等高線圖的閱讀理解與眼動表現：地圖複雜度與讀者表現水準之影響

Abstract

等高線圖閱讀之所以困難，除了必須掌握等高線圖的表徵方式，同時還需進行空間視覺化與空間定位。傳統測驗只能得知其對錯，卻無法了解讀者「是怎麼讀圖的」。本研究以等高線圖判讀為主題，透過測驗表現、眼動追蹤來了解不同複雜度等高線圖與表現水準對讀者閱讀表現與眼動型態的影響，前者包含答對率、主觀難度與答題時間等行為，後者包含總凝視時間、總凝視次數、平均凝視時間、總跳移次數、圖區跳移距離、圖區跳移次數、第一次進入AOI延宕時間等七項眼動指標與熱區圖。本研究的有效樣本為82名成人，被隨機分派到高、中、低三種複雜度等高線圖的情境。在前兩個題組，所有受試者皆觀看搭配中複雜度等高線圖的試題，用以評估受試者的起始狀態；後五個題組，三種情境的受試者分別接受不同複雜度的等高線圖。研究材料的平均答對率為0.58，屬於中等難度；平均鑑別度為 0.37，題目區辨能力高低的表現為「良好」。共變數分析的結果顯示：在控制起始狀態後，複雜度與表現水準的交互作用全都不顯著，代表兩種自變項是獨立作用的。就等高線圖複雜度的主要效果而言，大部分的行為表現與眼動指標並未顯著，有兩個眼動指標達顯著，一為中複雜度在圖區跳移距離顯著長於另外兩組，該情境受試者能進行較大距離的跳移，反映其視覺搜尋的負擔較低；二為中複雜度在第一次進入AOI延宕時間顯著少於另外兩組，該情境可能提供適量的線索與提示，資訊負荷剛好，讓受試者可以快速聚焦在關鍵區域。考量三種複雜度操弄的差異不夠大，研究者也對未達顯著的多數指標比較三種情境的調節平均數，發現多數呈現一致趨勢：當地圖為高複雜度時，答對率最低、主觀難度最高、答題時間最長，而且總凝視時間、總凝視次數與總跳移次數最長或多，推測受試者投入更多認知資源於任務中。相對地，地圖為低複雜度時，答對率最高、平均答題時間最短，而且總凝視次數、總跳移次數、圖區跳移次數最少，顯示受試者投入的認知資源最少。研究者原本假設，中複雜度圖不像高複雜度圖資訊過載，也比低複雜度圖有更多線索進行判斷，因此會有最佳的表現，但是目前只有複雜度效果唯二顯著的圖區跳移距離與第一次進入AOI延宕時間支持此一假設。但其餘未達顯著的多數指標的調節平均數則比較支持低複雜度（而非中複雜度）的等高線圖最佳。就表現水準的主要效果而言，則多數達顯著。高表現水準者顯著地有較高的答對率、較長的總凝視時間、較多的總凝視次數與總跳移次數，且有較短的圖區跳移距離與較多的圖區跳移次數。顯示高表現水準者雖然答題正確率較高，但也投入較多的認知資源在解題上。本研究顯示地圖複雜度的確影響閱讀歷程，而眼動研究有助於等高線圖設計。本研究限制來自等高線圖複雜度的操弄不夠大，使得多數複雜度主要效果未達顯著，也無法有效回答複雜度是越低越好還是有中度複雜度較佳的問題，值得後續研究再加驗證。本研究確認讀圖者表現水準的影響，但未顯示與複雜度的交互作用；未來加大複雜度差異的研究中，有必要再次檢驗表現水準是否為複雜度效果的調節因子。Reading contour maps is often challenging, as it requires not only an understanding of the representational features of contour lines but also the ability to perform spatial visualization and spatial localization. Traditional tests assess only whether a response is correct or incorrect, offering little insight into how participants interpret the map. This study focused on contour map interpretation, utilizing both performance outcomes and eye-tracking data to examine the effects of map complexity and participant performance levels on reading behavior and eye movement patterns. Behavioral indicators included accuracy, subjective difficulty, and response time, while eye-tracking indicators included total fixation duration, total fixation count, average fixation duration, total saccade count, saccade distance within the map area, saccade count within the map area, time to first fixation in the area of interest (AOI), and heatmaps. The valid sample consisted of 82 adults randomly assigned to one of three contour map complexity conditions: high, medium, or low. In the first two sets, all participants viewed medium-complexity maps to assess their initial states. In the five subsequent sets, participants in each condition were exposed to maps with varying complexity levels. The average accuracy across test items was 0.58, indicating moderate difficulty, and the average item discrimination index was 0.37, which falls within the “good” range of item quality. ANCOVA results indicated that, after controlling for the initial state, the interaction between map complexity and performance level was not significant, suggesting that the two variables operated independently. Regarding the main effects of map complexity, most behavioral and eye-tracking indicators did not reach statistical significance. However, two eye-tracking indicators showed significant differences. First, participants in the medium-complexity condition had significantly longer saccade distances within the map area than those in the other two groups, suggesting a lower visual search load and more efficient map scanning. Second, they had a significantly shorter time to first fixation on the AOI, possibly due to the optimal number of cues provided by the map, which facilitated faster orientation toward key information. Given that the manipulation of map complexity may not have been sufficiently strong, adjusted means for non-significant indicators were also examined. A consistent trend emerged: under the high-complexity condition, participants had the lowest accuracy, highest subjective difficulty, longest response times, and the highest values for total fixation duration, fixation count, and saccade count—indicating greater cognitive effort. In contrast, under the low-complexity condition, participants demonstrated the highest accuracy, shortest response times, and the lowest fixation and saccade counts, suggesting minimal cognitive load. The original hypothesis was that medium-complexity maps would yield optimal performance, offering more cues than low-complexity maps but without the information overload of high-complexity maps. This hypothesis was supported by two significant indicators (saccade distance and time to first fixation in the AOI), though the pattern of adjusted means in the non-significant indicators appeared to favor the low-complexity condition. Regarding the main effects of performance level, most indicators reached significance. High-performing participants showed significantly higher accuracy, longer total fixation durations, more fixations and saccades, shorter saccade distances within the map area, and more saccades within the map area. These results suggest that while high performers are more accurate, they also invest more cognitive resources during the task. This study demonstrates the value of eye-tracking methods for improving contour map design. However, one limitation is that the manipulation of map complexity may not have been strong enough, resulting in mostly non-significant main effects. Therefore, the study could not fully determine whether low- or moderate-complexity maps yielded better outcomes. Future research should consider increasing the differences in map complexity to re-examine whether performance level moderates the effects of complexity.