Conducting the Survey

A. Method

There are several survey methods which could be used. One is to mail questionnaires to a random sample or even all of the subscribers of your cable system. The main problem with this method is that you are dependent on people taking the time to fill out the questionnaire and return it. Response rates vary, but generally, you would be lucky to have even a 30 percent return. Further, given present postal rates, you may find the cost of mailing to be prohibitive.

Another is a random face-to-face survey conducted in people's homes. If properly done, this method can be the most accurate. Its main drawback is that it is very time consuming. Also, the interviewers must be very well trained. This method is best left for experienced researchers with funds to pay trained interviewers for their time and travel expenses.

The method advocated here is a random telephone survey. This method is comparatively easy to execute, can be very low cost, and the actual interviewing process need take no more than two or three evenings. Also, the response rate is usually very high with 80 percent to 90 percent about average.

B. The Sample

The ideal would be to survey every person in a home with a television set hooked up to cable and able to watch your community programming. Obviously, unless your system has only a few hundred subscribers, this would be impractical. Most statistical research is based on the theory of probability. That is, the data gathered from a random sample of a "universe" (in this case, local cable subscribers) would be about the same as if the entire universe had been polled. The important criterion in deriving a truly random sample is that each person in the universe must have an equal opportunity to be chosen.

The first step is to get access to a complete list of all cable subscribers from the cable company. Approach the local manager with caution. He or she may be reluctant to provide you with a subscriber list--and with good reason. That list could be a valuable commodity to anyone wishing to sell TV sets, repair services, cable program listing guides or who knows what else to subscribers. Indiscriminate dissemination of a subscriber list by an operator could mean irate subscribers and even some disconnects.

Clearly explain your reason for requesting the list and assure the operator that the list will be kept confidential. Explain also that you are not seeking names or addresses, but only phone numbers. (He or she may be more willing to give you just phone numbers.) Also, you might offer to include in your questionnaire questions the operator would like answered about subscribers, and share your results. Be sure to request the most current list available, since a few subscribers will be hooking up or disconnecting every month.

A few cable franchises contain a provision expressly prohibiting the operator from-releasing subscriber lists to anyone. In such instances, it would be necessary to obtain permission from your local government in order for the operator to release this information to you. If permission to use the subscriber list cannot be obtained, it is possible, although considerably more difficult, to derive your sample from the phone book as will be explained later in this section.

If you have permission to use the cable company's list, the next step is to derive a random sample from it. Actually, there are many ways to do this which will give you equally valid results. The method explained here was chosen for its simplicity. All you'll need is paper, pencil, a ruler or a piece of acetate, rudimentary mathematical skills, the attached random number table, and a finger.

To avoid a lengthy discussion of probability, an adequate number of phone numbers necessary for a reasonably accurate survey of this kind is about 400. Significantly fewer would raise the margin of error of your final results to an unacceptable level. More numbers would not improve your results enough to offset the time and effort necessary. It should be noted that accuracy is related to the size of the sample, not the number of subscribers from which the sample is derived. It does not matter whether the sample is taken from a 3,000 or 80,000 subscriber universe.

Most cable companies use some form of computerized billing. Therefore, its subscriber list exists either as a computer print-out or on microfiche cards. Let us assume, for example, that your list consists of 200 computer print-out pages. By dividing 400 (the desired number of phone numbers) by 200 pages, you find that you need two subscribers from each page.

Next, assume that there are about 20 subscribers listed on each page. By dividing 20 subscribers per page by two, you find that taking every tenth name will give you-the desired two names per page. This is called a "skip interval." In most cases, you will find that the number of subscribers needed per page and the skip interval will not be a convenient whole number, but rather a fraction (e.g., 1 7/8 or 9 3/4). If this proves to be the case, round up to the nearest whole number as rounding down would result in a smaller sample than you need.

Now you need to determine at what point on the page you will begin to apply your skip interval. Arbitrarily deciding to begin with the first name would eliminate the chance that the first nine names on the page would be chosen. You would be inflicting your own bias into the random selection process.

