1 Abstract This diploma thesis subjects on rating scales that measure respondent's value of the agreement with given statement. Particularly it focuses on a number of scale points and usage of mid-point and option "Don't know". Due to the fact, there is not an optimal number of points that scale should contain the author focuses on comparing scales of different length. This work includes research that uses four types of the questionnaire to test differences between odd and even scales of different lengths. The thesis shows that share of extreme points is based on scale length. With the use of shorter scales, respondents are more likely to choose extreme points. Based on respondents' answers, the difficulty to choose from scale options is similar in both even and odd scales. Option "Don't know" was mostly chosen by indecisive and poorly-informed respondents when the even scale with above-mentioned option was used. In the case of using an odd scale, the mid-point was chosen also mainly by indecisive respondents, so odd scales should be combined with option "Don't know" for better detection of neutrality. Keywords Rating scale, attitude measurement, odd scale, even scale, mid-point, number of scale points, Web survey, online questionnaire