It is difficult to supply a useful answer, because the question, and some details of the situation, are so amorphous.
Let me first rephrase what I think your situation is; if parts of this are incorrect, my advice may not be very helpful: You wish to evaluate the performance of 3 workers, whose work supports that of 50 salespersons. When salespersons seek this support, or these services, they first come to worker A. When A has finished with whatever A does for the salespersons, the salespersons then go to worker B or C for the second installment of this support (or these services). [Aside: Possibly worker A directs salespersons to B or C as appropriate, in which case part of your evaluation may want to involve this aspect of A's work, as part of (or in addition to?) the 14 questions in the "first section".] 50 questionnaires have been distributed, presumably one each to the 50 salespersons (?). It may be relevant to know whether a salesperson ever encounters workers B and/or C only once, or more than once; and if more than once, whether you're asking them for separate information for each occasion; and if more than once, whether a given salesperson only ever encounters worker B (or C), or may sometimes encounter one, sometimes the other. You seem to state, however, that the salesperson is asked to name the worker in question; so that even if the salesperson is sometimes served by both workers, the responses are always with respect to one particular worker. To rephrase part of that last paragraph: are you dealing with a group of 50 salespersons, or with two groups of 25 (or so?), the groups identified by the worker who serves them? The 14 questions in each section of the questionnaire "are much alike". Does that mean that it is reasonable (and/or that you wish) to compare the services of A with more or less corresponding services of B and C? (I presume that the questions in the second section apply equally to both B anc C.) Now as to your purpose(s). (1) You may want merely to verify that the services offered are being offered adequately well: in which case some standard summary of each of the 28 questions would be in order. It is also possible, but you haven't provided evidence, that it is reasonable to cumulate responses over subsets of the 14 questions in each section for this purpose. (2) You may wish to identify particular areas of service in which to seek or encourage improvement. In that case, presumably the questions include specific items on these particular areas, and it may well be UNreasonable to think of cumulating responses over several questions. For this purpose, probably the only useful information is an analysis of the item-by-item responses. (3) You may wish to compare the performances of B and C in the second section, on particular services. Perhaps if C is offering superior service in the area covered by item 1, and B is not doing so well, you may want C to tutor B in that area. Or perhaps you have some other action in mind. (4) You may want to make a more global comparison between B and C. Now, the questionnaire itself: You have not described anything about the questionnaire except the number of items and that each item has 5 possible responses that you have coded 1 to 5. Without knowing the responses, it is impossible to offer useful advice on summarizing the information from these items, except in very general terms. For example, IF the responses are degrees of agreement with a statement of some kind (Strongly agree, Agree, No opinion, Disagree, Strongly disagree), it MAY be reasonable to treat the numbers 1 to 5 as though they could be averaged, and to report such averages (which may be done, with some clevereness, rather more succinctly than reporting that the five options were chosen by, respectively, 12, 18, 5, 8, and 4 persons, with 3 persons not responding, for example). BUT that may well depend on what the middle category is perceived as representing (which may differ from what you had in mind in framing the item). "No opinion" (and similar locutions) may mean a neutral opinion more or less midway between "Agree" and "Disagree"; but it may be used by a respondent to mean "I don't know anything about this topic, so I can't tell whether I agree or disagree", or "This item is irrelevant to my work, so my opinion is not useful", or "This topic was not dealt with in the interaction(s) I have had with worker B", and in these latter meanings the response "3" is much more like a "missing" response than a "neutral" response. OTOH, of course, the five options may be different kinds of responses altogether, in which case the whole last paragraph is virtually vacuous. Finally: You write "I do not expect men and women (or any other two groups) to have systematically different answers." Is this to be understood as meaning also "if there are any such differences I don't want to know about them"? I know, you thought you were asking a simple question. In a certain sense you were; but simple questions do not necessarily have simple answers, and the nature of the questions you want to ask of your data (as distinct from what you thought you wanted to ask of us!) is not, as I have tried to illustrate, very clear from your description so far. Good luck! -- DFB. On Thu, 1 Jan 2004, Uziel wrote (and on 13 Jan asked again): > I have 3 workers who serve about 50 saleswomen and salesmen in my > company. The first worker serves all of the 50 (First section). second > and third workers serves about 25 each, same service to different > saleswomen/men (Second section). "No relationship between the two > sections." > > My intention was to check (internal) customer satisfaction. I have > distributed 50 questionnaires with 28 questions, 14 for each section. > The saleswomen/men were asked to tell which worker (second or third) > serves them. > The questions in the two sections are much alike. They all with > scaled answers 1 - 5. > I do not "expect men and women (or any other two groups) to have > systematically different answers." > > Hope this clarifies my first post. > > Thank you again. > Uziel > > > > > > Stan Brown <[EMAIL PROTECTED]> wrote in message news:<[EMAIL PROTECTED]>... > > In article <[EMAIL PROTECTED]> in > > sci.stat.edu, Uziel wrote: > > >I have about 40 returned questionnaires, which contain 28 questions, > > >divided into two sections of 14 questions each. > > >Each Question has 1 to 5 scale answers. > > >Now, my question is, besides summing up: how many picked X answer in Y > > >question, > > >What kind of analysis can I perform? Is there a Tutorial/guide I can > > >find on the web? > > > > Meaning no disrespect, I think you're putting the cart before the > > horse. You're saying, in effect, "Here's a bunch of data; how do I > > analyze them?" The answer to that depends on what you're trying to > > find, and _that_ should be your first question, before even you > > start collecting data. > > > > So what are you trying to find?> Is there supposed to be some > > relationship between the two sections? Do you expect men and women > > (or any other two groups) to have systematically different answers? > > Is this some sort of attitude survey, and you want to make > > statements about the attitudes of the population your sample came > > from? > > > > Data analysis is a toolbox. Which tool you select depends on what > > you're trying to accomplish. So tell us more about _that_ and we may > > be able to make useful suggestions. > . > . > ================================================================= > Instructions for joining and leaving this list, remarks about the > problem of INAPPROPRIATE MESSAGES, and archives are available at: > . http://jse.stat.ncsu.edu/ . > ================================================================= > ----------------------------------------------------------------------- Donald F. Burrill [EMAIL PROTECTED] 56 Sebbins Pond Drive, Bedford, NH 03110 (603) 626-0816 . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
