Population aging has recently come to the fore as a prevalent social issue all over the world. The case of Korean Society is not an exception in the world-wide phenomenon. However, there are two distinct differences in Korea as compared with other countries. The first distinct difference is the intensity of aging as a result of a very low fertility rate (1.24 for Korea in 2010, 1.87 for France, 2.07 for U.S. 1.37 for Japan). According to the official population projections, the elderly population will reach 14% (aged society) of the national population in 2018, only 18 years after it first exceeded 7% (aging society) in 2000. It took France 115 years and the U.S. 73 years to go from an ageing society to an aged society; it took Japan 26 years for this change. Second, Korea has a traditional custom of adult children living with and supporting their old parents. As a result, despite of rapid population aging, the associated problems have not been frequently observed and consequently, the issue of aging has not received enough attention from the Korean society. However, recently, the trend of living with their old parents has been rapidly decreasing due to a low fertility rate, rapid industrialization, and the influence of western culture. According to "Social Survey, 2008", 55% of household heads in 1998 were living with their parents when one of their parents was still alive and about 40% in 2008. From these changes, it is clear that single and couple elderly people will proliferate who cannot receive support from their adult children. In this context, this thesis aims to understand the effects of household structure changes on elderly travel needs under variations in mobility and accessibility. In addition, this study focused on non-work activities rather than work-related activities because non-work activities account for the majority of elderly activities. To achieve the research goal, three main works were conducted as follows.

First, majority of previous findings reviewed in the literature review part were not from Korea due to few amounts of published studies on this field in Korea. In addition, it is expected that more Korean elderly people will live in single or couple households without their adult children. As such, this step first divided households into single, couple, and multi-generational households in order to look at the differences in elderly travel behavior by household structure based on the previous findings in the literature. As data source, two time period data, 2002 and 2006 travel survey data were chosen to look at the effects of household structure changes. Based on the previous findings from other countries, elderly travel behavior was analyzed and compared by household type, year, and age group using cross-tabulations and trip-based going-out models (binary logit). Although the time difference between two datasets is only 4 years which can be considered quite small, there were very clear differences in number of trips, meaning very rapid changes. Compared to the previous findings in the literature, most of them corresponded to those of this step in the context of Korea; however several differences were also found.

Second, it is generally agreed across accessibility-related studies that "greater accessibility leads to more travel". However, interestingly, some studies showed inconsistent findings with the common hypothesis. In addition, the previous step did not take into consideration the effects of other household members on elderly travel behavior. Therefore, this step focused on the intra-household interaction and the impact of accessibility on tour frequencies. First of all, utility-based accessibility measures were developed using 2006 household survey data and transportation network and O/D data from KTDC. The descriptive analysis section showed that accessibility is negatively related to maintenance tour frequency while it is positively related to discretionary tour frequency. To confirm this uncommon relationship, tour engagement models were developed using ordered probit model. This model also showed the same uncommon pattern.

Third, the final step started with the findings and limitations of the previous steps which were mainly caused by the existing dataset, thus an original survey was conducted based on the limitations of the existing dataset and also 5 hypotheses. The hypotheses were tested through two types of approaches: descriptive analysis and structural equation modeling. The descriptive analysis part examined the variations in the frequencies, accompanying behavior, and enjoyment level of elderly grocery shopping among elderly segments, mainly using basic cross-tabulations. The analysis showed many differences among elderly segments and many variables are strongly interrelated. In other words, elderly grocery shopping behavior is very complicated and many factors cannot be meaningfully be interpreted separately. Therefore, a structural equation modeling technique was adopted to statistically test the hypotheses because this technique has strengths in accommodating complex causal relationships among variables. First of all, using the entire sample, a full structural equation model was estimated to test the hypotheses. Although the model showed a quite good level of goodness-of-fit statistics and confirmed many hypotheses of this study, this model ignored the possibility that there might be two or more subpopulations in the dataset. In this sense, the entire sample was classified into 8 segment groups by 4 segment variables. 4 comparison groups were considered and all segment groups were specified in the same way for consistent comparison purpose. Through comparing the estimations results, several differences among segments were found which were not found in the baseline model.

Through those steps above, this study identified a number of general findings and also provided several interesting original findings. Key findings were presented below with the implications for both policy makers and practitioners.

First, it was found that accessibility do not always lead to more tours despite the easiness of reaching destinations due to lack of vertical movement consideration in accessibility measures, the impacts of intra-household interactions and different travel needs by activity type. This finding will contribute to better evaluating transportation planning projects or policies and more accurate travel-demand forecasting. Second, household structure change in Korea will increasingly force the elderly to travel for unpleasing activities rather than for pleasure. Three policy solutions for this expectation were suggested: proper policy establishment to encourage adult children to live near to their old parents, regulations for opening new discount stores near to small grocery stores, and more investment to make senior centers into more pleasant entertainment places. Third, there is clear trade-offs among shopping places due to the nature of relatively fixed grocery amount by household and this study explicitly explained the mechanism which has important implications, especially for transportation planners (e.g. more accurate travel-demand forecasting). Fourth, grocery shopping is a pleasing leisure activity to particular elderly segments to improve their quality of life better than neighborhood leisure activities which account for the majority of Korean elderly leisure activity time. This finding urges policy makers to give more careful attention to elderly grocery shopping behavior, especially for the elderly who are less mobile or live without their adult children.

