha and the interclass correlation coefficient.

1.3. Report structure

The background study for this report is presented in chapter two, followed by the methodology in chapter three. The results substantiate by the discussion will be described in chapter four. Chapter five describes the conclusion followed by the recommendation presented in chapter six. Chapter seven contains the references used to justify this report. This report also contains appendices which includes the forms and information sheets used in this project. As an addition to the this study, the abstract from the previous study was added to the appendices.

(12)

6

2. Theoretical background

2.1. Forensic podiatry

Disorders, medical and surgical treatment of the foot is a separate professionalism in the world of medicines. The specialism of a podiatrist includes human anatomy, physiology, pathophysiology, surgery, sociological and psychological perspectives. The first school of podiatry was opened in 1911 in the United States of America. In 1912 podiatry was established in the United Kingdom. At that time podiatry sought full professional recognition with a specialist knowledge base for many years. Professional groups had both a theoretical and a practical knowledge basis. The knowledge available for podiatrists was not only a scientific basis, but also pre-scientific. After podiatry became a

recognized profession, podiatrists specialize themselves in forensic podiatry. Contemporary forensic podiatry is currently a discipline of forensic science. The additional value of podiatric involvement in bare footprint examination is related to podiatrist appreciation of foot function and it is possible effect on the footprint, the ability to use podiatric survey data and extensive clinical experience in relation to the prevalence of foot type, conditions and pathologies that may be apparent from bare footprint examination. The main purpose of a forensic podiatrist is human identification by the comparison of unknown bare footprints from a crime scene with known bare footprints(1).

2.1.1. Bare footprints in a forensic context

It is not usual for a forensic podiatrist to recover evidence from a crime scene. This part is being the duty of the Crime Scene Officer, Crime Scene Investigator or the Scene of Crime Officer1. Bare footprints collected from a crime scene indicates the unknown prints. These prints are often compared with known prints which been collected from a suspect or a victim. Bare footprint examination is a simple method with a great visual impact(5). Bare footprint impressions are impressions made by walking or standing on bare feet. Bare footprints made without any type of motion are suggest as static bare footprints (fig. 1 ). When bare footprints made by walking or any type of motion , the samples are suggest as dynamic bare footprints (fig. 2). It is important for forensic podiatrists to collected both types of bare footprints. A previous research determined statistically significant difference between measurements of static and dynamic bare footprints (t=23.17, p<0.01)(4). These types of bare footprints are more often found at homicide or crime scene were sexual assault is being involved(6).

(13)

7

The contour and shape of a bare footprint impressions is related to different variables. Variables as height, weight and ethnic origin can show different types of bare footprint impressions as expected. A previous study from Krishan et al. in 2008(8) concluded that body weight has an influence on a bare footprint impressions. A bare footprint impression not only relate directly to the feet of a person but can also match with an insole footprint impression in a person’s shoe. The Meredith Kercher case in 2007 showed that forensic podiatry is very important in forensic science. A bare footprint in blood was found at the crime scene which matched with the insole footprint impression in the shoe of the main suspect(9).

2.1.2. Forensic podiatry and the role in forensic research

Throughout the previous years, the role of forensic podiatry in forensic research became more important. As bare footprints are the main evidence for forensic podiatrist to examine, previous research improved the analysis of bare footprints. The use of bare foot impressions for forensic examination was proposed by Kennedy et al. in 2003(3). This study examined the comparison of the shapes of bare footprints from an individual with footprints from a crime scene to include or exclude an individual at the scene of crime. The study confirmed the normality of the set of measurements from footprint outlines. The methodology for measuring the outlines of a bare footprints was proposed by Norman Gunn, which is called the GUNN method. This particular method was re-examined by Reel et al. in 2010(4). The study established a reliable two-dimensional measuring technique that could be used for the comparison of bare footprints in further research.

Fig. 2: Method to collect static bare footprints(7)

Fig. 1: Method to collect dynamic bare footprints(7)

(14)

8

2.2. Statistical analysis of bare footprints

Statistics plays an important role in forensic science. A forensic scientist must be able to determine the magnitude and significance of the variability in forensic work. The statistical analysis of data is based on a set of repeated measurements on a single object. The statistical analysis of bare

footprints was developed by Kennedy et al. in 2003(3). There are several types of statistics, the main parts are the descriptive statistics and the inferential statistics. Descriptive statistics and inferential statistics are disjunctive. Descriptive statistics includes procedures to describe a set of data. Inferential statistics is concerned with making predictions or inferences about a set of data(10).

2.2.1. Descriptive statistics

The mean and standard deviation are the two main parameters to describe the variability in a set of data. The mean describes the location of the data set along the axis of possible values. The standard deviation describes the range of values around the location of the data. The mean is an unbiased estimate of the population of samples mean (µ), which is calculated in the same way but including all possible measurements, n. As the sample size n increases the mean will get closer to (µ) in

magnitude. The mean is the most useful and widely applicable assessment. The median and the quartile points are also providing different approaches to an average. The median value is found by ranking all the measurements in order of magnitude and identifying the value halfway the order. The quartile points can be split up in the lower quartile and the upper quartile. The lower quartile will be the value one-quarter of the way up the rank list. The upper quartile will be the value three-quarters of the way along. The interquartile range gives the range of values around the mean. The quartile points together give a measure of the statistical distribution of the data around the median. An alternative approach to provide the description of the mean is to identify the range of data for example the minimum and maximum values in the sample or population. It is essential to calculate the value of the mean first before calculate the standard deviation when using this formula. The symbol s can be substitute by the symbol σ for the standard deviation of populations. For the comparison of the different techniques for the analysis of the same subject, it is useful to evaluate the ratio of the standard deviation to the mean for a particular set of measurements. The relative standard deviation is a measure of the relative precision expressed as a percentage(10). To choose the right statistical test, the mean of the samples must be plot to create the sampling distribution of the mean.

