DF User perception of upper interior space in the rear seat of a car A human centric approach to assess head roominess Master’s thesis in Product Development KARTHIK GUNASEKARAN CORNELIA WONG NYLANDER Department of Industrial and Material Science CHALMERS UNIVERSITY OF TECHNOLOGY Gothenburg, Sweden 2020 Master’s thesis 2020 User perception of upper interior space in the rear seat of a car A human centric approach to assess head roominess KARTHIK GUNASEKARAN CORNELIA WONG NYLANDER DF Department of Industrial and Material Science Division Design and Human Factors Chalmers University of Technology Gothenburg, Sweden 2020 User perception of upper interior space in the rear seat of a car A human centric approach to assess head roominess KARTHIK GUNASEKARAN CORNELIA WONG NYLANDER © KARTHIK GUNASEKARAN & CORNELIA WONG NYLANDER, 2020. Supervisor: Sara Alpsten and Pernilla Nurbo, Volvo cars, Gothenburg Examiner: Lars-Ola Bligård, Department of Industrial and Material Science, Chalmers University of Technology Master’s Thesis 2020 Department of Industrial and Material Science Division Design and Human Factors Chalmers University of Technology SE-412 96 Gothenburg Telephone +46 31 772 1000 Cover: Visualization of the acceptance ranges for the position of Cant rail, Belt line and C-pillar Printed by Chalmers Digitaltryck Gothenburg, Sweden 2020 iv Abstract The upper interior space in a car is an important aspect to consider in the design process for creating overall riding comfort. During the development of sufficient roominess a population’s anthropometrics together with ergonomic dimensions and measuring standards are utilized. The interest underlying this thesis study is that head roominess evaluation is considered as a physically measured entity, when it is substantially a combination of physical measurement (objective) and perceived roominess which is subjective to the user. Investigative studies were done at Volvo Cars and recommendations were generated based on the results from the study. The master thesis study started with literature study followed by three research phases which included qualitative and quantitative studies; Awareness, Identification and Investigation phase. The initial phase (Awareness) was focused in obtaining wider knowledge, and in the final phase (Investigation) involving specific investigation to attain in-depth knowledge. During the second phase (Identification), a qualitative study was applied, and six parameters of the category Vision and Spatial distances were identified as critical for the perceived roominess. The parameters were analyzed in the Investigation phase, the third and last phase, using a quantitative user study with 34 test persons, divided into two groups based on their sitting height; Short and Tall. The rear-outer seat was studied (behind the driver seat on the left side) in a XC90 car. The rear seat and the driver seat were set in design position (SgRP), the car had light interior trim with a covered panorama roof and the test was performed with the Swedish population. In total thirteen manipulations were randomly performed and a Visual Analogue Scale (VAS) was utilized to assess test person’s perceived upper interior space for every manipulation. Nine manipulations were intended for the Vision parameters; Cant rail height, Belt line height and C-pillar forward/rearward position, one change in the Spatial distance (vertical height) and four VAS rating were obtained when the car was in a nominal state. This data together with additionally retained angles and eye points were then statistically analyzed to find trends, tendencies and correlations for all the studied parameters. For the Vision parameters, the change in Belt line had the highest (negative) influence on the VAS rating irrespective of the Short and Tall group. A high correlation was established between VAS rating and retained angle for the Cant rail. Moreover, C-pillar had the same influence on the Short and Tall group and a common borderline was acknowledge. However, for the Spatial distances, the Tall group had a stronger correlation with these distances than the Short group, who did not establish any correlation with the VAS rating. As recommendations, the acceptable angles for all the parameters have been translated into a definite designed colour range guidance of what would be acceptable, critical or poor for the Vision parameters when perceiving the interior space. Keywords: perceived head roominess, upper interior space, car, anthropometirc, VAS, vision parameters, spatial distances, statistical analysis v Acknowledgements This study was performed as a master’s thesis at Chalmers University of Technology under the department of Industrial and Material Science for the master’s programme of Product Development. The scope of the thesis work was 30 credits with a work plan of 20 weeks. The study was done in collaboration with the Ergonomics team of Volvo Cars, Gothenburg. We like to take this opportunity to thank Sara Alpsten and Pernilla Nurbo from the Ergonomics team for setting up the thesis work in this interesting area of head roominess. It was their guidance and continuous support in supervision that lead the way to successfully complete the study. With gratitude, we thank the people who were involved in the user studies that were conducted for the thesis work. To mention, experts from the Product Validation team and experts from Ergonomics team who were involved in the earlier phases of the study. It was with their expertise and feedback that enabled to design the later phases of the study. We also like to thank all the test participants from various departments of Volvo Cars who have participated in user study in later phases amidst the uncertainty caused by Covid19. A special mention of thanks to Mikael Fransson from Complete Vehicle Engineering Quality team at Volvo Cars, who helped us by providing initial knowledge to progress with statistical analysis of subjective data. This thesis could not have been possible without the academic supervision from our supervisor and examiner Lars-Ola Bligård. We thank him for his approachable nature helping us out whenever we were held back in a doubt, constantly entrusting in our skills and abilities. Finally, we would like to thank our family and friends for constant support and encouragement throughout this time. Karthik Gunasekaran and Cornelia Wong Nylander, Gothenburg, 2020 vii Nomenclature This chapter contains the definition and explanation to symbols, terms and abbreviations used in the thesis. This page can be referred if the abbreviations and symbols are not explained in the particular section or context. %-ile – Percentile, A value (0-100%) that divides a group so that one part of the data falls below that value and the other part falls above it. CAD- Computer aided design Catia V5- Computer Aided Three-dimensional Interactive Application ERQ- Exploratory research question EX – Position of eye in the X-coordinates EZ – Position of eye in the Y-coordinates H - Hypothesis, a supposition or proposed explanation made on the basis of limited evidence as a starting point for further investigation. NRS - Numerical Rating Scale P-value - Probability value, measure of significance value Ramsis- Digital HumanModeling tool used for anthropometric simulation and seating posture prediction RQ – Research questions r-value – Correlation coefficient S – Short group of test participants SAE – Society of Automotive Engineers SgRP – Seating reference point StDev - Standard deviation T – Tall group of test participants TP- Test person User test- An experimental study involving TPs to get data on desired aspects VAS – Visual Analog Scale VRS - Visual Rating Scale V90/S60/XC40/XC60/XC90 – Names of Volvo Cars in different segment that are commercially available Q1 - Lower quartile, 25th %-ile value (statistics) Q2 - Median, 50th %-ile value (statistics) Q3 - Upper quartile, 75th %-ile value (statistics) viii Contents 1 Introduction 1 1.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 1.2 Problem Description . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 1.3 Project Aim . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 1.4 Delimitation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 1.5 Specification of Issue Under Investigation . . . . . . . . . . . . . . . . 4 1.6 Research Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 1.7 Disposition of Report . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 2 Theoretical Framework 7 2.1 Anthropometrics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 2.2 Sitting Posture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 2.3 Vehicle Packaging and Measuring Standards . . . . . . . . . . . . . . 10 2.3.1 Seating Reference Point, SgRP . . . . . . . . . . . . . . . . . 10 2.3.2 Torso Angle, A40 . . . . . . . . . . . . . . . . . . . . . . . . . 11 2.3.3 Head Position Contour . . . . . . . . . . . . . . . . . . . . . . 11 2.3.4 Belt line . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 2.3.5 Eye location . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 2.4 Perceived Interior Space . . . . . . . . . . . . . . . . . . . . . . . . . 14 3 Research Method 17 3.1 Qualitative Research Design . . . . . . . . . . . . . . . . . . . . . . . 17 3.2 Quantitative Research Design . . . . . . . . . . . . . . . . . . . . . . 18 3.2.1 Random Sampling of Test Runs . . . . . . . . . . . . . . . . . 18 3.3 Interview Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 3.4 Measurement Scales . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 3.4.1 Visual Analogue Scale . . . . . . . . . . . . . . . . . . . . . . 20 3.4.2 Verbal Rating Scale . . . . . . . . . . . . . . . . . . . . . . . . 21 3.4.3 Numerical Rating Scale . . . . . . . . . . . . . . . . . . . . . . 21 3.5 Paired Comparison . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 3.6 Analysis of Statistical Data . . . . . . . . . . . . . . . . . . . . . . . 23 3.6.1 Data Analysis Methodology . . . . . . . . . . . . . . . . . . . 23 3.6.2 Continuous Data . . . . . . . . . . . . . . . . . . . . . . . . . 24 3.6.3 P-value . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 3.6.4 Anderson-Darling Test for Normality . . . . . . . . . . . . . . 25 3.6.5 2-sample T-Test . . . . . . . . . . . . . . . . . . . . . . . . . . 25 ix Contents 3.6.6 Pearson Correlation Coefficient . . . . . . . . . . . . . . . . . 26 3.6.7 Crobach’s Alpha Test . . . . . . . . . . . . . . . . . . . . . . . 27 4 Phase 1 - Awareness 29 4.1 Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 4.2 Result . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 4.3 Discussion and Conclusion . . . . . . . . . . . . . . . . . . . . . . . . 30 5 Phase 2 - Identification 31 5.1 Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 5.1.1 User Study - Identification . . . . . . . . . . . . . . . . . . . . 32 5.1.2 Expectation of the Car . . . . . . . . . . . . . . . . . . . . . . 33 5.1.3 Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 5.2 Result . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 5.2.1 Expectation of the Car . . . . . . . . . . . . . . . . . . . . . . 35 5.2.2 Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 5.3 Discussion and Conclusion . . . . . . . . . . . . . . . . . . . . . . . . 41 5.3.1 Expectation of the Car . . . . . . . . . . . . . . . . . . . . . . 41 5.3.2 Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 6 Hypothesis and Research Questions 45 6.1 Hypothesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45 6.2 Exploratory Research Questions . . . . . . . . . . . . . . . . . . . . . 46 6.3 Research Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 7 Phase 3 - Investigation 49 7.1 Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 7.1.1 User Study - Investigation . . . . . . . . . . . . . . . . . . . . 49 7.1.2 Analysis Methodology . . . . . . . . . . . . . . . . . . . . . . 58 7.2 Disposition of Results . . . . . . . . . . . . . . . . . . . . . . . . . . . 62 7.3 Data Grouping - Short and Tall . . . . . . . . . . . . . . . . . . . . . 63 7.4 Quality of Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63 7.4.1 Distribution Study and Normality Test . . . . . . . . . . . . . 