4 learning & growth
These are the 4 broader areas but it again depends on organization to organization. so once you do the job analysis & identify the KRA's of each person . it will become easier for you to identify KPI's
so first of all you need to do job analysis of the person's.i don't have any specific questionnaire set but i hope people at citeHR help you wid this
All the best
1st June 2005 From United States, Boston
I am happy that you are going to develop a fresh questinnaire. Hope, in its finished form you will post it for others use. Presuming that you know psychometry/questionnaire construction I am giving you the steps below.
Step 1- Collection of content.
Go to your possible population of study. Approach about 50 persons from among them. Hand over them a plain paper and ask them to make 3 positive statements and 3 negative statements on the theme you want to make the questionnaire. It does not matter if some of them could not write six statements. Collect all the papers.
Step2- Preparation of the preliminary questinnaire.
Convert the statements you have got to question forms. All qustions must be direct, easily communicable to the subjects, as short as possible, unambiguous and must elicit the answers you want. Normally questionnares do have yes and no categories. You should have at least 100 to 150 questions in this preliminary questionnaire. The written request to the subjects what you want them to do should be before the questions
Step3- Collection of data for standardisation.
Administer this questionnaire on a sizable sample and get the responses. The standardisation sample size is not prescribed but my experience shows that when the sample size goes above 200 the statistical values go on getting stabilized.
Step 4. Item variance
Tabulate the frequencies of responses in each category of responses for every item. If your response categories are 2 then find out the probability of responses in both the response categories, one becomes p and other q. pq gives you the item variance. In such cases the value of item variance varies from 0 to .25. If you have more than one response categories then go for quartile deviation to get the dispertion values.
Step5- First selection of items.
First you decide how many questions you want to keep in your final questonnare. Now go on selecting the questions starting from the highest value. Select some more than the number you have decided because in the next step you will eliminate some more questions. Step 4 and 5 are necessary eyeing on normal distribution of scores of the sample.
Step6-Tentative direction of scoring and internal consistency.
Suppose you have yes and no response categories. Then tentatively decide for which response you want ot give higher value and for which response you give want to give lower value, like 1 and 0. Now score each response and make a total. The total is the tentative score one gets as per your judgement. But this scre may not be true. Therefore, you compute item total correlation. If the response categories are two you may go for point bi-serial correlation. If there are more response categories you may go for pearson correlation. Now for each question you are getting one correlation. The item which is having the highest item total correlation is most consistent with the construct which you are measuring. If your judgment on the direction of scoring was correct then you will get all positive cerrelations. In case you get negative correlations in some items then change the direction of scring for those items and repeat step 6.Then go on selecting from higher to lower value items till you reach the requisite number of questions you have decided to retain in your final question.
If you are all the more serious about internal consistency then you can go for factor analysis in place of step 6.
Step7- Distribution and norms
Now get the distribution of the test scores. It will be a normal distribution. Then find out the percintile norms.
Step-8 Go to establish the reliabilty of your questionnaire.
Step-9 Go to establish the validity of the questionnaire.
Ask any other clarification if you need.
1st June 2005 From India, Delhi
Thank you for your assistance.
Dr.Jogeswar, from your input, I would gather the steps are applicable to all types of questionnaire? (guess, I should have paid more attention to statistic class) :oops:
Actually, I made a mistake, I should have clearly mentioned that I am to design a questionnaire to determine the Department KPIs. A standard format to assist the department heads in determining their KPIs and to be used as a guideline to ensure that they have all areas in their department covered when they are considering their KPIs.
:oops: I also like to apologise for a typo mistake, I actually meant, that I have NOT done this before, instead of I have done this before.
Thanks again for your input, I have found it as an additional information for my later project in preparing questionnaires for individual KPIs.
2nd June 2005 From Malaysia, Johor Bahru