Psychometric Foundation of KYKO Profiler
KYKO measures the intensity of five dimensions of human traits and to make comparisons between them as to which is dominant all the way to which is least intense. For valid comparisons, these five dimensions must necessarily be measured on the same scale. This is very much like comparing the heights of two people where valid comparison of height can only be made if they are both measured in centimeters or both in inches. If one is measured in inches and the other in centimeters, we will need to know the “equation” between centimeters and inches, convert one into the other, and then make the valid comparison. The idea of measurement is that there must be objectivity and hence it requires the construction of a variable on a linear scale. For the measurement of the KYKO dimensions, all five are equated on to the same logit scale using Rasch Model. Rasch Model puts each dimension on a logit scale, but to ensure that they are on the same logit scale; these dimensions should not be measured independently for each respondent. They must first be equated, through which each item on the five dimensions are calibrated on a single logit scale, and hence all person measures for the five dimensions are also on a common logit scale. This is the accuracy we need for KYKO interpretations and profiles to be more accurate. Note that raw scores are not linear. Depending on how a person obtains his or her raw score total for each dimension, whether or not they are from items that are easy to agree to or more difficult to agree to, will affect their final measure on that dimension.
While raw scores are not linear, they are the necessary and sufficient statistics for the construction of measures. To obtain the necessary raw scores, the usual ratings using Likert Scale is used. The Likert Scale used by KYKO, a seven -point scale. The number of points on the Likert Scale depends on whether it is possible for respondents to distinguish between the intensities of each adjacent pair of scale points. For example, if we use a nine-point scale, could the respondent tell the difference between 8 and 9? Or the difference between 6 and 7? If the answer is yes, then a nine-point scale should be used. The more the number of points on the scale the sharper the difference that can be detected between people. However, if the differences between adjacent pairs of scale points cannot be told apart meaningfully, we should reduce the number of scale points to five points, or even three points if that becomes necessary. For the instruments used by KYKO, the seven-point scale is found to be ideal.
For each dimension, we may use say, 15 or 20 items to measure the intensity of that dimension quite meaningfully. However, the more the number of items, the better the estimation of the measures of the respondents. That is, larger number of items will reduce the error of estimates. To reduce this standard error, KYKO finds it ideal to use 30 items per dimension resulting in a total of 150 items for the five dimensions. While items are constructed as carefully as possible to target the dimension concerned, we should be cognizant of the fact there may be items that do not really fit into that dimension. If this happens, Rasch analysis will be able to flag out the misfitting item through its report on the infit and outfit mean squares. Misfitting items are removed and order to have only “clean” items on that dimension remaining in the instrument. KYKO uses only clean items for the five dimensions with 30 items per dimension.
Prof Dr. Lee Ong Kim, Vice President of World Education Research Association