64 7.4.2 Test for Differentiation . . . . . . . . . . . . . . . . . . . . . . 69 7.4.3 Cluster Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . 70 7.4.4 Internal Consistency . . . . . . . . . . . . . . . . . . . . . . . 72 7.5 Analysis of Ranking Trend . . . . . . . . . . . . . . . . . . . . . . . . 72 7.5.1 Sum of Percentage . . . . . . . . . . . . . . . . . . . . . . . . 73 7.5.2 Top 3 Parameters . . . . . . . . . . . . . . . . . . . . . . . . . 74 7.6 Analysis of Vision Parameters . . . . . . . . . . . . . . . . . . . . . . 75 7.6.1 Cant Rail . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75 7.6.2 Belt line . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85 7.6.3 C-pillar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93 7.7 Analysis of Spatial Distances . . . . . . . . . . . . . . . . . . . . . . . 102 7.7.1 Vertical Distance . . . . . . . . . . . . . . . . . . . . . . . . . 103 7.7.2 Diagonal Distance . . . . . . . . . . . . . . . . . . . . . . . . . 106 7.7.3 Lateral Distance . . . . . . . . . . . . . . . . . . . . . . . . . 107 x Contents 7.7.4 Discussion and Conclusion . . . . . . . . . . . . . . . . . . . . 109 7.8 Analysis of Eye Position . . . . . . . . . . . . . . . . . . . . . . . . . 111 7.8.1 Eyellipse . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111 7.9 Analysis of Sitting posture . . . . . . . . . . . . . . . . . . . . . . . . 117 7.9.1 Discussion and Conclusion . . . . . . . . . . . . . . . . . . . . 118 8 Compared Analysis 121 9 Recommendation 127 9.1 C-pillar’s Acceptance Range . . . . . . . . . . . . . . . . . . . . . . . 127 9.2 Cant rail’s Acceptance Range . . . . . . . . . . . . . . . . . . . . . . 130 9.3 Belt line’s Acceptance Range . . . . . . . . . . . . . . . . . . . . . . . 132 10 Discussion 135 10.1 Research Appraoch . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135 10.2 Research Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135 10.3 Research Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136 10.4 Ethical and Social Aspects . . . . . . . . . . . . . . . . . . . . . . . . 137 10.5 Sustainability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138 11 Conclusion 141 11.1 Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142 References i A Ranking List from nine Experts (Identification phase) I B Documentation Sheet - Questionnaire- Phase 2 III C Documentation Sheet - User test - Phase 3 VII D Manipulation of Vision Parameters IX E Analysis of Extreme groups XI E.1 Cant Rail . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XI E.2 Belt line . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XII E.3 C-pillar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XII F VAS rating vs Angle to vision parameters (Short and Tall) XIII F.1 Cant rail . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XIII F.2 Belt line . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XIII F.3 C-pillar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XIV G 2-sample t-test for acceptance angles XV G.1 Cant rail . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XV G.2 Belt line . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XVI G.3 C-pillar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XVI H Scatterplots of VAS rating vs Spatial distance XVII xi Contents I Analysis of Eye Position XIX J Expected VAS rating for the acceptance ranges XXI K Acceptance range - extended views for the Vision parameters XXV L Research Questions XXXI xii List of Figures 2.1 Measuring sitting height (erect) . . . . . . . . . . . . . . . . . . . . . 8 2.2 Sagittal sitting postures, reprinted fromOlder children’s sitting postures when riding in the rear seat . . . . . . . . . . . . . . . . . . . . . . . 9 2.3 Vehicle interior measures and reference points, figure adapted from Ergonomics in the automotive design process . . . . . . . . . . . . . . 11 2.4 Rear view section(XX) showing the clearance measurements, figure adapted from SAE Standards, J1100 . . . . . . . . . . . . . . . . . . 12 2.5 Side view (Y-plane) of head position contour with clearance measurement of H61, figure adapted from SAE Standards, J1100 . . . . . . . . . . 12 2.6 Belt line is represented along with DGO, figure adapted from SAE Standards, J1100 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 2.7 The right and left eyellipse with the mid-eye centroid point located between the ellipses, figure adapted from SAE Standards, J941 . . . . 14 3.1 The Visual Analogue Scale with anchoring words used in the Investigation phase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 4.1 Visualizing the listed parameters A-J . . . . . . . . . . . . . . . . . . 30 5.1 Visualizing the VRS and NRS scale when assessing the expectation and preconception of the interior space . . . . . . . . . . . . . . . . . 33 5.2 The question which is asked in the end of an analysis of a car and the five level VRS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 5.3 Bar chart representing the rating given for the expectation for XC60 . 36 5.4 Bar Chart representing the rating given for the expectation for S60 . 36 7.1 The indoor environment of the user study . . . . . . . . . . . . . . . 50 7.2 The Visual Analog Scale used in the user study of phase 3 . . . . . . 50 7.3 Test set-up inside the test car showing the checkboard tapes and covered rear-rear window . . . . . . . . . . . . . . . . . . . . . . . . . 51 7.4 Test setup on the left rear window to manipulate the parameters . . 52 7.5 Interior of XC90 showing the 6 parameters in the outer-rear seat of a XC90 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 7.6 Pairwise comparison application . . . . . . . . . . . . . . . . . . . . . 54 7.7 Manipulation of the Vision parameters in the extreme position (A3/B3/C3) in reference to the nominal state . . . . . . . . . . . . . . . . . . . . . 55 7.8 Visualizing how α, β and θ are determined . . . . . . . . . . . . . . . 56 xiii List of Figures 7.9 Visualizing the vertical distance manipulation with flat mock-up roof 56 7.10 The checkboard positioned inside the XC90. The upper red marker represent the mid-eye centorid point and the lower the SgRP point . . 59 7.11 Two images are overlaying each other; reference picture and the test person side picture. By having the upper layer transparent it is possible to match the images and determine the difference in eye position from the reference point . . . . . . . . . . . . . . . . . . . . 60 7.12 Presenting all the vision angles and how they were determined in a CAD file . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 7.13 Presenting the four sitting posture categorizes used in this study. ‘a’ head on head rest, ‘b’ head off and body in a upright/natural posture, ‘c’ head positioned far forward and ‘d’ sitting in an extreme/awkward posture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62 7.14 Histogram plot showing the distribution of whole group based on sitting height. Contains results from Anderson Normality test . . . . 64 7.15 Histogram graphs showing the distribution of Short and Tall group based on sitting height. Contains results from Anderson Normality test 65 7.16 Histogram showing female test persons’ distribution in sitting height . 65 7.17 Histogram showing male test persons’ distribution in sitting height . 66 7.18 2 sample t-test with boxplot for Short (blue box) and Tall (red box) group based on sitting height . . . . . . . . . . . . . . . . . . . . . . 69 7.19 Dendrogram (initial) with all parameters showing clusters and similarity between the parameters . . . . . . . . . . . . . . . . . . . . . . . . . 70 7.20 Dendrogram with six parameters showing the cluster and similarity between the parameters . . . . . . . . . . . . . . . . . . . . . . . . . 71 7.21 Histogram presenting the summarized percentage for respective parameter in Short (S) and Tall (T) group . . . . . . . . . . . . . . . . . . . . . 73 7.22 Presenting the amount of times a specific parameter has been ranked as either 1, 2 or 3 as the most influential parameter, where 1 indicates the highest rank . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74 7.23 Boxplot showing the distribution of VAS rating for the various positions of Cant rail (nominal state, A1,A2 and A3) . . . . . . . . . . . . . . . 76 7.24 Box plot showing the distribution of VAS rating for the various positions of Cant rail (nominal state, A1, A2 and A3) between Short and Tall group . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77 7.25 Line plot visualizing the difference in VAS rating compared to the given rating in nominal state of the Cant rail. The blue coloured dots and lines represent the Short group and red coloured marks represent the Tall group . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79 7.26 Line plot showing the difference in VAS rating compared to previous change of Cant rail. The blue coloured dots and lines represent the Short group and red coloured marks represent the Tall group . . . . . 79 7.27 Scatterplot of α-angle and VAS rating for all the determined Cant rail changes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80 7.28 Histogram with a distribution fit for the whole test group . . . . . . . 82 7.29 Histogram with a distribution fit for acceptance angle . . . . . . . . . 82 xiv List of Figures 7.30 Box plot showing the distribution of VAS rating for the various positions of Belt line (nominal state, B1,B2 and B3) . . . . . . . . . . . . . . . 85 7.31 Box plot showing the distribution of VAS rating for the various positions of Belt line (nominal state, A1, A2 and A3) between the Short (blue marked) and Tall group (red marked) . . . . . . . . . . . . . . . . . . 87 7.32 Line plot visualizing the difference in VAS rating for the different Belt line positions, compared to the mean of the nominal state. Blue represent Short group, Red represent Tall group . . . . . . . . . . . . 88 7.33 Line plot visualizing the difference in VAS rating for the changes in the position of Belt line. Redline represents the Tall group and the blue line represents the Short group . . . . . . . . . . . . . . . . . . . 89 7.34 Scatterplot of β angle and VAS rating for all the determined Belt line changes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90 7.35 Histogram with a distribution fit for the whole test group of Belt line’s acceptance angles . . . . . . . . . . . . . . . . . . . . . . . . . . 91 7.36 Histogram with a distribution fit for acceptance angle . . . . . . . . . 91 7.37 Box plot showing the distribution of VAS rating for the various positions of C-pillar (nominal state, C1, C2 and C3) . . . . . . . . . . . . . . . 94 7.38 Box plot for VAS rating distribution for Short and Tall group . . . . 95 7.39 Line plot visualizing the difference in VAS rating for the changes in the position of C-pillar from nominal state. Red line represents the Tall group and the blue line represents the Short group . . . . . . . . 96 7.40 Line plot visualizing the difference in VAS rating for the changes in the position of C-pillar . Red line represents the tall group and the blue line represents the short group . . . . . . . . . . . . . . . . . . . 97 7.41 Scatterplot of θ-angle and VAS rating for all the determined C-pillar changes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98 7.42 Histogram with a distribution fit for the whole test group of C-pillar’s acceptance angles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99 7.43 Histogram with a distribution fit for acceptance angle in S/T . . . . . 100 7.44 Scatterplot for the whole group determining the correlation between the VAS rating and the Vertical distance. The regression line is presented as well with a r-value at 0.49 . . . . . . . . . . . . . . . . . 103 7.45 Scatterplot of Vertical distance vs VAS rating, with differentiation of having the vertical change done (red dots) or not (blue dots) . . . . . 104 7.46 Scatterplot for the whole group where the VAS rating is plotted against the Diagonal distance, the regression line is visualized as well with a r-value at 0.48 . . . . . . . . . . . . . . . . . . . . . . . . . . . 107 7.47 Short group’s scatterplot of the Lateral distance and the VAS rating, plotted with the regression line . . . . . . . . . . . . . . . . . . . . . 