2.2.2. Analysis of the normal distribution

The Kolmogorov-Smirnov test makes the prediction of the distribution of the datasets. The form of the distribution has been made up out of three characteristics; the number of modes, the degree of symmetry and the kurtosis. The number of modes is indicative of the normality of the distribution. The degree of symmetry of the distribution has three categories; normal distribution as seen in fig.3, positively skewed and negatively skewed as seen in fig.4.

(15)

9

The kurtosis represent the frequency of a histogram. Normal distributed data shows a normal curve (Fig. 3). A normal curve has certain characteristics; a normal curve is symmetrical, a normal curve is unimodal, this means there is only one peak shown in the curve; the normal curve has merits

discussion, the area under the curve is always 100% of all value in the distribution(11). The purpose of an statistical test is to test the hypothesis. There is always a possibility to reach the wrong

conclusion. Rejecting the null hypothesis when it is true, is called a type I error. A type II error is the opposite of type I error. When the null hypothesis is false and failed to be rejected, it is called a type II error (12). There are two concepts which should be taken into account when the statistical test are performed and why these tests should be preferred over other tests (Fig. 5).

Fig. 5: How to choose the right statistical test

(16)

10 2.2.3. Parametric and non-parametric tests

The assumption of the independent sample test is that the data is at interval level. The sample t-test is based on a calculation of one sample compared to a population(11). The independent sample t-test is used to compare the mean on continuous variables for two different groups of participants. The paired sample t-test is based on the same elements as the independent sample t-test. The difference between an independent sample t-test and a paired sample t-test is the comparison of the samples. The independent sample t-test is based on a calculation of one sample compared to a population. The paired sample t-test use two samples – one sample as the subject of the research and one sample that will represent the population. In the case of this pilot study , the samples on fax paper are the subject of this research which will make the samples on treated paper the control group of this project. The paired t-test is used to test the hypothesis of two sets of measurements where a direct pairing exists between individual values in each set. In other words , the paired sample t-test is used to compare the mean scores for the same group of participants on two different occasions or when the data includes matched pairs. The paired t-test is based on the empirical rule. If the distribution is approximately normally distributed, this means 68% of the values in the dataset falls within ±1 standard deviation of the mean. With the empirical rule it is possible to locate the relative position of a data point in terms of the standard deviation . The sample mean, the population mean and the standard error mean are used to computing value of t(13) _{. When using the paired}

t-test on non-normal distributed data the risk of error increase. If the value of the data is normally distributed, the mean of the samples will also be normally distributed.

A significant level is used to examine whether the data has a significant difference or has not a significant difference. Scientist assume that a probability of 5% is small enough to be a useful cut-off point. This means that the statistical test must give a p-value of 0,05. Every number under the 0,05 means there is a significant difference. Every number above 0,05 means there is no significant difference. The p-value is a conditional probability of the null hypothesis being true(14).

The effect size is a sample-based estimate of the quantity of the datasets. The effect size will give an indication if the difference between the groups is statistically significant. The calculation of an effect sizes will tell the magnitude of intervention’s effect. The calculation of the effect size is descriptive and quantifies the extent of the difference between two groups(13).

When the Kolmogorov-Smirnov test concludes the data is not normal distributed, a non-parametrical test used to examine the significant difference between two or more data sets. Non- parametric tests use the median of the data sets where parametric tests use the mean of the datasets. Non-parametric analysis make fewer assumptions but the applicability of a non-Non-parametric test is much wider than the of parametric tests. On the other hand, where a parametric test is more suitable to use, non-parametric tests have less power to analyse the data(15).

(17)

11 2.2.4. Reliability analysis

The reliability of a methodology or measurements indicates how free it is from random error. There are two indicators to examine the reliability. When the same subject will be analysed on two different occasions with the correlation between the two values obtained then test-retest is being assessed(13). Previous articles from Bland and Altman(16-19) published a graphical method to observe if two different datasets agree sufficiently with each other. According to Bland et al. in 1997(17), the Cronbach’s Alpha is a useful coefficient for assessing the reliability. Cronbach’s Alpha measures the internal consistency and independency of datasets(20). Previous studies discussed reliability research by the use of intraclass correlation coefficient (Chia-Cheng et al. in 2008(21),2013(22)). According to Reel et al. in 2010(4), the ICC calculated from a one-way ANOVA can reflects the amount of consistency and agreement between raters.

(18)

12

3. Methodology

3.1. Research design

Ethical approval (appendix III) and a PRA (appendix IV) form were signed by Staffordshire University before the study has been performed.

This research was divided in three different studies. All the studies were an addition on the previous treated paper versus thermal fax paper study (appendix VIII) but can be considered as independent studies. The overall experimental design strategy was to seek the null hypothesis is being true. The main part of this study was to assess the reliability of bare footprint during time. Therefore the original bare footprints were scanned for the second time. Measurements were taken from bare footprints samples scanned for the second time after 11 weeks, 12 weeks, 13 weeks and 14 weeks. To substantiate the results of the time study, two different studies were carried out. One dataset was taken to be measured repeatedly by the same participant on different occasions to observe if the measured bare footprints repeatedly performed by the same participant, on different times of the day during three weeks, reliable so no significant difference between the measurement could be observed. The same dataset was used again and measured by several practitioners to examine if measuring method is not personal sensitive and could be performed by any person.