韓国では、他国に類を見ない急速な少子化と高齢化が進行している。2010年の合計特殊出生率は1.23、高齢化率は2000年に7%を越え、わずか18年後の2018年に14%に達する予測である。さらに、子供世帯が親世帯と同居し日常生活を支援するという慣習の存在により、現時点では高齢化から生じる諸問題は潜在化し、韓国社会の関心も比較的低い。しかし、少子化に伴い、今後、子供世帯と同居する高齢者世帯の割合は急速に減少するため、高齢単身/夫婦世帯が急増することが予想される。以上の背景を踏まえて、本論文は、韓国ソウル都市圏を対象に、世帯構成およびモビリティとアクセシビリティが、高齢者の交通需要に与える影響を明らかにすることを目的とした。特に、高齢者の外出の多くを占める、通勤・業務以外の外出活動に焦点を当てた。

第一章で研究の背景と目的を述べた後、第二章で既往研究のレビューを行った。その結果、韓国では、世帯構成が高齢者の交通行動に与える影響を含めて、これまで高齢者の交通需要に関する研究はほとんど行われていないことが明らかとなった。そこで第三章では、既存の大規模交通調査データである、ソウル都市圏パーソン・トリップ調査データの分析により、世帯構成の差異による高齢者の交通行動の差異を明らかにした。また、2002年および2006年の二時点のパーソン・トリップ調査データを用いることで、世帯構成の変化の影響にも着目した。世帯構成、調査年次、年齢階層に着目したクロス集計および外出/非外出選択ロジット・モデル分析を通して、短期間にも拘らず二時点におけるトリップ数に差異が確認され、急速な変化が起こったことが示唆された。

次に、「より高いアクセシビリティは、より多くの移動をもたらす」という交通研究分野で一般的な仮説が、当てはまらない場合があることを示す研究の存在に着目した。また、前章の分析では、他の世帯員が高齢者の交通行動へ与える影響を考慮していなかった。そこで第四章では、世帯内相互作用とアクセシビリティが、外出ツアー頻度に与える影響に焦点を当てた。まず、2006年のソウル都市圏パーソン・トリップ調査データ、交通ネットワーク・データ、ODデータ等を用いて、効用ベースのアクセシビリティ指標を開発した。クロス集計および外出ツアー頻度を説明するオーダード・プロビット・モデル分析の結果、アクセシビリティは買い物等の義務的活動ツアー頻度と負の相関がある一方、娯楽等の自由目的ツアー頻度とは正の相関があることが確認された。

そして第五章において、既存のパーソン・トリップ調査データを用いて可能な分析の限界を認識し、ソウル郊外のニュータウンを対象に、高齢者の買い物行動の実態と意識に関する独自のアンケート調査を実施した。まず、クロス集計分析により、外出頻度、同伴行動、買い物に対する楽しみが、高齢者の属性により異なることを明らかにした。続いて、多数の変数間の複雑な因果関係の仮説を検証するために、構造方程式モデルを構築した。全サンプルを用いたモデル推定の後、自動車利用可能性、同伴外出、義理の娘の存在、子供世帯と同居といった4つの属性に基づき、それぞれ2グループに分割して推定したモデル間の比較を行った。その結果、グループ間で異なる影響を示す変数が複数存在した。

最後に第六章で、結論と今後の課題を整理した。以下に、主要な知見と交通計画/政策への示唆をまとめる。第一に、目的地への到達しやすさを表現するアクセシビリティ指標が、必ずしも外出ツアー頻度と一定の関係があるわけではないことである。既存のアクセシビリティ指標の計測において、自己申告制のアンケート調査による短距離トリップの報告漏れや、世帯員間の相互作用、活動目的別に異なる交通需要、トリップ・チェインや建物内移動などの要素を考慮していないこと等が一因であると想定される。この知見は、交通需要予測の精度向上だけではなく、より適切な交通計画/政策の評価に貢献するものと考えられる。第二に、韓国における世帯構成の変化が、高齢者にとって楽しみでない外出活動を強いる可能性があることである。この知見は、韓国政府が高齢者の生活の質を改善するためには、高齢者の交通安全だけではなく、世帯構成の変化の影響にも十分配慮する必要があることを示唆する。第三に、世帯単位での食料品・日用品の買い物量が比較的一定であるという性質から、大規模ディスカウント・ストア、在来市場、スーパーマーケットといった、異なる種類の買い物施設への外出頻度間にトレード・オフの関係が存在することである。この知見も、交通計画/政策および施設配置に関しても示唆を与えるものである。第四に、食料品・日用品の買い物は、特定の高齢者層にとっては楽しみな活動であり、韓国の高齢者の多くが従事する近隣での娯楽活動よりも生活の質を改善する可能性があることを示した点である。この知見は、特にモビリティが低く子供世帯と同居していない高齢者の買い物行動に、政府はより配慮する必要性があることを示唆する。

以上、本論文は、韓国における急速な高齢化に伴う世帯構成の変化や、モビリティおよびアクセシビリティが、高齢者の交通行動および生活の質に与える影響を、既存の大規模調査データおよび独自の調査データの分析を通して明らかにし、新規性の高い結論と有用性の高い知見が得られたものと判断される。よって本論文は博士(工学)の学位請求論文として合格と認められる。