108 7.48 Scatterplot where the VAS rating is plotted against the Lateral distance, with identified people sitting parallel to the window divider bar (red dots) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109 7.49 The whole group’s eyellipse is fitted within the restricting rectangular by having its corners coincide to the 95th and 5th %-ile of eye point . 112 xv List of Figures 7.50 The eyellipses; yellow represent the whole group, blue the Short group and purple the Tall group . . . . . . . . . . . . . . . . . . . . . . . . 113 7.51 2D presentation of the test persons’ eye points, colour marked in Short/Tall group, in correlation to the mid-eye centroid point . . . . 113 7.52 Correlation analysis with the linear correlation visualized for the X-coordinate of the eye point and VAS rating of the nominal state of the car . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114 7.53 Scatterplot of X-coordinate of eye position(EX) vs VAS rating . . . . 114 7.54 The whole group’s Pearson Correlation Coefficient analysis resulting in a scatterplot with a given r-value at -0,53 . . . . . . . . . . . . . . 115 7.55 The four sitting posture categorize and their marker . . . . . . . . . . 117 7.56 Scatterplot of VAS rating vs X-coordinate of eye position . . . . . . . 118 8.1 Comparison lineplot for mean VAS rating vs positions of Vision parameters122 9.1 The whole group’s eyellipse with all the eye points. The green symbol is representing the mid-eye centorid point for the car and the back cross on the eyellipse’s line contour is representing the 95th %-ile (furthest back) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127 9.2 Histogram presenting C-pillar last acceptable angle for the whole group with the coloured acceptance range displayed . . . . . . . . . . 128 9.3 The coloured acceptance range created from the 95th %-ile of eye point intersecting the yellow eyellipse (whole group) at a furthest back position for C-pillar . . . . . . . . . . . . . . . . . . . . . . . . . 129 9.4 The bounding angles Q1 and Q2 for C-pillar (whole group) are marked out with a vertical line, where the line is intersecting the trendline is where the VAS rating is obtained, marked with a horizontal line . . . 129 9.5 The Short (blue) and Tall (purple) group’s eyellipse. The green symbol is representing the mid-eye centorid point for the car and the back cross on the eyellipse’s line contour is representing the 95th %-ile (furthers up) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130 9.6 With the colour scheme for the acceptance ranges the “Acceptable” /”Critical”/”Poor” are presented in the histogram for the a) Short group b) Tall group . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131 9.7 Images taken in CAD showing Cant rail’s acceptance ranges for the a) Short group and b) Tall group, from the 95th %-ile eye point . . . 131 9.8 Histogram presenting the last acceptable angles for β (Belt line) together with the representative colours for “Acceptable” (green), “Critical” (orange) and “Poor” (red) . . . . . . . . . . . . . . . . . . 133 9.9 Images taken in CAD showing Belt line’s acceptance ranges for the a) Short group and b) Tall group, from the 50th and 95th %-ile eye point . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134 10.1 Icons of the focused sustainability goals . . . . . . . . . . . . . . . . . 138 B.1 Documentation sheet in excel for the questionnaire of phase 2 - part 1 III B.2 Documentation sheet in excel for the questionnaire of phase 2 - part 2 IV xvi List of Figures B.3 Documentation sheet in excel for the questionnaire of phase 2 - part 3 V C.1 Documentation sheet - user test - phase3 (part 1) . . . . . . . . . . . VII C.2 Documentation sheet - user test - phase3 (part 2) . . . . . . . . . . . VIII D.1 Manipulation of the Cant rail positions . . . . . . . . . . . . . . . . . IX D.2 Manipulation of the Belt line positions . . . . . . . . . . . . . . . . . X D.3 Manipulation of the C-pillar positions . . . . . . . . . . . . . . . . . . X E.1 Line plots visualizing the change in VAS rating from nominal state and from previous positions . . . . . . . . . . . . . . . . . . . . . . . XI E.2 Line plots visualizing the change in VAS rating from nominal state and from previous positions . . . . . . . . . . . . . . . . . . . . . . . XII E.3 Line plots visualizing the change in VAS rating from nominal state and from previous positions . . . . . . . . . . . . . . . . . . . . . . . XII F.1 Scatterplot of VAS rating vs α-angle . . . . . . . . . . . . . . . . . . XIII F.2 Scatterplot of VAS rating vs β-angle . . . . . . . . . . . . . . . . . . XIII F.3 Scatterplot of VAS rating vs θ-angle . . . . . . . . . . . . . . . . . . . XIV G.1 Box plot of 2-sample t-test with test hypothesis . . . . . . . . . . . . XV G.2 Box plot of 2-sample t-test with test hypothesis . . . . . . . . . . . . XVI G.3 Box plot of 2-sample t-test with test hypothesis . . . . . . . . . . . . XVI H.1 Scatterplot of VAS rating vs Vertical distance . . . . . . . . . . . . . XVII H.2 Scatterplot of VAS rating vs Diagonal distance . . . . . . . . . . . . . XVII H.3 Scatterplot of VAS rating vs Lateral distance . . . . . . . . . . . . . . XVIII I.1 Scatterplot of Eye coordinates (EX vs EZ) with regression fit line for Short and Tall group . . . . . . . . . . . . . . . . . . . . . . . . . . . XIX I.2 Contour plot of VAS rating for EX vs EZ . . . . . . . . . . . . . . . . XX J.0 The bounding angles Q1 and Q2 for Cant rail and Belt line, in respective groups; Short and Tall, are marked out with vertical lines. Where the lines are intersecting the trendline is where the VAS rating is obtained, marked with horizontal lines. . . . . . . . . . . . . . . . . XXIII K.-3 The Vision parameters acceptance range designed from the 95th %-ile for different groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . XXX xvii List of Figures xviii List of Tables 2.1 Anthropometric data for the Swedish population . . . . . . . . . . . . 8 3.1 The data analysis process with the purpose of the steps . . . . . . . . 23 3.2 The inference for respective Pearson’s correlation coefficients . . . . . 27 5.1 Fourteen parameters which were used in the application “Pairwise comparison” . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 5.2 All 26 parameters identified which influence the perceived upper interior space are presented. The parameters are separated in two groups as a high level (Car parameters and Human factors) and in sub-levels within the group . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40 5.3 Top five areas listed as most influential . . . . . . . . . . . . . . . . . 43 7.1 The data analysis steps with the respective analysis done for each step 58 7.2 Anthropometric data for the Swedish population . . . . . . . . . . . . 63 7.3 Sitting heights values for the Swedish female population (Swe) and the 18 female test persons (TP). The row called “Difference” is the calculated difference between the data sets . . . . . . . . . . . . . . . 66 7.4 Sitting height values for the Swedish male population and 15 males participating the user test, excluding one outlier . . . . . . . . . . . . 67 7.5 Sitting height values of the calculated mean compared to the collected data of 34 participants . . . . . . . . . . . . . . . . . . . . . . . . . . 67 7.6 Summarized anthropometric data representing Swedish population and the collected data . . . . . . . . . . . . . . . . . . . . . . . . . . 68 7.7 All the ranked percentages added together and summarized in Short and Tall group for respective parameter . . . . . . . . . . . . . . . . . 73 7.8 Presenting Short and Tall group’s sequence of parameters that are affecting them the most in a falling order . . . . . . . . . . . . . . . . 74 7.9 The order of most affecting parameter in a falling arrangement from the count as a top 3 parameter . . . . . . . . . . . . . . . . . . . . . 75 7.10 Summarized Q1, Q3 and mean values for the various position of Cant rail . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76 7.11 Summarized P-values of the whole group for the comparing levels of VAS-rating . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77 7.12 Summarized Q1, Q3 and mean values for the various position of Cant rail for Short and Tall group . . . . . . . . . . . . . . . . . . . . . . . 78 xix List of Tables 7.13 P-value presented from the 2-sample t-test between the Short and Tall group for the Cant rail positions . . . . . . . . . . . . . . . . . . 78 7.14 Commonly mentioned comments as description and the respective counts for Short and Tall group . . . . . . . . . . . . . . . . . . . . . 80 7.15 The count of the identified last acceptable level for Short and Tall group . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81 7.16 Cant rail change presented in degrees, determined by the mean from all 34 participants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83 7.17 Summarized Q1, Q3 and mean values for the various position of Belt Line. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86 7.18 Summarized P-values of the whole group for the comparing Belt line’s levels of VAS rating . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86 7.19 Summarized Q1, Q3 and mean values for the various position of Belt line for Short and Tall group . . . . . . . . . . . . . . . . . . . . . . . 87 7.20 P-value presented from the 2-sample t-test between the Short and Tall group for the Belt line positions . . . . . . . . . . . . . . . . . . 87 7.21 Commonly mentioned comments as description and the respective counts for Short and Tall group . . . . . . . . . . . . . . . . . . . . . 89 7.22 The count of the identified last acceptable level for Short and Tall group, regarding Belt line positions (nominal, B1-B3) . . . . . . . . . 90 7.23 Summarized Q1, Q3 and mean values for the various position of C-pillar 94 7.24 Summarized P-values of the whole group for the comparing C-pillar’s levels of VAS rating . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95 7.25 Summarized Q1, Q3 and mean values for the various position of C-pillar for Short and Tall group . . . . . . . . . . . . . . . . . . . . 96 7.26 P-value presented from the 2-sample t-test between the Short and Tall group for the C-pillar positions . . . . . . . . . . . . . . . . . . . 96 7.27 Commonly mentioned comments as description and the respective counts for Short and Tall group . . . . . . . . . . . . . . . . . . . . . 98 7.28 The count of the identified last acceptable level for Short and Tall group, regarding Belt line positions (nominal, C1-C3) . . . . . . . . . 99 7.29 Mean θ-angle determined for every C-pillar level, separated in S/T group . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100 7.30 Summarized the mean and Q3 values in respective groups; Short and Tall. The parenthesis under the row description of “Count” in “VAS rating of changed vertical height” indicate how many went through with the change . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105 7.31 Commonly mentioned comments as description and the respective counts for Short and Tall group . . . . . . . . . . . . . . . . . . . . . 106 7.32 Summarized r-values for the VAS rating and the Diagonal distance in a nominal state for the whole group, Short and Tall group . . . . . 106 7.33 The calculated Pearson correlation coefficient for the Lateral distance and VAS rating in the nominal state of the car . . . . . . . . . . . . . 107 7.34 Summarized all the calculated correlation coefficient for the Spatial distances for Short, Tall and the whole group . . . . . . . . . . . . . . 110 7.35 Specification of the eyellipses . . . . . . . . . . . . . . . . . . . . . . . 112 xx List of Tables 7.36 Pearson correlation coefficient calculated for the 3 groups . . . . . . . 115 7.37 Summarized r-value between the VAS rating and the x/z-coordinates 116 7.38 Calculated ratio for a chosen sitting posture within the group S/T . . 119 8.1 Rank of the most influential parameters by the VAS rating . . . . . . 122 8.2 Rank of most influential parameters - rank analysis . . . . . . . . . . 122 8.3 Summarized table for Pearson’s correlation coefficient for all Vision parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123 8.4 The initial ranking for both Vision parameters and Spatial distances . 