3.2. Collection of bare footprints

For the collection of bare footprints in the previous study (appendix VIII) AN inkless shoe print kit (CSI Equipment LTD) was used for both types of paper. For the static prints the participant was asked to stand with the left foot en right foot on both sides of the inkless pad. The participant was asked to raise their right foot and place it onto the inkless pad whilst a piece of treated paper of fax paper where placed on the original position of the participants right foot. The participant was asked to place their right foot onto the paper to develop a static print. This method was repeated five times on treated paper and five times on fax paper. After the collection of static bare footprints, dynamic bare footprints has been taken. The participant was asked to stand with their right foot between the lines and started walking with their right foot. Whilst the participant was practicing the inkless pad and the paper were placed on the right position between the other two lines . After practicing the participant was asked to move themselves between the lines where the inkless pad and paper were positioned to repeated the same method as mentioned above. The method was repeated five times on treated paper (n=5) and five times on fax paper (n=5)(6). 20 bare footprints where collected from each participant. 30x20=600 bare footprints in total.

(19)

13

3.3. Dynamic changes of bare footprints over time

All samples were stored in a paper evidence bag and stored in an aluminium drawer at Staffordshire University, room 345 (appendix II for representative examples of bare footprint samples). The bare footprints were scanned on a Epson perfection 3490 PHOTO scanner and were stored in Tiff. Format on a Packard bell Easynote LV laptop with a backup on an Verbatim 53035 1TB external hard drive.

Keynote measurement codes: *TLT1: length heel-toe 1 on treated paper. ^TLT2: length heel-toe 2 on treated paper. #TLT3: length heel-toe

3 on treated paper. †TLT4: length heel-toe 4 on treated paper. ¶TLT5: length heel-toe 5 on treated paper. ‡TWB: width ball on treated

paper. ˟_{TWH: width heel on treated paper. FLT1: length heel-toe 1 on thermal fax paper.}^_{FLT2: length heel-toe 2 on thermal fax paper.}

#_{FLT3: length heel-toe 3 on thermal fax paper.}†_{FLT4: length heel-toe 4 on thermal fax paper.}¶_{FLT5: length heel-toe 5 on treated paper.}

‡_{FWB: width ball on treated paper.}˟_{FWH: width heel on treated paper.}

3.3.1. Measuring bare footprints using GNU Image Manipulation Program The bare footprints were measured with GNU Image Manipulation Program (appendix I for

measurements). The prints were opened in GNU Image Manipulation Program. The default settings were changed from px to mm and from 33% to 50%. The paintbrush was changed to size 3.00 and the painting colour was changed from black to red. The layer boundary size was changed over half the position of the original width and height of the print. The X axis was offset until the image moved to the right side. The Y axis was offset until the image moved downwards. The image was resized and fitted to canvas layer. At a viewing of 50%, a line was drawn skimming the medial edge of the foot. The drawing process was repeated for the lateral side of the foot, intersecting the medial line below the heel. At a 200% viewing the widest part of the heel was selected and en line was drawn from one side of the heel to the other side of the heel. The number of millimetres was divided by 2 and a line was drawn backwards until the line intersected the cross of the lateral and medial length lines at the bottom on fig. 10. The image was rotated until it aligns vertically. At a view of 200% the lowest pixels of the heel were marked with the grid as a guide. The highest pixel at the tip of the big toe was marked. A line was drawn from the central point of the heel to the tip of the big toe. This process was repeated for the other toes(6). The lines were measured and saved in a excel document for statistical analysis.

3.3.2. Statistical analysis: reliability of measuring bare footprints

The measurements of the bare footprints were analysed with different statistical tests. IBM Statistical Package for the Social Science (SPSS) version 19 was used to perform the Kolmogorov-Smirnov test (appendix IX) for testing the distribution of the datasets. A paired sample t-test (appendix IX) was performed to observe if the samples scanned after 11,12,13 and 14 weeks significantly increase or decrease over time. The mean differences as a result of the performed Kolmogorov-Smirnov test were plot into a histogram with Microsoft excel 2010.

(20)

14

3.4. Reliability of measuring bare footprints

This study was a repeated-measure design to examine the intra-rater reliability of measuring bare footprints differed in two different pilot studies. The first pilot study assessed the bare footprints repeatedly performed by the same practitioner, using the same samples on different occasions. The second pilot study examined the bare footprints measured by several practitioners using the same samples. The samples were made up of one participant out of the previous treated versus thermal fax paper study. The same measuring method was performed in both studies.

3.4.1. Measuring bare footprints using GNU Image Manipulation Program

The bare footprints were measured with GNU Image Manipulation Program. The prints were opened in GNU Image Manipulation Program. The default settings were changed from px to mm and from 33% to 50%. The paintbrush was changed to size 3.00 and the painting colour was changed from black to red. The layer boundary size was changed over half the position of the original width and height of the print. The X axis was offset until the image moved to the right side. The Y axis was offset until the image moved downwards. The image was resized and fitted to canvas layer. At a viewing of 50%, a line was drawn skimming the medial edge of the foot. The drawing process was repeated for the lateral side of the foot, intersecting the medial line below the heel. At a 200% viewing the widest part of the heel was selected and en line was drawn from one side of the heel to the other side of the heel. The number of millimetres was divided by 2 and a line was drawn backwards until the line intersected the cross of the lateral and medial length lines at the bottom on fig. 10. The image was rotated until it aligns vertically. At a view of 200% the lowest pixels of the heel were marked with the grid as a guide. The highest pixel at the tip of the big toe was marked. A line was drawn from the central point of the heel to the tip of the big toe. This process was repeated for the other toes(6). Both studies were examined using an intra-rater reliability test.