A solution is to utilize a random number table (supplied, see appendix) which is simply a computer print-out of random numbers. Apply the "blind-fold/finger" technique. Without looking, point your finger anywhere on the chart. Look at the first two digits of the number your finger is pointing to. If the two digits form a number that is less than your skip interval, you're home free. Simply count down that number of names on your print-out sheet. Then, proceed to count out your skip interval to get your list of phone numbers from that page. If your finger gave you a number equal or greater than your skip intervals, try again, and keep trying until you come up with a number that is less. If you do not have a random number chart available, an acceptable alternative would be the "pull the number out of a hat" technique.

Using our example, assume that your finger came up with the digits O and 7. Count down seven names and then 10 more names to arrive at the first name you will use for the survey. Jot down the phone number and continue counting until you have the two phone numbers you need from that page. You will probably run out of names on the list before you have the second phone number you need so simply go back to the top of the page and continue counting until you reach ten. If the printout does not give phone numbers, write the name down and be prepared to spend some time with a phone book later.

After you have gleaned the first page for names, proceed to the rest using the same starting point and the same skip interval. Since computer print-outs have equal spacing, try using a ruler to save time. Measure how many inches down the first column it takes to reach your starting point. Then measure how many inches it takes to achieve your skip interval. To further save time, you could make a template from a piece of clear acetate (or paper you can see through) the same size as the printout. Draw lines on the template at the appropriate spots, place the template over each page and copy down the phone numbers. Once you get going, the process is somewhat tedious but not difficult.

This procedure for compiling your list of phone numbers also applies to microfiche cards since each card is divided into many pages. You will find, however, that the cards are not as easy to work with as a computer print-out since you'll have to use the microfiche reader at the cable company's office. On the other hand, microfiche cards are often more current and provide additional information such as how long the subscriber has been hooked up.

If a subscriber list is not available to you, it is possible to derive your sample from the current phone book by adapting the selection method described above. Your sample size, however, must be increased significantly to ensure that the sample contains at least 400 actual cable subscribers. How much more depends, in part, on the subscriber penetration of the system (the ratio of actual subscribers to the total number of living units passed by the system). The operator should be able to tell you the system's penetration in the community. If, for example, the penetration of the system is about 50 percent, you will need at least twice as many phone numbers from the phone book; if it's about 25 percent, you will need at least four times as many. Actually, since most phone books reflect a much larger geographic area than that served by the cable system, this difference must also be accounted for by a further increase in the original sample. Thus, to survey 400 cable subscribers, you may need to actually survey a few thousand phone numbers! Obviously, access to a subscriber list will make the project a lot easier.

C. The Questionnaire

Nothing is more important to the success of your survey than the test instrument: the questionnaire. Preparing a questionnaire, like any writing, is more art than science. There are, however, some basic principles you should be aware of before you begin. First and foremost, keep it simple! That goes for any questionnaire, but especially for one to be used in a phone survey.

Begin by compiling a list of information you want to gather and questions you want answered. Then take a cold, hard look at the list and eliminate everything that is superfluous and redundant.

Next, distill the list into 10 to 20 short questions. Avoid, as much as possible, any wording which may bias the answer. For example: "Why do you like public access programming?" would be a poor question because it infers that the person does like access programming, when, in fact, he or she may not. A better question would be, "Can you recall any program you have seen on public access that you particularly like?" followed by the question, "Can you recall any program you have seen on public access they you particularly disliked?" By asking these two questions, you have avoided inserting your own bias and have given the person the opportunity to give both pro and con opinions.

Avoid double-barreled questions. For example, "Do you think public access is interesting and informative?" is really two questions which should be asked separately. Also, consider whether the respondent would have enough information to actually answer the question. "Why don't you watch public access?" would be a difficult question to answer for anyone not aware that viewing the channel is an option. Finally, avoid negative words in your questions. For example, what would a "yes" answer mean to a question such as "Don't you think access is a good idea?"

Now try to anticipate and list, if possible, the most common answers you expect for each question. These answers should be written below each respective question on the questionnaire. While these answers will not be given during the interviews, it will make it simpler for the interviewer to record each response. Also, include an "other" category with space to write down any answer(s) not anticipated. Also, assign a letter to each answer which will simplify tabulation of the results later.

Finally, write a brief introduction to be read by each interviewer at the beginning of each interview. This introduction should state clearly who is conducting the survey, that you're not selling anything. and, in vague terms, what the survey is about. If possible, it is helpful to find a "legitimate" organization such as the local cable commission, city council or university which would allow you to use its name as sponsor of the survey in the introduction as this can lend more credibility and, therefore, increase your response rate.