123 8.5 Summarized all the calculated correlation coefficient for the Spatial distances in Short, Tall and the whole group . . . . . . . . . . . . . . 124 8.6 Acceptance angles for the short and tall group . . . . . . . . . . . . . 125 9.1 Predicted VAS rating at the bounding angles Q1 and Q2 . . . . . . . 130 9.2 Bounding angles for the Short and Tall group for Cant rail . . . . . . 130 9.3 The predicted VAS rating for the three acceptable ranges, for Cant rail132 9.4 The predicted VAS rating for the three acceptable ranges, for Belt line133 10.1 Parameters with respective exposed angles . . . . . . . . . . . . . . . 137 A.1 Ranking list from the participants ordered in increasing sitting height. The number in brackets are the calculated percentage given from the “Pairwise comparison”. . . . . . . . . . . . . . . . . . . . . . . . . . . I xxi List of Tables xxii 1 Introduction The ergonomics engineering work in vehicle design covers a great deal of aspects. What this thesis investigates is covered within the area of occupant packaging (Bhise, 2011a), but in a more precise description it is about the upper interior space and the user perception of the space in the out-rear seat of a car. The development within vehicle packaging can be grouped in six areas but just two of them will be studied in depth; comfortable seated posture covering roominess aspects, and visibility of interior and exterior areas. Because these two areas are separated at the company, there was an inquiry to investigate if they are affecting each other and in what extent. A generated knowledge would afford new insights and work as a foundation to refine requirements based on the human’s experience of interior space. Little research has been done specifically in this area regarding the perception of interior space in vehicles. However, numerous studies have been conducted in the two subjects separately; perception and interior space, and roominess in vehicles. Firstly, the two areas were studied to collect knowledge of how to connect them and analyze, as well what area would be possible to include in the time frame and resources of this thesis. This are presented in the Theoretical framework chapter (2) and as limitations in this chapter under Delimitation (1.4). Specific limitations for different phases and sub-studies are presented in the Method chapter for respective study. Two studies were conducted in this research to conclude upon parameters to analyze and to find trends, different correlations, measures and angles to arrive at recommendation and guidance for how to emphasis the perception of space in design parameters. This chapter begins with an overview of the background of the topic to provide a knowledge to understand the area which this thesis touches upon. Following this comes the problem description and the aim for this project. Listed limitations for this research are stated in the limitations followed by explaining the exploratory research questions and research questions for the different research phases. The chapter concludes with a research approach of how to reach the aim where different phases will be introduced with related sub-studies. As an additional section, the disposition of the report will be described for you, as a reader, to be able to follow the work in a straightforward manner. 1 1. Introduction 1.1 Background Volvo Cars is a well-known and respected premium car brand with an international market. The stakeholders of the thesis outcome is the Ergonomics team within the Customer Experience Center department at Volvo Cars. The Customer Experience Centre is constantly working towards the aim of satisfy their costumers’ desire of a complete car experience. The experts at the department are predicting customer expectations, behaviors and physical prerequisites to be able to provide the cars with what their customer want. Creating sufficient headroom inside the car is crucial for customer satisfaction. The available upper interior space is directly influenced by the seating position and posture, and is therefore one important part for creating overall riding comfort. The needed upper interior space is depending on both the anthropometry of the users, determined ergonomic dimensions (SAE standards) and also users’ expectations regarding type and size of vehicle, cultural differences and usage of the vehicles. Volvo Cars has identified a need to investigate the human perception of head roomi- -ness, which here is defined as the upper interior space. In other words, a new way of defining the head roominess is needed as well as creating a standardized way to measure the upper interior space. Looking into the ergonomics engineering work within the occupant packaging in vehicle design there are defined standards which the SAE International has launched that are commonly used in the vehicle industry. Three SAE papers have mainly been studied for the topic within seating posture and visibility for better understanding of relevant design parameters; SAE standard J1100 – Motor vehicle dimensions (SAE International, 2009c), SAE J1052 – Motor vehicle driver and passenger head position (SAE International, 2017) and, SAE J941 – Motor vehicle drivers’ eye locations (SAE International, 2010b). These papers will be further introduced in the Theoretical Framework, chapter 2, later in the report. However, the conclusion which was drawn after looking into the SAE standards was that only objective measures in respective areas of roominess and visibility were mentioned. A possibility of improving these guidelines was recognized and to identify common design parameters which would connect visibility with roominess. Due to a lack of knowledge and investigations based on the rear seat of the car along with less affecting parameters and complexity of investigation in this subject, compared to the front seat, the left outer-rear seat was chosen to be the fixed position of the study. 1.2 Problem Description The main problem underlying this thesis study is that head roominess evaluation is considered as a physically measured entity of SAE standards, when it is actually a substantially a combination of physical measurement (objective) and perceived 2 1. Introduction roominess which is subjective to the user. User’s personal and subjective perception is not only based on variations in anthro- -pometry such as sitting height and eye position. But it is also about cognitive factors such as compensational behavior, preconception and user expectation (Yanagisawa &Miyazaki, 2019). Therefore, evaluating upper interior space should include objective and subjective analysis. Subjective parameters alter with culture and market, model of a car as well with time (e.g. trends and preferences). One of the challenges with this thesis is how to measure the subjective perception of roominess and convert it to a objective guidance to be used in combination with the existing measures. 1.3 Project Aim The aim is to study the perception of the upper interior space of the user. This is done by utilizing methods to identify subjective impressions and objective parameters to conclude what and how parameters affect the perception of space. The desired result for this thesis is to create guidance and provide recommendations of how to assess and evaluate the upper interior space in the rear-outer seat of a car. This includes physical measurable design parameters with respect to the perceived roominess regarding the visibility. By developing these recommendations, the SAE standard is studied against perceptiveness and acts as supplement while designing interior spaces in a more human-centric approach. 1.4 Delimitation Time, available resources and previous study results are considered in arriving at limitations thus defining the scope of the thesis work. Limitations for the designed user studies have also been made but are presented separately in the method chapter respectively. The major limitation regarding the whole project are presented below. The headroom assessment is performed at Volvo Cars, as well usage of the company’s own cars for evaluation. However, only cars with standard roof are used in this research due to proven impact of a panoramic roof on the perceived roominess (Bentioulis & Forsberg, 2015). Volvo Cars’ internal test persons are preferred to be used in the studies. However, other participants are involved, but all test persons fall under the category of adults, from the age of 18. The study is performed with the Swedish population and is thereby restricted to the Swedish market. 3 1. Introduction 1.5 Specification of Issue Under Investigation This research includes the identification of parameters that influence the perceived upper interior space in the rear-outer seat. Furthermore, the identified parameters are investigated to find their relations with two differentiated groups dividing the test persons in Short and Tall. What also is investigated is the level of influence of the parameters, interdependencies between them, and the influence of preconception on the user regarding the upper interior space, among others. Based on these research areas, the recommendations consider both objective and subjective parameters to assist the design process to follow a more human-centric approach. Several exploratory research questions (ERQ) for phase 2 and 3 are developed which are essential to elaborate on to enable conclusive recommendations in the end. The aim for all these exploratory research questions is not only to provide a basis of how the execution and methodology should focus on. But also relate to a set of questions which are the main finding, the research questions (RQ). The ERQ will be presented and answered in the respective chapters. The thirteen research questions address back to the project aim and lead the research to desired results which are attained in the analysis part of phase 3 – the Investigation phase (chapter 7). These research questions are presented in chapter 6. Hypothesis and Research Questions, before entering phase 3 along with 6 hypotheses (H1-6) that were extracted and created from the result of phase 2, the Identification phase. 1.6 Research Approach The research approach this project follows is divided into three main phases; Awareness, Identification and Investigation. Phase 1: Get adequate knowledge within the subject and the research area to plan and design a relevant framework for this thesis. This first phase called Awareness is covered in the Theoretical Framework (chapter 2), as well in Phase 1 – Awareness (chapter 4) where an evaluation is designed for four cars to create awareness of parameters affecting the perceived roominess. Phase 2: Named Identification phase, is where a qualitative user study was designed to assess important perspectives and identify the most influential parameters for the perceived roominess. In this study two car models were used where 9 experts within Customer Experience Centre were invited. Specific exploratory research questions (ERQ 2.1-2.4) were answered within this phase and acted as preparative work to design and specify the substantial study in the third phase; Investigation. Phase 3: The purpose of phase 3, the Investigation, is to understand the impact of the influential parameters on the perceived upper interior space. A quantitative user study was conducted with 34 test persons to collect large amount of data 4 1. Introduction to do statistical analyses. Five exploratory research questions (ERQ 3.1-3.5) were generated to design the user test to provide desirable data to answer the stated research questions later. In the result part of this phase is where the statistical analyses are presented (chapter 7). Description of the used statistical analyses methods are presenter in the Research Method (chapter 3), for better understanding of the statistical data. 1.7 Disposition of Report This report is structured in a chronological order where the fundamental and interest- -ing literature studies for this research are presented in chapter 2 Theoretical Framework. All substantial research methods which are used within this thesis are described in chapter 3 Research Method. Afterwards will the research phases