3.4.2. Statistical analysis: intra-rater reliability test

The measurements of the bare footprints were analysed with different statistical tests. IBM Statistical Package for the Social Science (SPSS) version 19 was used to perform the Kolmogorov-Smirnov test for testing the distribution of the datasets. An intra-rater reliability test has been performed to assess the reliability of the measurements performed by the same practitioner.

(21)

15

4. Results and discussion

The results of these studies are displayed in this chapter and the significant will be discussed. All the results were obtained from the 30 different participants (n=600) from the previous study, the comparison between treated paper and thermal fax paper (appendix VIII). Five repeats of each participant were collected. Five static and dynamic bare footprints on treated paper and five static and dynamic bare footprints on thermal fax paper. The measurements were examined by the use of Statistical Package for the Social Science (SPSS IBM 19).

4.1. Dynamic changes of bare footprints over time

This paragraph presents the results of the statistical analysis from the bare footprints over a time period on treated paper and thermal fax paper. The measurements obtained from the original scanned bare footprint samples on treated paper (n=150) where compared with the measurements obtained from the same bare footprint samples on treated paper after the second time scanning (n=150) (appendix I). The measurement obtained from the original scanned bare footprint samples on thermal paper (n=150) where compared with the measurements obtained from the same bare footprint samples on thermal fax paper after the second time scanning (n=150) (appendix I). All samples were stored under equal conditions. According to Archer et al. in 2005(23), as it is difficult to obtain homogeneous samples from humans, an absolute statement about changes in time is difficult to make and needs to be considered in the conclusion. However, records on paper are exposed to chemical reactions. Increase or decrease of records on paper can be due to activation such as acid hydrolysis and oxidation of cellulose(24).

(22)

16 4.1.1. Static bare footprints on treated paper

The static bare footprints on treated paper were examined by the performance of the Kolmogorov-Smirnov test for normality. According to Curran et al. in 2006(25) ,a paired sample t-test was

performed to observe whether there is a significant difference in the mean scores between the first time of scanning the static bare footprints and the second time of scanning the static bare footprints.

Table 1: Descriptive statistics and Kolmogorov-Smirnov test static bare footprints on treated paper over time (n=150)

Samples 11 weeks Samples 12 weeks Samples 13 weeks Samples 14 weeks

Measurement (mm)

Mean SD Distribution Mean SD Distribution Mean SD Distribution Mean SD Distribution

*_TLT1[1] _226,18 _13,28 _Normal _241,48 _8,40 _Normal _232,02 _19,22 _Normal _236,53 _13,73 _Normal

TLT1[2] 226,24 13,27 Normal 242,39 8,33 Normal 232,67 19,34 Normal 237,28 13,73 Normal ^_TLT2[1] _229,31 _13,45 _Normal _220,29 _70,95 _Normal _233,38 _17,11 _Normal _236,11 _12,10 _Normal

TLT2[2] 229,35 13,41 Normal 221,22 71,15 Normal 234,02 17,32 Normal 236,58 12,06 Normal #_TLT3[1] _220,29 _13,81 _Normal _232,07 _9,05 _Normal _225,24 _16,32 _Normal _227,96 _11,79 _Normal

TLT3[2] 220,34 13,78 Normal 232,94 8,15 Normal 225,93 16,52 Normal 228,59 11,72 Normal

†_TLT4[1] 206,20 12,17 Normal 220,64 10,08 Normal 213,02 13,35 Normal 216,17 11,32 Normal

¶_TLT5[1] 190,65 10,01 Normal 187,71 53,65 Normal 137,34 91,62 Normal 154,59 84,18 Normal

‡_TWB[1] 89,19 6,50 Normal 94,48 4,71 Normal 90,59 8,00 Normal 90,65 8,11 Normal

TWB[2] 90,18 6,06 Normal 94,67 5,08 Normal 91,87 8,15 Normal 92,88 7,20 Normal

˟_TWH[1] 45,94 4,13 Normal 48,66 3,86 Normal 46,50 7,38 Normal 48,35 4,93 Normal

TWH[2] 45,75 4,12 Normal 49,11 3,04 Normal 46,26 7,49 Normal 48,96 4,91 Normal

Table 1: *TLT1: length heel-toe 1 on treated paper. ^TLT2: length heel-toe 2 on treated paper. #TLT3: length heel-toe 3 on treated paper.

†_{TLT4: length heel-toe 4 on treated paper.}¶_{TLT5: length heel-toe 5 on treated paper.} ‡_{TWB: width ball on treated paper.}˟_{TWH: width heel}

on treated paper

Table 1 provides the results of the Kolmogorov-Smirnov test to confirm the distribution of the static bare footprints on treated paper. According to Reel et al. in 2010(4), when the Kolmogorov-Smirnov test confirm normality for the datasets, a parametric test is suitable to be performed. The static bare footprints on treated paper used for re-examination in this study were not significantly different from normal. These results confirm the null hypothesis for normality so a parametric test was suitable to be performed on the datasets (n=150). The descriptive statistics are also represented in table 1. Small variation in the differences of the mean measurements from the samples with the same code divided by [1] and [2] has been observed. The standard deviation of all dynamic bare footprint samples on treated paper considered a low standard deviation compared to the mean. The datasets are more reliable as the data points reclined closely around the mean. According to Biau in2011(26) ,when the datasets are not significant from normal, 95% of the samples falls within 1.96

(23)

17

standard deviation. The static bare footprint datasets treated paper appear to be not significant from normal. 95% of the sample population in this study falls within 1.96 standard deviation(26).

