Traditional statisticians would advise against calculating the mean on Likert-scale data, which is ordinal data. Numbers represent choices in Likert-scale data, such as strongly agree = 1 and moderately agree = 2. Ordinal data does not have equal distances between the numbers and therefore violates one of the assumptions necessary to use the mean as a measure of central tendency. The median or the mode are preferable. If you are looking for general trends and there is a large sample, however, such as in market research, calculating the mean is acceptable.

Assign a numeric value to each response. Response choices and number assignments could be strongly agree = 2, moderately agree = 1, neither agree nor disagree = 0, moderately disagree = (-1) and strongly disagree = (-2). The numbers could be coded in any way, the numbers here are just used as a symbol for the response --- however, coding with a middle value of zero makes results easier to interpret through graphs.

Add the total responses to each question. If six people responded yielding answers of (-2), 1, 1, 0, 0, and (-1), the total would be (-1).

Divide by the total number of answers: (-1)/6 = (-0.167).

Interpret results. As none of the response choices corresponds with (-0.167), there are two ways to interpret. You could say that most people responded "Neither Disagree or Agree", which is the rounding method, or you could give a range of value, such as most people responded either neutrally or with moderate disagreement.

#### Warning

Be careful when calculating the mean with ordinal data, such as Likert scales. If precision is needed the median or mode are better measures of central tendency.

#### Tips and warnings

- Be careful when calculating the mean with ordinal data, such as Likert scales. If precision is needed the median or mode are better measures of central tendency.

#### References

- " International Journal of Market Research"; Do Data Characteristics Change According to the Number of Scale Points Used?" An experiment using 5 point, 7 point and 10 point scale; John Dawes; 2008
- Quick MBA: Central Tendency
- Open Content: An Exploratory Study of the Statistical and Educational Implications of Violations of the Assumptions of Parametric Analysis Techniques; David Wiley, et al.
- C.Q. Press: Essential Statistics for Public Managers and Policy Analysts
- The Math Forum At Drexel: Finding a Mean in a Survey