Awareness, Identification and Investigation be divided into separate chapters (4/5/7) where the method, result, discussion & conclusion will be presented individually for every research phase. Between the second phase and the third phase chapter 6 Hypothesis and research questions is intended to bring clarity of what are investigated in phase 3. All hypotheses, exploratory research questions and the main research question will be presented in this chapter. Chapter 8 Compared Analysis is where the results from the Investigation phase are being compared with each other to bring context to the result presented individually in chapter 7 Phase 3 - Investigation. Interrelationships are also identified between various findings for better understanding of the result. In chapter 9 Recommendation, an extended analysis of the critical angles are developed into recommendations for acceptable angles for all parameters. This chapter will act as a concretized guidance when assessing the perceived roominess by manipulating the Vision parameters. In chapter 10 Discussion the bigger discussions regarding the whole research and findings are discussed together with some ethical, social and sustainability aspects. Finally, a short and conclusive description of the main findings are described in chapter 11 Conclusion, ending with suggestions of future work. 5 1. Introduction 6 2 Theoretical Framework Relevant research papers and information with importance for this project are presented in this chapter. It starts with explaining the anthropometrics, which was of high relevance for this project, where as well the Swedish population was studied. It continues to present a study regarding the sitting posture that was the main source of inspiration for categorizing these posture in the analysis stage of phase 3 (Sitting posture). Furthermore, essential measurement from the SAE standards of how roominess and visibility are acknowledged in Vehicle packaging and measuring standards. Also, research papers about how to interpret perception and expectations are presented in Perceived interior space. 2.1 Anthropometrics According to Haslegrave (2005) and Bhise (2011a), anthropometrics and biomechanics is the branch of science that deals with body measurements related to size, shape, strength, mobility, flexibility and working capacity. It is based on the fact that humans are variable in dimensions and proportions and that there is a limit of motion and applicable forces before stretching beyond the capacity of the muscles and tendons and where the human is in risk of injure herself. Hence, the human-centric approach within the department of Customer Experience Centre, needs the knowledge of this variability to design appropriate products for the specific user. By knowing a populations anthropometrics, several dimensions and requirements for a product can be set to accommodate the user. When studying a population’s anthropometrics it reflects a normal distribution that is symmetrical about its highest point, which is used to calculate the percentile values of populations. Due to the normal distribution 50% of the population is shorter than the mean and 50% is taller, the mean can be characterize as the fiftieth percentile, abbreviated as 50th %-ile. Commonly extremes are usually determined below the fifth percentile (5th %-ile) and above ninety-fifth percentile (95th %-ile) which makes it common to design products within the range of 5th to 95th %-ile to include 90% of the population (Pheasant & Haslegrave, 2005). Because this study aims to create guidelines and provide recommendation to accom- -modate the Swedish population, Lars Hanson’s research about the Swedish anthropometrics for the product and workplace design was studied (Hanson et al., 2009). In his 7 2. Theoretical Framework research he investigated if the anthropometrics of the Swedish population have changed compared with data from 1969. 367 people participated, aged 18-65, representing the Swedish population. 43 anthropometrics data points where measured for 262 female and 105 males and the data was presented in different percentile values of the Swedish population. The measure of interest for this thesis study was the sitting height. These values were used to differentiate between two data sets which were compared with each other in the the data analysis part of phase 3. Different percentile values as well the minimum and maximum values for the sitting height covering both female and male of a Swedish population are presented in the table 2.1. Table 2.1: Anthropometric data for the Swedish population Sitting Height (mm) Min 5th %ile 50th %ile 95th %ile Max Female 773 832 892 949 993 Male 841 883 946 1006 1029 According to ISO 7250 the sitting height (erect) is a vertical distance taken from the vertex of the head to the sitting surface (Taifa & Desai, 2017). The figure below (figure 2.1) visualizes the measure. Figure 2.1: Measuring sitting height (erect) In the book Bodyspace (Pheasant & Haslegrave, 2005), four different sets of constraints were described within the anthropometrics which are usually causing conflicts between criteria: clearance, reach, posture and strength. Clearance was the constraint that was studied in this thesis, which focuses in providing sufficient space and access. These constraints are usually determined as a minimum acceptable dimension for the larger members in a population (95th %-ile and above). 8 2. Theoretical Framework 2.2 Sitting Posture During the evaluation and investigation of the result obtained from phase 3 in this research, awareness was raised on how the participants were sitting in the seat, in particular where the head was placed in relation to the head rest. In the research paper Older children’s sitting postures when riding in the rear seat (2011a) were observed with children aged 8-13 years old. Both lateral sitting postures, fore and aft differentiation were documented and a classification system was designed. Three different head positions were identified and combined together with numerous torso position in fore and aft direction as well lateral. Thirteen different sagittal sitting posture were identified and did act as an inspiration source to how this project categorised the sitting postures together with the head positions (Sitting posture section 7.9). Figure 2.2: Sagittal sitting postures, reprinted from Older children’s sitting postures when riding in the rear seat (Jakobsson, Bohman, Stockman, Andersson, & Osvalder, 2011b) In the figure 2.2, the 13 different sitting postures are visualized. The head position were defined in three positions; ‘a’ head against the head rest, ‘b’ head upright relative to the torso, and ‘c’ head leaning forward relative to the torso. When it comes to the torso position it was categorized in five position (Torso a-e). ‘a’ was defined as the entire back against the backrest, ‘b’ the entire back except shoulders are against the backrest, ‘c’ sitting in a upright position but is not leaning against the backrest, ‘d’ no contact with the backrest while the torso is leaning forward, and ‘e’ sitting in a slouched position without having any contact with the backrest and the lower back. From the figure 2.2 the illustrative sitting postures’ for ‘aa’, ‘bb’, ‘bc’ and ‘ec’ were used to describe the head and sitting posture of the test person in this project. However, other descriptions for these illustration as well categories were used, due to less complex analysis of the person’s sitting posture. This is further described in phase 3, Sitting posture. 9 2. Theoretical Framework 2.3 Vehicle Packaging and Measuring Standards When talking about clearance and space in terms of a vehicle, occupant packaging is usually the term which is used to describe the placement of systems and components in the vehicle architecture. Apart from the consideration of fitting various components together, for example the powertrain, chassis, electrical components and a body (interior, climate control systems and exterior) the main focus is that the vehicle design has to accommodate the user – the driver and passengers (Gkikas, 2012). The objective of ergonomics is the evaluation and design of the interaction between people and systems. The interaction can be in the form of physical contact, movement and cognitive interactions. In the literature “Design for ergonomics” (Tosi, 2020) the ergonomic engineering work is described as putting the user and his needs in the centre of the design process by using knowledge and methodological tools. It involves emphasis on anthropometry, biomechanics and psychology to apply the human capabilities and characteristics to design a suitable vehicle (Gkikas, 2012). Computer Aided Design (CAD) is a tool commonly used for designing a product (Encarnação, Lindner, & Schlechtendahl, 1990). It supports the design process and the architecture of a whole product and produces information which can be accessed globally in a company. When looking into these three-dimensional graphics different design interfaces can be visualized together with defined criteria and standards. Society of Automobile Engineers International (SAE International) is a global associ- -ation leading the development of standards in the automotive, commercial-vehicle and aerospace industries (SAE International, 2020). Three technical research papers, published by the SAE, have been studied to access ergonomics standards regarding motor vehicle dimensions (SAE J1100) (2009c), head position (SAE J1052) (2017) and eye locations (SAE J941) (2010b). The measurement standards that were considered as relevant were within the area of Visibility of interior and exterior areas, which study the location and movement of eye and head during gathering of visual information, and available fields of view. The other area was the Comforable seated posture, which studying head and shoulder room. Both areas are consider within the Occupant packaging which was described in Bhise book Ergonomics in the automotive design process (2011a). Relevant measures are presented below. 2.3.1 Seating Reference Point, SgRP The seating reference point (SgRP) is a critical reference point which is used to position other feature and defining other important vehicle dimensions, see figure 2.3. It is established at one specific location of its designated seating positions within the range of seat up/down and fore/aft adjustment. Any other location in the travel path of the seat is referred as a designed H-point. 10 2. Theoretical Framework Figure 2.3: Vehicle interior measures and reference points, figure adapted from Ergonomics in the automotive design process (Bhise, 2011b) 2.3.2 Torso Angle, A40 The torso angle, also defined as “back angle” (A40) is referring to a designated seating position of an angle created by the vertical line in SgRP and a line which is parallel to the back pan contour, defined as the torso line. In figure 2.3 both the torso line and A40 are presented. 2.3.3 Head Position Contour The surface of the head position contour, also called head clearance envelop, is used to help establish an upper interior space with enough room around the head with additional hair, see figure 2.3. The desired head position contour is determined by the three following factors: H-point travel path (fore/aft and up/down adjustment from SgRP), percentile value of accommodation and torso angle (A40). Once the contour is set, different views of normal cut sections are used to determine distances to interior surfaces to create directions for head clearance. The following distances are determined as head clearance measurements following the head position contour of the appropriate percentile. Head clearance diagonal (W27) The minimum diagonal distance, at an angle 30 degrees above horizontal, to any limiting surface in X-plane, see figure 2.4. Head clearance lateral (W35) The minimum lateral distance to any limiting surface in X-plane, see figure 2.4. 