Two examples of access phone survey questionnaires successfully used in East Lansing, Michigan, are contained in the appendix. The first was used to obtain general information about community awareness of the access channel and services. The second was used to measure the audience of a well-publicized series of public affairs programs. However, as discussed in the introduction, audience surveys for individual programs should be limited to those for which a relatively large audience was expected.

D. Pre-test

Pre-testing is simply trying out your questionnaire under "battle conditions" before you actually conduct the survey. It is an opportunity to see which questions work and which need to be re-written or eliminated. You may also find that you need to add to or modify your lists of anticipated answers. To conduct the pre-test, simply take about twenty phone numbers randomly from the subscriber list which you have not selected for the actual survey, make the calls, and modify your questionnaire accordingly. Then go ahead and print up the number of questionnaires you will need for the actual survey.

E. Training the Interviewers

It is assumed that, as with most access projects, you'll be using volunteers. If you have 400 phone numbers to call, you'll need about 20 interviewers. Select people who can speak clearly, are dependable, and who will follow directions.

Ideally, the training session should take place the day before or even the same day you begin the survey. Every interviewer should attend. Begin by going through the questionnaire, question-by-question. Explain as clearly as possible the purpose of each question and emphasize that each must be read exactly as it is written--no ad-libbing.

Some questions may have more than one answer and therefore it is desirable for an interviewer to probe for additional responses. Such "probing," however, should not be done in such a way as to bias the answer. Let's use our earlier example, "can you recall any program you have seen on public access which you particularly liked?" If the person names one program, the interviewer could probe for additional responses by saying: "Is there any other program you recall?" The interviewers should be instructed that such probing should not in any way suggest an answer. An unacceptable probe would be: "Well, what about last week's coverage of the city council meeting?"

Above all, stress courtesy. An overly persistent or rude interviewer will not help your survey or public image. If a person does not wish to participate in the survey, simply say "thank you" and hang up. If the person is willing but busy at the moment, the interviewer should schedule a more convenient time and call back promptly. When making a "call back," an attempt should be made to try to talk to the same person who answered the phone when the call was first made. If there is no answer, the interviewer should again call later that evening or the following day, if necessary. Obviously, a few people will be working or out of town while the survey is being conducted. If, after a reasonable number of attempts, no one at the number can be reached, have the interviewer make an appropriate notation on the questionnaire and set it aside.

The interviewers should be instructed to question the first person who answers the phone. That goes for children, too, assuming the child is able to understand and respond to the questions. Young people should have the same opportunity to express their opinions about television and access as adults. Again, any subjective judgment as to who should or should not be interviewed must be avoided.

Be sure to instruct the interviewers to write down everything said in response to a question. At the end of each interview, the interviewer should go over the questionnaire to make sure the notes are legible and complete. Role playing is a very useful training technique. Have the interviewers pair up and give each other the questions. It's surprising how clear the process of conducting an interview becomes once you have tried it a few times.

At the end of the training session, give each of the interviewers their allotment of questionnaires and phone numbers. Be sure to give them a phone number to call in case they encounter problems or have questions. Make definite arrangements for the return of the questionnaires when they have completed the interviews.

Also, explain to them that you will be verifying their completed questionnaires as will be discussed later. Verification is not "distrust." Rather, verification is a standard procedure which is necessary to assure the accuracy of the survey. It is insurance against possible claims that you had "rigged" the results of the survey.

F. Conducting the Survey

Assuming that you have done a thorough job of training the interviewers, conducting the actual survey should be easy. Make sure to have someone who is familiar with the questionnaire near a phone to answer any last-minute questions the interviewers may have.

The process should take only two nights. At the end of the first evening, have each interviewer call in to report any problems and let you know how many surveys they were able to complete. Some interviewers are faster than others, so you may want to reassign questionnaires before the second night.

After the second evening, collect all questionnaires and separate those which were not completed. You may want to make further attempts to reach the "no answers" since the more completed questionnaires you have, the more accurate will be your results. An 80 percent response rate should be considered about minimum for a telephone survey. Such attempts should be made as soon as possible, preferably within a day or two, to avoid any subsequent events such as a favorable (or unfavorable) newspaper article which might alter the results.