The difference between the mean scores of the samples with the same code divided by [1] and [2] were taken to plot into a histogram in order to assess the variation between the mean scores of the samples.

Fig. 6: Mean differences for static bare footprints on treated paper plot into a histogram (n=150)

The variations in the differences of the mean measurements from the static bare footprint

measurements on treated paper with a time range from 11-14 weeks are presented in fig. 6. The x-axis references to the time in weeks and the y-x-axis to the differences in the mean measurements of the datasets. According to the results, the datasets show a high variety in mean measurements between the samples. A trend is being observed with an increase in size of all the samples with an exception of the samples with the code TWB. The samples with code TWB shows a variety of increase and decrease over a period of time. According to the mean measurements of the samples, the samples with the code TWB were found as the less reliable samples within the static bare footprints on treated paper for re-examined after 14 weeks.

-0,50 0,00 0,50 1,00 1,50 2,00 2,50 11 12 13 14 M e an d iff e re n ce ( m m ) Weeks

Mean measurements for static bare

footprints

TLT1 TLT2 TLT3 TLT4 TLT5 TWH TWB

(24)

18

Table 2: Paired sample t-test static bare footprints on treated paper over time (n=150)

Samples 11 weeks Samples 12 weeks Samples 13 weeks Samples 14 weeks Measurement (mm) Mean t-value Sig. Mean t-value

Sig. Mean t-value Sig. Mean t-value Sig.

*_{TLT1 [1]-TLT1[2]} _-0,06 _-1,371 _0,186 _-0,91 _-2,753 _0,008 _-0,65 -10,398 0,000 -0,75 -6,625 0,000 ^_{TLT2 [1]- TLT2[2]} _-0,04 _-0,715 _0,483 _-0,92 _-2,728 _0,009 _-0,64 _-8,984 _0,000 _-0,48 _-5,900 _0,000 #_{TLT3 [1]- TLT3[2]} _-0,06 _-1,238 _0,231 _-0,87 _-2,885 _0,006 _-0,69 _-9,596 _0,000 _-0,63 _-6,557 _0,000 †_{TLT4 [1]- TLT4[2]} 0,03 0,584 0,566 -0,83 -2,916 0,005 -0,65 -12,152 0,000 -0,66 -8,473 0,000 ¶_{TLT5 [1]- TLT5[2]} 0,01 0,098 0,923 -0,88 -2,914 0,005 -0,46 -5,415 0,000 -0,68 -6,314 0,000 ‡_{TWB [1]-TWB[2]} -1,00 -2,461 0,024 -0,20 -0,895 0,375 -1,28 -4,056 0,000 -2,23 -9,241 0,000 ˟_{TWH [1]-TWH[2]} 0,20 1,418 0,172 -0,45 -1,666 0,101 0,24 2,233 0,033 -0,61 -3,525 0,001

Table 2: *TLT1: length heel-toe 1 on treated paper. ^TLT2: length heel-toe 2 on treated paper. #TLT3: length heel-toe 3 on treated paper.

†_{TLT4: length heel-toe 4 on treated paper.}¶_{TLT5: length heel-toe 5 on treated paper.} ‡_{TWB: width ball on treated paper.}˟_{TWH: width heel}

on treated paper.

Table 2 presents the results of the paired sample t-test for static bare footprint on treated paper. As shown in the table, the samples scanned after 11 weeks showed significant difference compared to the original scanned samples. Significant difference is being observed in one case (14% of the samples). Samples scanned after 12 and 13 weeks showed that significant difference appears when time increased. In the results of the samples scanned after 14 weeks, significant difference in all cases (100% of the samples) is being observed. The p-value of the paired sample t-test were plot into a graph so a possible trend could be observed.

Fig. 7: p-value for static bare footprints on treated paper plot into a graph (n=150)

0,000 0,100 0,200 0,300 0,400 0,500 0,600 0,700 0,800 0,900 1,000 11 12 13 14 p -v al u e p ai re d sam p le t -te st Weeks

P-value for static bare footprints

(25)

19

Fig. 7 presents the p-values of the paired sample t-test plot into a graph with a time range from 11-14 weeks for static bare footprints on treated paper. The x-axis references to the time in weeks and the y-axis to the p-value of the datasets. Fig 7 shows a high variety whether significant difference is being observed. A trend is being observed in the samples scanned after 14 weeks. All samples showed significant difference after 14 weeks. There can be considered that static bare footprints on treated paper show significant difference after 14 weeks.

(26)

20 4.1.2. Dynamic bare footprints on treated paper

The dynamic bare footprints on treated paper were examined by the performance of the

Kolmogorov-Smirnov test for normality. A paired sample t-test was performed to observe whether there is a statistically significant difference in the mean scores between the first time of scanning the dynamic bare footprints and the second time of scanning the dynamic bare footprints(25).