11 2. Theoretical Framework (a) W27 (b) W35 (c) H35 Figure 2.4: Rear view section(XX) showing the clearance measurements, figure adapted from SAE Standards, J1100 (SAE International, 2009b) Head clearance vertical (H35) The minimum vertical shift of the head contour section to any limiting surfaces. H35 is regarding the cut on the X-plane, see figure 2.4. Effective head room (H61) Along a line through SgRP with a vertical angle leaning 8 degrees rear (Y-plane) creating a distance from the head contour to the first limiting surface, see figrue 2.5. Head clearance minimum (W38) Compared to W27 and W35 will W38 refer to the true minimum 3D distance to any surface from the head contour. Figure 2.5: Side view (Y-plane) of head position contour with clearance measurement of H61, figure adapted from SAE Standards, J1100 (SAE International, 2009b) 12 2. Theoretical Framework 2.3.4 Belt line The Belt line is defined by a side view (Y-plane) as the lower edge of a boundary of a transparent area on the surface of a vehicle glass assembly, also known as the design glass outline (DGO), see figure 2.6. Related measures to the Belt line is the Belt height (H25), a vertical distance from the SgRP to the bottom of the side window DGO (in X-plane). The property H25 can be seen in figure 2.3. Interior Belt height (H26) is another vertical distance but is measured from the top surface of the interior trim (excluding window seals in the inside of the car) to the SgRP. The DGO was used when identifying the parameters Belt line and Cant rail during the research phases Identification and Investigation. Also, the manipulations for the Belt line and Cant rail had the glass outline as the reference point for the different changing levels, see figure 2.6. Figure 2.6: Belt line is represented along with DGO, figure adapted from SAE Standards, J1100 (SAE International, 2009a) 2.3.5 Eye location By locating the eyes which can be described as eyellipses, visibility analyses can be done as well determine the placement of objects to ensure adequate vision. The mid-eye centroid point which is determined between the right and left eyellipse centroids are located at the centre line of occupant (in line with the SgRP), and can be used as a reference point. In figure 2.7,the mid-eye centoid point are located on the Y-axis of the eyellipse and between the left and right eyellipses’ Z/X-axis. When determining the eye position in Investigation phase (7.1.2.2) of this study, the designed mid-eye centroid point where used as a reference point to understand the participants eye position in relationship to its environment. A side view of the eyellipse and the mid-eye centroid point in context of the car can be seen in figure 2.3. All the measurements can be determined in different seating position; driver, second row passenger or third row. Since the study was specified to the second row, all the presented measures are referred to the second row outboard passenger on the driver side of the vehicle. 13 2. Theoretical Framework Figure 2.7: The right and left eyellipse with the mid-eye centroid point located between the ellipses, figure adapted from SAE Standards, J941 (SAE International, 2010a) 2.4 Perceived Interior Space In the book Perception (2003), Maund states perception to be a natural process by which humans understand and acquire knowledge of the objective world around them. But this process of perception is highly subjective, that it differs from person to person or differs in several occasions for the same person. Hence, study about perception would require two different views, scientific as well as philosophical. In the book Human-Computer Interaction (Hwang, Kim, Ahn, & Jung, 2011), it is discussed that there is a tendency for the user of the motor vehicles to perceive the interior space more as a psychological space rather than physical space. Though different cars have similar interior space by volume, they can be perceived differently due to other characteristics of the interior space. Thus it is important to study this perceived psychological space in addition to the headroom measures, which has been discussed earlier (section(2.3.3)). In the same study it was concluded that the feeling of narrowness creates a negative feeling with the users. This traces back to the necessity to create a feeling of broader space to create positive feeling among the users. In Yanagisawa and Miyazaki’s paper regarding extracting expectation effect in user- product interactions (2019) the Kano-model with the definition of Must-be and Attractive qualities are used to define design with perceived qualities. The Must-be and Attractive qualities need to be translated into engineering properties to be able to meet customers’ requirements and expectations. Studying the user in a time sequence of user-product interaction a cyclic interaction of action, sensation and meaning will be performed. Inbetween the states the sensor modality is changing. For the different transition of states, the user usually predicts these in advance from previous experience as well from evaluation of the perceived qualities on perceived features. For example, by just looking at some object the person will create an 14 2. Theoretical Framework assumption and expecting it to smell in a specific way, as well taste or predict its weight. Whenever the expectation does not meet the actual experience a disconfirm- -ation exists and evokes feelings as dissatisfaction, disappointment or surprise. However, the level of prior expectation does affect the experience which is due to cognitive processes, for example emotions and a desire for rewards. In Deliza and MacFie’s study (1996) two patterns of expectation effect were distin- -guished; contrast and assimilation. When the user is experiencing a difference between prior expectation and post perception, then contrast is defined as a bias that amplifies the difference; low expectation leads to a high perception. While assimilation on the other hand is a bias that diminishes the difference. If applying these expectation effects into an example with a vehicle’s size and the expectation of perceived interior space. What will happen when contrast occurs is that the user has a low expectation of the upper interior space due to a small car and get surprised when perceiving the space to be bigger than expected. Assimilation will then be when the user having high expectation of the interior space due to a big size car, and gets disappointed when perceiving it. Tolerance can also be linked with the expectation effect where in contrast tolerance will consequently increased and in assimilation decreased the perceived feature that satisfies product qualities as Must-be and Attractive. Sensory organs play the vital role in information intake for humans, the process of perception is also predominantly governed by multiple senses, so-called perceived qualities. These senses are what evoke feelings, impressions and emotions toward a specific matter (Yanagisawa & Miyazaki, 2019). However, in the book The enigma of perception (2013) Maclachlan mentions that vision and audition to be the dominating senses for perception and other senses can be used to confirm the initial perception through vision or audition. But this can change based on the environment in which the user is present and senses that provide more information are prioritized. From this discussion the perception of the upper interior space in a car is dominated by the visual sense; the eyes. The upper interior space inside the car is affected by the space around the head, which is constrained by the surrounding objects (features of the car interior). Also, theoretically as mentioned in (2.3.3), headroom is the measure of distance and it can be perceived by accessing the distance to this surrounding object. Perception of real spaces depends on a pictorial cues to provide information about the distance from the user. These cues specifically affecting size and distance are called depth cues. In an article discussing perception of hedonic attributes (Lageat, Czellar, & Laurent, 2003) depth cues are explained together with four important cues; occlusion, relative size, height in visual field and aerial perspective. These cues can be extended to the interior of the car to understand the perception of headroom. However studying these cues were not included in the thesis study since 15 2. Theoretical Framework it would be too big of a research scope if also addressing these parameters. In the experimental study done by Yang et al. (2015), the authors have aimed to improve perceived roominess by applying principles of optical illusions to specific interior car parts. As a result of their study, they have concluded that perspective, geometric, and concave/convex illusions can be used to improve the perceived roomi- -ness in the car. But the optical illusions have to be applied to the parameters that affect the headroom and these parameters can change with different cars. Hence, the thesis study constrained in identifying the parameters that affect the headroom and their subjective relation to the users. In a thesis study done previously at Volvo Cars (2015), Bentioulis and Forsberg conducted the evaluation of headroom in three different cars. As a result of their subjective study, it was concluded that the panoramic roof had improved the percep- -tion of the headroom. Important discussion is that the measured vertical headroom in the panoramic roof car was less than the car with no panoramic roof. However the presence of panoramic roof had positively improved the perceptive head roominess. The users felt that the presence of panoramic roof made the space airier and lighter and improved the roominess.Bentioulis and Forsberg have also said that it may be due to the reason that the panoramic roof had let more light inside the car, in the essence to improve the feeling of space. By this argument we can arrive at other important factors that affect the perception of roominess to be lighting and colour in the interior space of the car. Manav and Yener have also studied the effect of lighting in the spatial arrangement (1999). They discussed that the measure of illuminance inside the car affected the roominess feeling for the users. Since there were conclusive evidence from previous study and literature, it was decided to not include the panoramic roof in the study. The lighting was kept constant by performing the study in a controlled indoor environment. 16 3 Research Method The research approach of the thesis was divided into three phases. These phases required to adapt different research design approaches to arrive at the intended results of the phase. Some phase adopted only a single research design, and some used a combination of several research designs. The research designs, statistics related methods and the appropriate tools that were used are stated and explained in this chapter. 3.1 Qualitative Research Design Due to the new research area where there was a lack of theoretical paradigms describing the perceived roominess in a vehicle it was necessary to establish and develop a set of working theories. In the book Global Research Design (Darian-Smith & McCarty, 2017), it is stated that an exploratory research would then be the approach to be employed to produce useful knowledge. In another research paper the author relates this approach to a flexible design of how the data collection is gathered. The role of the researcher is described as an "instrument of data collection" hence to be skilled to absorb new information quickly, be adaptive and flexible to intervene with new interesting aspects or unexpected issues (Boeren, 2015). Qualitative methods are usually associated with a flexible design and are used in exploratory researches. In Check and Schutt’s book Research Methods in Education (Check & Schutt, 2011), qualitative methods are described to “capture educational reality” and where the subjectivity and the experience of the participants are the focus points. It is about understanding what the participants really felt or did at some point in time. Furthermore, focus groups, participant observations and intensive interviewing are activities referred to qualitative methods. Analyzing raw qualitative data usually means study words, expressions and texts, where there is no existing formula of how to transform the data into findings though unique for every inquirer. Small-scale of data and participants is enough to achieve desired results, hence qualitative data analysis should “focus on meanings rather than on quantifiable phenomena” and “collection of many data on a few cases” (Denzin & Lincoln, 2002). Phase 1 and 2; Awareness (chapter 4) and Identification (chapter 5), was aimed to gain knowledge and were both designed to be an exploratory research. However, it was in phase 2, Identification, where a qualitative study was conducted. The test 17 3. Research Method persons’ expressions and words were studied and were used to identify influential parameters. 