G. Tabulating the Results

First, take the time to verify the accuracy of the completed questionnaires. Randomly select at least one questionnaire from each set returned by the interviewers, call back, and ask the questions again. Be sure to speak to the person who was interviewed the first time. If you find any major discrepancies between what was noted on the questionnaire and the answers given to you, select a few more questionnaires from that interviewer's set and re-administer. If you continue to find discrepancies, all you can do is throw out that interviewer's set of questionnaires. Hopefully, this will not be necessary. It is possible, however, that a well-meaning interviewer may have chosen to "improve" the results of the survey by making the answers more positive than they really were.

Next, go through each questionnaire and look for any which were improperly or illegibly filled out. If you find any which cannot be corrected by checking with the interviewer or calling back the number, throw these out also.

Now you'll need to establish categories for what you'll probably find to be a wide variety of answers which were not anticipated when the questionnaire was prepared. Write down all the "other" answers to each question and try to group them together. For example, if you had asked people to recall programs they had seen, you might have gotten answers like "bicycle show," "bicycle safety," and "how to ride a bicycle." All three answers may be referring to a tape programmed recently called "How to Ride a Bike Safely," and should be combined into one category.

In many cases, you'll have to make subjective judgments into what category an answer fits. Try to be as accurate as possible. If you find an answer that is too vague such as "political program" or "music," it is best to relegate it to an "other" category in your final results. If the answer is totally unintelligible, you may have to discount it entirely.

Finally, add up the number of responses, by category, to each question and convert the totals to percentages.

H. Applying the Results

Now that you have your data compiled, what does it mean? How can you use this information to improve your access channel as a community service? Obviously, if you do not apply what you've learned, you've spent a lot of time and effort for nothing.

Formulate specific recommendations based on your survey results. For instance, how good is overall awareness of public access in your community? In almost all cases, outreach efforts could be improved. Of those who are aware, what has been the most effective means of reaching them? Newspaper articles? Consider regular press releases about programming and access activities of special interest. Are most people finding out about access by just tuning in? Try focusing a camera on a card or better still, a message wheel at times when there is no programming so people will not find a blank channel when they turn to the channel.

Did your results indicate transmission or reception problems which are reducing viewership? What is the nature of these problems? Perhaps improved equipment and production workshops, better equipment maintenance, or even "time-base" correction is needed.

Above all, convert your recommendations into action. Perhaps in six months or a year you'll want to do the survey again to gauge the success of your efforts. Improvement is always possible. It is even more possible if you base your efforts on solid information.

Appendix A: Random Number Table

RANDOM NUMBER TABLE

07	54	38	68	69	23	39	77	54	52	84	19	77	61	76	20	53	88	52	61	87	57	33	05	08
08	49	11	13	18	79	61	36	71	62	61	53	36	68	96	01	30	25	71	53	39	41	54	97	11
77	53	58	66	95	02	06	79	52	28	23	02	74	14	95	09	60	35	25	41	39	21	25	74	29
40	06	32	93	26	38	82	93	48	95	96	16	86	52	98	86	43	70	03	52	52	96	39	100	09
37	52	27	40	23	81	38	85	11	15	76	02	39	05	45	92	62	30	17	66	45	79	19	32	68
63	65	71	57	62	54	76	33	19	69	67	23	06	02	28	47	06	61	58	85	11	39	15	13	28

64	39	03	06	61	82	53	38	95	44	17	55	07	45	84	15	41	76	01	99	37	23	56	35	56
86	24	85	85	18	57	21	95	60	63	62	63	60	23	67	67	69	41	50	03	68	14	42	64	69
07	76	80	40	71	81	80	66	82	94	01	05	67	56	23	50	86	00	75	40	94	46	60	94	88
85	81	29	15	45	07	19	82	04	35	10	69	56	83	62	17	84	45	22	72	78	41	29	55	77
62	32	79	09	86	81	55	51	100	13	37	56	82	08	14	95	80	13	67	03	06	16	97	40	14
100	78	01	58	86	65	80	71	92	91	85	73	28	26	43	14	86	97	72	46	17	84	74	78	85