Table 3: Descriptive statistics and Kolmogorov-Smirnov test dynamic bare footprints on treated paper over time (n=150)

*_TLT1[1] _245,06 _12,45 _Normal _255,64 _9,26 _Normal _246,47 _16,54 _Normal _250,89 _16,75 _Normal

TLT1[2] 244,94 12,71 Normal 255,82 8,95 Normal 247,04 16,71 Normal 253,39 14,72 Normal ^_TLT2[1] ₂₄₃ _11,85 _Normal _230,75 _74,40 _Normal _242,99 _16,11 _Normal _247,69 _12,67 _Normal

TLT2[2] 242,88 12,15 Normal 231,06 74,42 Normal 243,27 16,32 Normal 249,07 11,94 Normal #_TLT3[1] _232,56 _13,25 _Normal _241,91 _9,93 _Normal _233,94 _15,54 _Normal _238,67 _12,79 _Normal

†_TLT4[1] 217,27 10,88 Normal 229,70 10,96 Normal 221,05 12,32 Normal 225,98 12,25 Normal

¶_TLT5[1] 200,98 8,58 Normal 183,96 71,45 Normal 155,62 87,67 Normal 208,96 10,94 Normal

‡_TWB[1] 88,75 5,79 Normal 93,76 3,85 Normal 90,94 8,65 Normal 91,86 8,48 Normal

TWB[2] 89,12 5,86 Normal 93,76 3,63 Normal 90,22 7,95 Normal 92,42 8,07 Normal

˟_TWH[1] 49,50 2,05 Normal 49,98 2,96 Normal 48,57 8,75 Normal 50,96 5,98 Normal

TWH[2] 49,09 2,35 Normal 49,55 3,62 Normal 48,25 8,72 Normal 51,58 5,16 Normal

Table 3: *TLT1: length heel-toe 1 on treated paper. ^TLT2: length heel-toe 2 on treated paper. #TLT3: length heel-toe 3 on treated paper. †TLT4: length heel-toe 4 on treated paper. ¶TLT5: length heel-toe 5 on treated paper. ‡TWB: width ball on treated paper. ˟TWH: width heel on treated paper

Table 3 provides the results of the descriptive statistics and the Kolmogorov-Smirnov test to confirm the distributions from the dynamic bare footprints on treated paper. The original data and the data used for re-examination were not significantly different from normal. These results confirm the null hypothesis of normality so a parametrical test was suitable to be performed on the datasets

(n=150)4). The descriptive statistics are also represented in table 3 . Small variation in the differences of the mean measurements from the samples with the same code divided in [1] and [2] has been observed. The standard deviation of all dynamic bare footprint samples on treated paper considered a low standard deviation compared to the mean. The datasets are more reliable as the data points recline closely around the mean. The static bare footprint datasets treated paper appear to be not significant from normal. 95% of the sample population in this study falls within 1.96 standard deviation(26).

(27)

21

The difference between the mean of the samples with the same code divided in [1] and [2] were taken to plot into a histogram in order to assess the variation in the means.

Fig. 8: Mean differences for dynamic bare footprints on treated paper plot into a histogram (n=150)

Fig. 8 presents the variations in the differences of the mean measurements from the dynamic bare footprints measurements on treated paper with a time range from 11 weeks till 14 weeks. The x-axis references to the time in weeks and the y-axis to the differences in the mean measurements of the datasets. According to the results, the datasets show a high variety in mean difference between the samples. No trend is being observed within the results. However, the samples with the code TLT5 (Length heel-toe 5) can be considered to show the highest dynamic changes within dynamic bare footprint samples on treated paper according to the mean difference of the samples.

-4,00 -2,00 0,00 2,00 4,00 6,00 11 12 13 14 M e an d iff e re n ce ( m m ) Weeks

Mean measurements for dynamic

bare footprints

(28)

22

Table 4: Paired sample t-test dynamic bare footprints on treated paper over time (n=150)

Samples 11 weeks Samples 12 weeks Samples 13 weeks Samples 14 weeks Measurement (mm) Mean t-value Sig. Mean t-value Sig. Mean t-value

Sig. Mean t-value Sig.

*_{TLT1 [1]-TLT1[2]} _0,12 _1,047 _0,308 _-0,19 _-0,565 _0,575 _-0,57 _-4,695 _0,000 _-2,5 _-5,325 _0,000 ^_{TLT2 [1]- TLT2[2]} _0,11 _0,832 _0,416 _-0,31 _-1,122 _0,267 _-0,28 _-3,268 _0,003 _-1,38 _-5,827 _0,000 #_{TLT3 [1]- TLT3[2]} _0,08 _0,644 _0,528 _-0,30 _-1,589 _0,118 _-0,16 _-3,134 _0,004 _-0,84 _-4,186 _0,000 †_{TLT4 [1]- TLT4[2]} 0,07 0,729 0,475 -0,13 -0,770 0,445 -0,12 -2,701 0,011 -0,67 -4,564 0,000 ¶_{TLT5 [1]- TLT5[2]} 0,04 0,354 0,727 3,61 0,956 0,343 -0,02 -0,620 0,540 -0,66 -3,551 0,001 ‡_{TWB [1]-TWB[2]} -0,37 -1,151 0,146 0,00 0,000 1,000 0,72 2,036 0,051 -0,56 -1,960 0,056 ˟_{TWH [1]-TWH[2]} 0,41 1,751 0,096 0,42 2,254 0,029 0,32 1,721 0,096 -0,62 -2,967 0,005

Table 4: *TLT1: length heel-toe 1 on treated paper. ^TLT2: length heel-toe 2 on treated paper. #TLT3: length heel-toe 3 on treated paper. †

TLT4: length heel-toe 4 on treated paper. ¶TLT5: length heel-toe 5 on treated paper. ‡TWB: width ball on treated paper. ˟TWH: width heel

on treated paper.

Table 4 presents the results of the paired sample t-test for dynamic bare footprint on treated paper. As shown in the table, the samples scanned after 11 suggest significant difference in one case (14% of the samples). The results of the samples scanned after 14 weeks suggest significant difference in six cases (86% of the samples). The p-value of the paired sample t-test were plot into a graph so a possible trend could be observed.