3.2 Quantitative Research Design Compare to the qualitative data analysis quantitative analysis are more focused on the numbers rather than interpret figures and subjective matters. Quantitative research is utilized to confirm assumptions and theories, hence being theory driven (Bhandari, 2020). Hypotheses are commonly stated in advance and tested in a quantitative research design. It falls under methods that are used during explanatory researches to explain and predict existing theories (Darian-Smith & McCarty, 2017). The higher aim with the explanatory research would be to provide new insights and challenge existing theory. In Boeren’s research paper she presents that quantitative research design also falls under a fixed term where the data collection process would be considered to not be flexible for the researcher to intervene (Boeren, 2015). Furthermore, it is a strict design with a fixed approach with a probability sampling (random sampling). In order to present representativeness in the result, unbiased representation of the subject/population probability sampling is needed (further presented in section 3.2.1). It was also discussed that the probability sampling is not achievable with a small-scale survey, though require a large data collection with many respondents. Moreover, quantitative data analysis can also be associated with statistical techniques to determine variation in numerical measures as well relationships between two or more variables (Check & Schutt, 2011). Example of statistical analysis which can be done are cluster analysis, frequency distributions, measures of central tendency, correlation coefficient and reliability tests. Phase 3, Investigation, incorporated the use of a explanatory research design to do statistical analyses to investigate the formulated hypothesis and research questions. Significant number of test persons were involved in the study and the collected data was processed to arrive at results. 3.2.1 Random Sampling of Test Runs Considered to be an important sampling method is the simple random sample. In an experimental setting, a simple random sample of size n is obtained when items are selected from a fixed population or a process in a way that every group of items of size n has an equal chance of being selected as the sample (Mason, Gunst, & Hess, 2003). In addition to its use in the sampling of observations from a population, simple random sampling has another important application in the subject of scientific and engineering experiments. Among the more prominent uses of simple random sampling in experimental work are the selection of experimental units and the randomization of test runs. By adhering the concept of simple random sample, all the test persons were tested with a fixed set of test setups but in a randomized test run (Mason et al., 2003). The main goal of this test run sampling is to eliminate the effect of biasing and clustering. 18 3. Research Method The authors of journalHandbook of medical imaging (Van Metter, 2000), had discussed upon the importance of presenting the various test setups in random while conducting a subjective test. This random sampling of the test setups was stressed in order to avoid bias of the users during rating in subjective assessments. 3.3 Interview Approach Three commonly known interview approaches are structured, unstructured and semi-structured interviews which are used in different scenarios and research approach- -es. During this project a semi-structured interview has been conducted to extract important findings during the Identification phase. While a more structured interview approach was conducted in the third phase, Investigation, to fit the quantitative research design. What differentiate the interview approach is mainly the kind of questions that are to be asked during the interview. Closed-end and open- -ended questions are distinguished where closed-ended questions are used in structured interviews, also called standardized interviews. The questioner/interview is rather strict and not deviating between respondents and are used for larger sample groups. While open-ended questions are used to allow extended probing and where the interview is very flexible to be able to make the interview more personalized (Adams, 2015). Unstructured interviews are characterized with mainly open-ended questions. The advantage with this approach is that the interview becomes more casual and free- -flowing hence creating a relaxed environment for the interviewee and as well obtain honest answers with inductive reasoning (Boeren, 2015). Disadvantages will then be that it is a time-consuming approach where the interview takes time as well the analysing in the end. The approach does also require interviewer sophistication and techniques to be able to create an environment for the interviewee to open him-/herself up and keep the conversation within the topic (Adams, 2015). In one of Leech’s article where he presents different techniques when asking interview questions, he recommends the interviewer to be professional and generally knowledge- -able. But in some cases on the particular topic of the interview the interviewer should act less knowledgeable than the interviewee (Leech, 2002). Leading, presuming, loaded and double-barrelled questions are stated in the article to avoid. Furthermore, the interview and setting are aimed to be pleasant and neutral to make the respondent comfortable. Semi-structured interviews are following what is called an interview guide which is creating an outline of the planned topic where some questions are prepared (Adams, 2015). In this way it follows a bit of a structured interview, however in a semi-structured interview it is allowed to ask follow-up questions which makes it also following an unstructured interview approach (Leech, 2002). With this approach the data can be more easily compared between the respondents comparing to unstructured interviews and its data where objective comparison of data is not 19 3. Research Method possible (Pollock, 2019). 3.4 Measurement Scales For the three phases of the research approach appropriate tools were used. One such important tool was the scales. Different scales were used to measure the perception of upper interior space as in a feedback from the test participants. The scales also served as an input medium to receive intended data from the test persons that were later used in the analysis. The various scales that were used in the different phases of the research are explained in this chapter. 3.4.1 Visual Analogue Scale Visual Analogue Scale (VAS) is used to gather information about internal feelings, perceptions, or sensations that are difficult to measure. It is for example a very popular tool to use to assess pain intensity (Eliav & Gracely, 2008). The VAS takes a form of a linear line anchored at both ends with the two extremes of the attributes to be assessed. Several authors have discussed the length of the line used in VAS to be 100 mm as satisfactory (Eliav & Gracely, 2008) (Briggs & Closs, 1999). Hence the same length was applied to develop the VAS used in this evaluation during the Investigation phase. The line was then anchored with “Very narrow” on the left end and “Very spacious” on the other side (Figure 3.1). It was then used to indicate how the respondent was perceiving the upper interior space by placing a marker somewhere on the line. A ruler is commonly used to determine the score from where the marker is placed to the end of the line (Lazaridou, Elbaridi, Edwards, & Berde, 2018). Figure 3.1: The Visual Analogue Scale with anchoring words used in the Investigation phase Due to working with multidimensional experience it is an advantage to work with a VAS compared to a scale with predefined intervals (Lee & Kieckhefer, 1989). Another benefit of the VAS is that the scores have the qualities of ratio data and can be used in statistical studies(Correll, 2007). Furthermore, the VAS has large number of response categories when compared to other measuring scales, for example the Verbal Rating Scale. This means that it can avoid confusion of the choice of descriptive words and provides accurate data about the test person’s subjective opinion. The scale is sensitive to changes and is simple to use which are seen as beneficial properties (Caballero, Trugo, & Finglas, 2003). 20 3. Research Method One of the disadvantages with this method is that it is too simple for some people, making it difficult to find a correlated answer on the line translating a multidimensional experience (Briggs & Closs, 1999). Another troublesome characteristic is that the VAS is having top ceiling, which makes it difficult for a participant to document an even worse experience if the upper-end is already used (Correll, 2007). 3.4.2 Verbal Rating Scale Verbal rating scale (VRS) is another subjective method which is commonly used to measure pain as well attitude (Lazaridou et al., 2018) (Roberts, 1997). What is distinguishing VRS form other subjective scales is that descriptive adjectives are used to determine and describe different levels of what is called an item. Two different approaches are usually presented when designing a VRS, the Thurstone (Thurstone & Chave, 1929) and Likert (Likert, 1932). The difference between them is that Likert’s approach is a multi-item scale which refer to a statement or research question as an item and having a standardized way of measuring the respond to these statement with levels of agreement/approval and disagreement/disapproval (Bartikowski, Kamei, & Chandon, 2010). While Thurstone approach, a calibrated verbal rating scale, is a more extensive scale which is unique to every question, hence appropriate items needs to be determined to cover a full attitudinal continuum. What is called as a calibration phase is required for a Thurstone scale to not only determine appropriate items for indicating different levels but as well construct an equivalence of internal-level. Comparing Thurstone and Likert scale between each other, Likert’s scale is more commonly used in surveys due to the ease of use. But Thurstone is however a valid and reliable method that are supported with empirical data which Likert’s method is lacking (Likert, 1932) (Bartikowski et al., 2010). Furthermore, discussing the advantages and disadvantages, VRS are most used due to the high compliance rate while being easy to use, interpret and administer for different application and responder (Hjermstad et al., 2011). Disadvantages might then be the distinctive categories that can be rather discriminative for senses and being less sensitive to changes compare to a Visual Analouge Scale. Due to using descriptive words the VRS is dependent on the patient ability to read as well interpret the words. This might cause error and inconsistency in the result. 3.4.3 Numerical Rating Scale If comparing a Numerical Rating Scale (NRS) with the already described scales (VAS and VRS), NRS would be considered as the simplest scale method. With just having a series of numbers, written in ascending order, together with some anchoring descriptive words on both ends of the scale to represent the best and worst scenario, it is easy understood and administered as well having cross-cultural validity to it (Selvaratnam, Niere, & Zuluaga, 2009) (Roberts, 1997) (Brunelli et al., 2010). 21 3. Research Method By having eleven levels it can be compared to have the same sensitivity as a VAS and acknowledge small changes (Brunelli et al., 2010). Other advantages with NRS is that the method has a well-documented validity and correlates positively with other measures (Roberts, 1997). It can be used for all ages, from a young age as 5 years old to elderly. The main disadvantage with NRS is that it does not have ratio qualities, similar problems as VRS (Thurstone approach). In other words, if comparing the numerical interval between 2-4 and 7-9, even if the interval is numerically equal, it may not represent equivalent intervals in terms of scaling the intensity of the subjective feeling (Lazaridou et al., 2018) (Selvaratnam et al., 2009). 