47	03	55	22	36	46	80	47	20	22	72	68	78	03	60	53	93	61	23	44	96	01	100	25	40
84	28	65	35	75	71	48	16	83	24	67	24	80	37	89	83	38	23	44	25	63	77	06	25	29
35	09	84	52	34	81	80	99	16	46	48	89	05	60	19	66	19	70	14	75	62	78	09	85	21
91	41	15	55	04	57	76	26	10	39	29	05	10	07	30	13	31	93	97	33	59	75	35	10	13
31	60	91	78	43	05	35	61	37	57	29	14	81	65	68	50	28	60	69	06	06	79	03	71	36
06	11	34	42	10	11	45	71	48	98	57	70	49	22	91	12	03	80	67	14	89	42	51	60	86

01	54	92	30	79	06	53	78	72	86	67	80	51	26	51	09	37	62	98	16	89	34	89	82	21
72	82	19	33	54	67	07	36	56	39	07	28	59	08	84	48	50	79	78	33	91	100	46	34	78
04	19	01	01	74	14	36	81	13	41	18	74	63	43	29	90	52	06	37	92	66	94	71	59	11
13	89	01	60	91	59	04	57	16	11	05	70	36	33	02	12	61	72	16	46	98	84	59	70	19
47	83	70	53	28	01	55	24	52	53	41	54	21	11	57	79	47	64	53	85	29	70	94	87	11
60	80	14	32	17	52	74	90	50	76	46	27	68	78	71	92	12	31	80	28	47	64	48	36	26

100	67	88	53	83	50	16	88	46	75	65	16	76	94	56	59	80	69	19	09	24	57	95	93	18
98	36	01	31	80	06	08	57	48	43	50	24	54	54	78	17	54	16	10	08	57	35	22	81	19
84	74	55	25	13	42	54	19	57	71	47	24	01	57	26	82	43	38	17	03	83	91	29	80	20
87	42	31	10	22	43	30	81	56	48	43	45	20	13	84	68	11	59	46	96	47	02	71	98	85
15	26	15	53	42	29	71	84	46	12	02	10	07	57	67	32	35	100	82	44	80	47	70	50	84
33	04	81	55	30	11	48	47	94	60	66	99	11	85	81	38	99	16	56	22	13	23	34	15	23

93	97	06	35	55	62	93	89	76	76	89	49	32	13	32	08	41	01	44	52	25	10	63	69	88
99	68	65	50	54	74	80	11	66	56	97	34	20	29	44	53	19	44	36	51	57	18	05	94	34
37	82	63	69	11	31	42	14	77	40	07	91	52	76	69	50	80	16	21	02	78	90	34	76	43
51	83	15	78	80	89	81	80	43	92	27	80	60	87	37	89	10	54	42	25	24	58	05	69	91
32	14	89	81	61	13	74	44	39	57	99	72	03	26	05	32	28	44	38	70	00	76	19	13	45
67	32	89	45	00	13	70	38	39	40	23	35	64	50	21	13	29	44	60	04	40	70	71	10	75

92	40	86	86	60	77	37	10	26	58	88	35	75	35	71	96	32	75	92	70	76	58	20	28	61
71	70	16	30	98	40	51	53	30	33	52	48	80	56	36	25	62	70	00	12	20	82	16	62	49
76	76	97	47	08	28	77	41	95	85	81	10	41	68	61	28	71	98	01	95	99	62	62	84	52
55	09	36	03	89	18	32	51	68	84	32	04	53	63	68	46	87	61	30	91	85	98	100	23	83
58	25	28	62	93	33	42	18	29	42	72	80	58	90	58	74	37	09	53	02	48	05	32	90	48
56	29	83	48	52	90	52	68	36	69	03	40	90	16	07	49	29	54	63	21	32	25	09	12	38

83	54	68	21	16	80	63	66	88	83	29	75	33	48	82	86	85	55	89	01	86	74	03	11	19
77	40	67	84	45	16	25	73	89	97	91	36	69	43	31	87	12	49	28	77	05	84	52	73	76
58	04	18	38	29	95	51	94	86	43	82	08	80	77	34	58	53	01	57	76	78	87	56	72	16
66	95	93	13	40	19	16	46	72	16	100	28	87	96	66	75	08	73	13	80	91	12	100	72	17
08	14	70	17	50	13	10	36	51	87	51	71	25	40	50	24	36	89	80	53	11	62	34	18	14
25	10	97	40	19	14	81	92	22	88	71	74	61	79	99	93	56	99	33	68	91	36	48	73	86

Introduction