Fig. 9: p-value for dynamic bare footprints on treated paper plot into a graph (n=150)

Fig. 9 presents the p-value of the paired sample t-test plot into a graph with a time range from 11 weeks till 14 weeks for dynamic bare footprints on treated paper. The x-axis references to the time in weeks and the y-axis to the p-value of the datasets. Fig 9 shows a high variety within the samples of significant difference between the original samples and the samples scanned for the second time. No

0 0,2 0,4 0,6 0,8 1 1,2 11 12 13 14 p -v al u e p ai re d sam p le t -te st Weeks

P-value for dynamic bare footprints

(29)

23

trend is being observed within the samples on the different occasions. However, there can be considered that the samples with the code TWB (width of the ball from the foot) are the most reliable samples as the samples show no dynamic changes after 14 weeks.

(30)

24

4.1.3. Static bare footprints on thermal fax paper

The static bare footprints on fax paper were examined by the performance of the Kolmogorov-Smirnov test for normality. A paired sample t-test was performed to observe whether there is a statistically significant difference in the mean measurements between the first time of scanning the static bare footprints and the second time of scanning the static bare footprints(25).

Table 5: Descriptive statistics and Kolmogorov-Smirnov test static bare footprints on thermal fax paper over time (n=150)

*_FLT1[1] _226,66 _13,28 _Normal _241,36 _7,30 _Normal _232,67 _18,52 _Normal _237,18 _14,45 _Normal

FLT1[2] 226,78 13,28 Normal 241,83 7,34 Normal 234,21 18,67 Normal 237,97 14,43 Normal ^_FLT2[1] _229,88 _13,54 _Normal _219,93 _70,72 _Normal _234,25 _17,05 _Normal _236,21 _12,73 _Normal

FLT2[2] 229,96 13,43 Normal 220,15 70,77 Normal 235,24 17,25 Normal 237,01 12,69 Normal #_FLT3[1] _220,92 _14,10 _Normal _227,32 _32,26 _Normal _226,08 _16,28 _Normal _228,57 _12,69 _Normal

FLT3[2] 221,04 14,00 Normal 231,92 8,06 Normal 227,03 16,54 Normal 229,31 12,73 Normal

†_FLT4[1] 206,48 12,43 Normal 220,36 9,65 Normal 213,80 13,01 Normal 216,68 11,81 Normal

¶_FLT5[1] 191,21 9,81 Normal 172,28 72,07 Normal 130,61 94,02 Normal 150,85 87,35 Normal

‡_FWB[1] 88,14 5,69 Normal 94,08 4,33 Normal 89,84 7,60 Normal 90,64 8,05 Normal

FWB[2] 89,30 5,76 Normal 93,99 4,36 Normal 91,05 7,98 Normal 92,59 7,39 Normal

˟_FWH[1] 46,52 3,44 Normal 49,08 3,50 Normal 45,04 11,43 Normal 48,25 5,00 Normal

FWH[2] 46,60 3,39 Normal 48,96 2,94 Normal 46,58 7,75 Normal 48,67 5,22 Normal

Table 5: *FLT1: length heel-toe 1 on thermal fax paper. ^FLT2: length heel-toe 2 on thermal fax paper. #FLT3: length heel-toe 3 on thermal

fax paper. †FLT4: length heel-toe 4 on thermal fax paper. ¶FLT5: length heel-toe 5 on treated paper. ‡FWB: width ball on treated paper.

˟_{FWH: width heel on treated paper}

Table 5 provides the results of the descriptive statistics and the Kolmogorov-Smirnov test to confirm the distributions from the static bare footprints on thermal fax paper. The original data and the data used for re-examination were not significantly different from normal. These results confirm the null hypothesis of normality so a parametrical test was suitable to be performed on the datasets

(n=150)(4). The standard deviation of all dynamic bare footprint samples on treated paper considered a low standard deviation compared to the mean. The datasets are more reliable as the data points recline closely around the mean. The static bare footprint datasets treated paper appear to be not significant from normal. 95% of the sample population in this study falls within 1.96 standard deviation(26).

(31)

25

Fig. 10: Mean differences for static bare footprints on thermal fax paper plot into a histogram (n=150)

Fig. 10 presents the variations in the differences of the means measurements from the static bare footprints measurements on thermal fax paper with a time range from 11-14 weeks. The x-axis references to the time in weeks and the y-axis to the differences in the mean scores of the datasets. According to the results, the datasets show an increase in mean difference between the samples. A trend is being observed within the samples, as almost all the samples appear to be increased over time. -1,00 0,00 1,00 2,00 3,00 4,00 5,00 6,00 11 12 13 14 M e an d iff e re n ce ( m m ) Weeks

Mean measurements for static bare

footprints

(32)

26

Table 6: Paired sample t-test static bare footprints on thermal fax paper over time (n=150)

Samples 11 weeks Samples 12 weeks Samples 13 weeks Samples 14 weeks Measurement (mm) Mean t-value Sig. Mean t-value Sig. Mean t-value

Sig. Mean t-value Sig.