3.5 Paired Comparison According to Stone, Bleibaum, and Thomas, paired comparison is the first developed method to access preference (2012). Further described in Iijima et al.’s research paper (2020), a paired comparison method is considered to have a quality to identify the hierarchy of personal values hence comparing values to one another. While evaluating products, the test involves one or more pairs of products and the respondent may evaluate one or more pairs of samples. The test is extensively used in the area of test design, statistical analysis and mathematical models (Stone et al., 2012). Explained in Ljubuncic book Problem-Solving in High Performance Computing (Ljubuncic, 2015) the amount of binary comparisons of N elements would be determined by using following equation: x(N) = N(N − 1)/2 (3.1) Bradley and Terry discussed about the flexibility of this test procedure in subjective testing in their research paper (1952). By using this procedure, it can be assumed that the standards of judging is uniform, eliminating respond style bias while ranking the preferences in a comparative nature (Bradley & Terry, 1952). Due to this reason the method was used in the thesis study when studying the influence of various parameters affecting the upper interior space by ranking them in the order of influence in phase 2 and 3’s user test. Moreover, this method provided a structural test procedure considering the high number of parameters without making it difficult for the test person since they had to compare all of them mentally, assuming weighted score before providing the ranking to them. Using this method two parameters were compared at a given instance of time, for which the test person must give their preference based on the question "Which of these two options is more important". At the end of the test, every parameter was compared with each other and based on aggregation, the final ranked list of parameters was obtained. 22 3. Research Method 3.6 Analysis of Statistical Data A quantitative approach was taken for phase 3 that required processing of data to arrive at statistical results. The following section contains the methodology that was used to handle the data and to process it. Also, several statistical concepts were adopted that included predictive tests, hypothesis tests and use of standard coefficient. These are stated and explained in this part to create a clear view of how they were used and when they were used in the analysis of the data in phase 3. 3.6.1 Data Analysis Methodology Due to a large possession of data after the quantitative user study in the Investigation phase a structured methodology was required to handle the data. The data analysis process as explained by Huber in the book Data Analysis: What Can Be Learned from the Past 50 Years (2012) was adopted as the data analysis process. This procedure guided to handle the data and provided a step by step process to analyse it. Table 3.1: The data analysis process with the purpose of the steps Step Number Step Purpose 1 Identification To get familiar with the new dataset, quality control including identifying distributions, outliers, clusters and preliminary error checking 2 Error Checking Checking for errors including plausibility and consistency checking of data and the dataset 3 Modification Arriving at derived dataset, preparation of the data for the upcoming analysis and simulations 4 Comparison Comparison of data and a relevant model. In case of moderate sized samples, serves as signal to avoid over-interpretation of data 5 Modelling and Model Fitting Modelling of the data for the purpose to interface in data analysis packages 6 Simulation To create simulated data, prepare synthetic datasets and to analyse them, if required 7 What-If analysis Includes analysis that are alternatives based on theory, amended dataset or subset 8 Interpretation Quantification of conclusions based on theoritical values (P-values, significance levels, co-efficients, etc.) 9 Presentation of Conclusion Presentation of the results from the above mentioned steps in a chronological way that is more appropriate for the analysis work 23 3. Research Method The data analysis process as discussed by Huber is summarized in the below table 3.1 and the purpose of each steps are explained against it. It is also stated that the below-mentioned steps are more of a common approach and some steps can be revised or skipped based on the purpose of the data analysis. 3.6.2 Continuous Data Before doing any statistical data analysis an understanding of the collected data was needed to know what kind of analysis was able to be done. This also provides a basic understanding in choosing the test types that holds valid for the specific type of data. Numerical data is divided into being discrete or continuous data. The distinct difference of them is that discrete data is values that can be counted and can take only the possible or finite values, e.g. number of heads while tossing a coin for 10 times. While for continuous data, the possible values cannot be counted and can be described only as intervals on the real number line, e.g. the height of a person measured in mm (Rumsey-Johnson, 2011). Thus, the data collected during the user test in the Investigation phase (measuring Spatial distances and the VAS rating of the perceived upper interior space) fall under the data category of numerical continuous data. Hence be used in statistical analysis or tests that require the data to be numerical and continuous in nature. 3.6.3 P-value The probability value, P-value, is the measure of significance of the results obtained from hypothesis tests. All hypothesis tests use the P-value to determine the strength of the result to provide statistical evidence. In hypothesis test, the claim is considered as the null hypothesis (H0) and its validity is tested against the alternative hypothesis (H1) which is the opposite of the null hypothesis. The P-value takes up a number between 0-1, and determines which hypotheses stays true for the selected data (McCormick, Salcedo, & Poh, 2015). • For P-value less than 0.05 (<0.05) indicates statistically significant. It exists strong evidence against the null hypothesis (H0), hence be rejected. Therefore H1 will be accepted. • For P-value higher than 0.05 (>0.05) means not statistically significant. It exists strong evidence for the null hypothesis (H0), hence H0 can be claimed to be true and the H1 is rejected. The same interpretation of the P-value was used in all hypotheses testing and the same being followed in the hypothesis tests for this thesis. 24 3. Research Method 3.6.4 Anderson-Darling Test for Normality The Anderson-Darling normality test is used to determine the normality of the curve which fits a particular data set. This test gives the degree of approximation to which curve would represent a normal curve. The cumulative distribution function for the data set is numerically compared with that for a fitted normal curve. The inference that can be drawn from the Anderson-Darling test is based on the following hypotheses: • H0: Data follow a normal distribution • H1: Data do not follow a normal distribution For H0 to be true, the P-value obtained from the test should be greater than 0.05. This value of P (P>0.05) makes H0 true and denotes the curve follows normal distribution. When P-value is less than 0.05, the H0 becomes invalid and H1 is true, denoting the curve does not follows normal distribution. The main aim for doing this test is to check the distribution of the test population and compare it with the Swedish population. One another use is to understand the data distribution so that it can be used for statistical analysis that requires the data to be normally distributed. 3.6.5 2-sample T-Test The 2-sample t-test (classical) is a hypothesis test determining the difference of two population group’s data based on the 95% Confidence Interval (CI). The test considers standard deviation (StDev) of the population sample and standard error based on confidence intervals to calculate the t-value. Based on obtained t-value and CI, the P-value for the hypothesis testing is determined. The test can be used to differentiate two population groups based on mean value of a specific data, e.g. if the two population groups differ based on their body height. The hypotheses for the 2-sample t-test is: • H0: µ1-µ2 equal to 0 • H1: µ1-µ2 not equal to 0 Where µ1 and µ2 are the mean values of the data pertaining to the two population groups. For H0 to be true, the P-value obtained from the test should be greater than 0.05. This means that µ1 and µ2 are similar and there is no statistical evidence for the two mean values to be difference. On the other hand, if P<0.05 then H0 becomes invalid and H1 becomes true. This means that the two population groups have statistically different mean values. 25 3. Research Method However, to be able to run a 2-sample t-test the following stated assumptions need to be true for the tested data. 1. The data are continuous (not discrete). 2. The data follow the normal probability distribution. 3. The variances of the two populations are equal. (If not, the Aspin-Welch Unequal-Variance test should be used.) 4. The two samples are independent. There is no relationship between the individuals in one sample as compared to the other (as there is in the paired t-test). 5. Both samples are simple random samples from their respective populations. Each individual in the population has an equal probability of being selected in the sample. The data used in this project had satisfied all the assumptions (refer section 7.4.2), except the variances of the two population to be equal. For this, the variances had to be checked before proceeding with classical 2-sample t-test or the Aspin-Welch test should be used. But on default, Minitab-19 uses Aspin-Welch for all 2-sample t-tests for the results. When the assumptions for the classical 2-sample t-test are true, Welch’s t-test performs as well or nearly as well as the classical 2-sample t-test (Welch, 1938) (Welch, 1951). Hence, the 2-sample t test using minitab 19 ((2-sample t-test, 2017)) provides similar result as classical 2 sample t test and holds good in determining the difference of two population means, without checking the variances of the two population’s data. 3.6.6 Pearson Correlation Coefficient The Bravais–Pearson correlation coefficient, commonly known as the Pearson Cor- -relation Coefficient, is a parametric test that can measure the strength and the direction relationship between two variables. These variables are values measured on an interval, ratio, or absolute scale. Pearson denotes the linear correlation between the two variables and cannot be extended to establish non-linear correlations such as quadratic or orthogonal (Spearman’s Correlation). It must be noted that the two variables have to be continuous in order to use this test (Mendes, 2007). In the book Introduction to Statistics and Data Analysis (2016) Pearson correlation was explained with the help of scatter plots. With the discussion by the authors, Heumann and Schomaker, upon the coefficient values the inference is summarised in table 3.2. 26 3. Research Method Table 3.2: The inference for respective Pearson’s correlation coefficients S.No Pearson’s Correlation Coefficient, r Inference 1 Positive r Positive and increasing correlation 2 Negative r Negative and decreasing correlation 3 1 Perfect correlation 4 >0.7 High to very high correlation 5 0.5 - 0.7 Moderate to high correlation 6 <0.5 Low correlation to no correlation The values in table 3.2 was used to establish the correlation between various variables in this thesis work. 3.6.7 Crobach’s Alpha Test The data collected during the user test were both subjective (VAS scale) and objective (measurement of vertical, diagonal and lateral distance). Reuzel et.al have mentioned that both measures shoul