Kuder Richardson's Reliability testing

During collection of data from the students of class V, I have noted their difficulty to understand discrepancy in their attitude from strongly agree to strongly disagree. As a result, I could not collect data following five point scale. Under such awkward position, I told them to respond yes or no. In this way, I found dichotomous responses.

Since the response pattern is dichotomous in nature, I used Kuder Richardson Reliability - 20.

Visit about the Kuder Richardson formula here:

from Wikipedia

The formula and calculation

I collected 226 data with 65 items

1. Polba Girls : 63
2. Shiakhala Girls : 50
3. Shiakhala Boys : 50
4. Polba Boys : 50
5. Vivekananda : 13

After item analysis using item -total correlation, I found only 40 items were significantly and positively correlated with total score.

And only 60 students out of 226 students responded to all the 40 items. Non response is due to unawareness of some infrastructures like school magazine, canteen, etc.

Finally I assessed reliability coefficient of whole questionnaire using only those 60 cases and found very high reliability, i.e., 0.89.

I am thankful to Gargi and Puskar for assisting me to understand different psychometric properties of the items of my questionnaire.

My next target will be to modify 25 items .