*_{FLT1 [1]-FLT1[2]} _-0,12 _-2,091 _0,050 _-0,47 _-3,469 _0,001 _-1,54 _-2,366 _0,025 _-0,79 _-3,687 _0,001 ^_{FLT2 [1]- FLT2[2]} _-0,09 _-1,034 _0,314 _-0,22 _-2,919 _0,005 _-0,99 _-3,204 _0,003 _-0,80 _-3,741 _0,001 #_{FLT3 [1]- FLT3[2]} _-0,13 _-1,073 _0,297 _-4,60 _-1,090 _0,281 _-0,95 _-3,395 _0,002 _-0,75 _-3,951 _0,000 †_{FLT4 [1]- FLT4[2]} -0,10 -1,188 0,249 -0,27 -3,118 0,003 -1,04 -3,366 0,002 -0,80 -3,976 0,000 ¶_{FLT5 [1]- FLT5[2]} -0,05 -0,322 0,751 -0,36 -3,129 0,003 -0,56 -2,416 0,022 -4,82 -1,153 0,255 ‡_{FWB [1]-FWB[2]} -1,16 -4,012 0,001 0,09 0,428 0,670 -1,20 -5,775 0,000 -1,95 -6,532 0,000 ˟_{FWH [1]-FWH[2]} -0,09 -0,640 0,530 0,12 0,515 0,608 -1,54 -1,060 0,298 -0,42 -1,988 0,053

Table 6 presents the results of the paired sample t-test of static bare footprint on thermal fax paper. As shown in the table, the samples scanned after 11 suggest significant difference compared to the original scanned samples in one case (14% of the samples). The results of the samples scanned after 14 weeks suggest no significant difference in two cases (28% of the samples). The static bare

footprints on thermal fax paper can be considered as not reliable for re-examination after 14 weeks. The p-value of the paired sample t-test were plot into a graph so a possible trend could be observed.

Fig. 11: p-value for static bare footprints on thermal fax paper plot into a graph (n=150)

0,000 0,100 0,200 0,300 0,400 0,500 0,600 0,700 0,800 11 12 13 14 p -v al u e p ai re d sam p le t -te st Weeks

P-value for static bare footprints

(33)

27

Fig. 11 presents the p-value of the paired sample t-test plot into a graph with a time range from 11 weeks till 14 weeks for dynamic bare footprints on treated paper. The x-axis references to the time in weeks and the y-axis to the p-value of the datasets. Fig 11 shows a high variety within the samples of significant difference between the original samples and the samples scanned for the second time. No trend is being observed within the samples on the different occasions. However, there can be considered that the samples with the code FLT5 and FWH are the most reliable samples as the samples showed no significant difference after 14 weeks.

(34)

28

4.1.4. Dynamic bare footprints on thermal fax paper

The dynamic bare footprints on thermal fax paper were examined by the performance of the Kolmogorov-Smirnov test for normality. A paired sample t-test was performed to observe whether there is a statistically significant difference in the mean scores between the first time of scanning the dynamic bare footprints and the second time of scanning the dynamic bare footprint25.

Table 7: Descriptive statistics and Kolmogorov-Smirnov test dynamic bare footprints on thermal fax paper over time (n=150)

*_FLT1[1] _243,81 _12,54 _Normal _259,12 _7,54 _Normal _246,65 _17,09 _Normal _251,40 _17,17 _Normal

FLT1[2] 243,34 12,49 Normal 259,32 7,57 Normal 247,92 17,04 Normal 254,45 16,27 Normal ^_FLT2[1] _241,6 _12,03 _Normal _232,77 _74,81 _Normal _242,48 _15,95 _Normal _247,95 _13,27 _Normal

FLT2[2] 241,48 12,13 Normal 233,39 74,99 Normal 243,47 16,08 Normal 249,83 13,11 Normal #_FLT3[1] _231,58 _13,63 _Normal _243,71 _7,94 _Normal _233,69 _15,68 _Normal _239,14 _13,08 _Normal

†_FLT4[1] 216,60 11,72 Normal 231,04 9,96 Normal 221,17 12,73 Normal 226,62 12,34 Normal

¶_FLT5[1] 200,93 9,30 Normal 197,14 56,29 Normal 168,89 77,37 Normal 210,05 11,05 Normal

‡_FWB[1] 89,09 5,96 Normal 94,32 3,50 Normal 90,41 8,15 Normal 91,87 9,00 Normal

FWB[2] 88,80 6,57 Normal 95,68 4,41 Normal 90,69 8,17 Normal 92,36 8,19 Normal

˟_FWH[1] 48,61 2,69 Normal 50,08 2,91 Normal 48,84 9,23 Normal 51,16 6,05 Normal

FWH[2] 48,02 2,29 Normal 50,20 2,90 Normal 48,89 9,04 Normal 51,68 5,23 Normal

Table 7 provides the results of the descriptive statistics and the Kolmogorov-Smirnov test to confirm the distributions from the dynamic bare footprints on thermal paper. The original data and the data used for re-examination were are not significantly different from normal. These results confirm the null hypothesis of normality so a parametrical test was suitable to be performed on the datasets (n=150)(4). The descriptive statistics are also represented in table . Small variation in the differences of the means of the samples with the same code divided in [1] and [2] has been observed. The standard deviation of all dynamic bare footprint samples on thermal fax paper samples considered a low standard deviation compared to the mean. The datasets are more reliable as the data points recline closely around the mean. The static bare footprint datasets treated paper appear to be not significant from normal. 95% of the sample population in this study falls within 1.96 standard deviation(26).

(35)

29

Fig. 12: Mean differences for dynamic bare footprints on thermal fax paper plot into a histogram (n=150)

The variations in the differences of the means of the dynamic bare footprints measurements are represented into a histogram with a time range of 11 weeks till 14 weeks. The x-axis references to the time in weeks and the y-axis to the differences in the mean scores of the datasets. According to the results, the datasets show a high variety in mean difference between the samples. No trend is being observed within the results. However, the samples with the code TLT5 (Length heel-toe 5) show a great decrease in size after 13 weeks. These samples can be considered as the less reliable dynamic bare footprint samples on thermal fax paper.

-20 -15 -10 -5 0 5 10 11 12 13 14 M e an d iff e re n ce ( m m ) Weeks