Color is one of the most effective visual variables to encode information. It is pre-attentively processed and encodes a variety of information such as categorical, ordinal, quantitative but also semantic information. However, the effectiveness of color encodings is not sufficiently defined and research proposes controversial guidelines. This thesis bridges the gap between the controversies by a novel definition of effectiveness and provides evidence that the effectiveness depends on the analysis task that is performed with color. Current guidelines provide effective color encodings for single elementary analysis tasks. However, for solving real-world problems, in most practical applications, single elementary analysis tasks are not sufficient but need to be combined. This thesis proposes a set of novel quality metrics, design guidelines, and methods to design effective color encodings for combined analysis tasks. First, for encoding single dimensions, and second, for high-dimensional data relations. For this purpose, the thesis provides novel tools that guide novice and expert designers through the creation of effective colormaps and allow the exploration of the design space of color encodings. The visualization expert is integrated into the design process to incorporate his/her design requirements, which may depend on the application, culture, and aesthetics. Despite a well-designed color map, optical illusions still bias the perception at the first level of the analysis process. For instance, in visualizations contrast effects let pixels appear brighter if surrounded by a darker area. This distorts the encoded metric quantity of the data points significantly, and even if the analyst is aware of these perceptual issues, the visual cognition system is not able to compensate for these effects accurately. To overcome these issues, this thesis presents the first methodology and the first methods to compensate for physiological biases such as contrast effects. The methodology is based on perceptual metrics and color perception models that can also be adapted to an individual target user. Experiments with over 40 participants reveal that the technique doubles the accuracy of users comparing and reading color encoded data values. Further experiments show that the introduced personalized perception models significantly outperform existing perception models applied in contrast effect compensation. Thereby, this thesis provides a solution to the problem of contrasts effects in information visualization. However, the thesis also presents how contrast effects can be exploited and used to enhance visualizations. First, by boosting the visibility of important data points, or second, by increasing the readability of high-frequency visualizations such as network visualizations. All methods, introduced in this thesis, can be used in any application or image without adapting to the visualization itself. Therefore, the effectiveness of the methods is demonstrated in use cases and case studies of different domains.