Mr. HYNES. I wouldn't agree with that. We don't suffer from a limitation to telephone homes. What we do, sir, you will find very explicitly in the front of the book. We are making a measurement based on telephone homes within 20 cities. It is simply what it says it is. And that is what we do.

Mr. Cox. Then, when it is compared to one of the national services, which I would assume would be either ARB or Nielsen, this may result in apparent discrepancies because of the difference in the procedure and the extent of the survey you undertake to make?

Mr. HYNES. Well, I would like to say, when it is compared by

whom?

Mr. Cox. By someone who criticizes what appears to be a marked discrepancy in the ratings. In other words, I referred to Mr. Seiler this matter of Nielsen's rating of "Fireside Theater" falling and yours rising.

Mr. HYNES. Who said that, though? This is a writer?

Mr. Cox. A writer in Time magazine.

Mr. HYNES. I don't believe, first of all, the comparison should be made. There is no reason to make a comparison between any of the services. That isn't the proof of whether we have good service or not, whether we come out the same way.

Mr. Cox. In other words, your service is purchased primarily because it rates network programing and it gives an immediate reponse? Mr. HYNES. And it does it under certain conditions.

Mr. Cox. Whereas the others are doing a nationwide survey and producing results at later dates; and they produce, therefore, a different thing that is measured?

Mr. HYNES. That is correct. They are measuring something different, they are doing it in a different way. There is no reason why anyone should ever compare even two national services, as far as I can see, if they are doing it with different techniques.

Mr. Cox. Now, it has been suggested that since both you and Nielsen rate essentially networks, or at least that is one important use that is made of the ratings-and I am afraid I can't even be as precise in telling you where I got this information as I was with respect to this other-but it is suggested that CBS does better than NBC on Trendex in comparison to Nielsen, because NBC has more stations and the fact that you use a measurement limited to 20 major markets causes them to lose their advantage in probable total viewing across the country. In other words, that you get a rating, which in some circles, at least, is used as a nationwide rating, and which seems to be disproportionately in favor of the other network, CBS.

Mr. HYNES. Well, there are an awful lot of "ifs," "ands," and "buts" in there.

The thing is that if you are going to compare NBC and CBS, use either Trendex or use Nielsen-use the same yardstick, don't use two different yardsticks and say one is better on one, and one is better on another. Quite possibly they are, and for very good reasons; for very demonstrable reasons they are different.

I don't know how to answer that question-one is better on Nielsen and one is better on Trendex. It is quite possible they are.

Mr. Cox. And therefore they should pick the service if it favors them.

Mr. HYNES. I think that would be very shortsighted. We are not selling a publicity report, we are selling research. They may be bad, and they might like to know. NBC might like to know why it is bad on the Nielsen and CBS might like to know why it is bad in 20 competitive markets. If I were running a network, I would. So I would probably buy both.

Mr. Cox. In the 20 markets, of course, the results would be affected by the whole job being done by the affiliates in those markets? Mr. HYNES. It is affected by many things, of course.

Mr. Cox. What percentage of the numbers called refuse to take the call, or don't answer?

Mr. HYNES. That is two questions. Let's talk about refusals, first of all. This would vary a bit with the time of day. I would say it never gets over five one-hundredths percent, probably, the refusals. The y are very low. They are a little higher later in the evening than they are at other times during the day; but basically and primarily the American people do not mind answering the phone. They love to think they are part of a national sample, that they are getting their opinion expressed.

The second question as to the number of "don't answers," that very definitely varies with the time of the day that you are calling. In the morning and during the day, the available homes in daytime may drop as low as 60 percent. In the evening, it may be up as high as 85 or 90 percent.

Mr. Cox. Those not answering are reported as not viewing?

Mr. HYNES. That is correct. If they are not home or awake, they couldn't be viewing.

Mr. Cox. Don't many of them say they are not watching, just to end the conversation? Or do the girls find they are usually able to complete whatever questions they are instructed and get a complete reply?

Mr. HYNES. An interview that starts, that gets through the first question, generally gets through all the questions, unless it is a very long survey. But speaking on this report, where the questions are short and concise and to the point; when they start they get through the whole survey. Usually, we get the refusal immediately, at the immediate outset of the call; and they run, as I said, about five onehundredths of 1 percent.

Mr. Cox. You don't add another name; she still calls only the four names when she gets a refusal?

Mr. HYNES. She is going to call 16 in 13 minutes.

Mr. Cox. You can't count the one that refuses to answer

Mr. HYNES. Oh, yes; it is a part of the sample.

Mr. Cox. You rate the first week in each month, as you indicated. What would be your comment, with respect particularly to network live shows, on the danger of presenting a particularly good show that week and employing extra advertising and promotion in order possibly to affect the rating that results?

Mr. HYNES. Well, you are asking me to theorize; but my theory would be that it is not nearly as great as people would like to believe it is.

First of all, we rate the first week of the month to produce this: report. We rate almost every night in the month, or at least some

time periods in every night of the month, to produce ratings on programs. The people who are going to buy shows, either advertising agencies or advertisers, are not naive enough to take one rating on any show, I don't think. Very often if an agency is pitched a show by a network, the network will contact us and we will survey the show-or the agency will contact us, we will survey the show, and the network will have no idea the survey is even going on.

I just don't feel that this problem exists. I don't think it is important. These people are not the newspapers that are writing about this thing; they are pretty astute individuals, spending a lot of money, and they are getting the information before they spend the money, I think.

Mr. Cox. In other words, you don't think that a regular weekly show-take the familiar example of Allen and Sullivan-would seek to get a particularly strong attraction?

Mr. HYNES. Yes, I do think it would do that; but I think purely for the promotion that they can get out of the higher rating, they might do it. I don't say they have done it, but they could do it, and it would get them a higher rating. But it wouldn't avail them anything. If Mr. Allen's agent goes to B.B.D. & O. to sell that show, they are not going to buy it on the basis of that one rating anyway. So I don't think it makes any difference whether he "hypos" that show or not; because the agency will probably come to us and we will do it for 3 or 4 weeks and find out more on the rating on the show before B.B.D. & O. will put their money in it.

Mr. Cox. One of the situations that has attracted most attention in connection with your service was the alleged exclusion from the Top Ten of the program "Twenty-one," because it never fell-for a period of at least 2 or 3 months in the first week, because spectaculars or something of that sort were in its time period. We have here and I am sure you are familiar with it-Jack Gould's comment in the New York Times of March 3 of last year.

Mr. HYNES. Yes.

Mr. Cox. Obviously the show is still on the air, so it wasn't put off by the fact it didn't get into your rating; but does this result in some misrepresentation, in view of the wide publicity that is given to the "Top Ten," you come up with, even though it is only what you say it is, but it is publicized as something else?

Mr. HYNES. Does it result in what, sir?

Mr. Cox. Is there a distortion, because some of these shows do not appear in the week in which you rate, and that therefore

Mr. HYNES. I wouldn't say so. We rate the first week of the month. We are publishing the Top Ten for that week. If a show isn't on the air, it can't be in the Top Ten.

Mr. Cox. It is published as March Trendex.

Mr. HYNES. It is published as March 1 through March 7 Trendex. Mr. Cox. By you?

Mr. HYNES. That is the way we release it for release.

Mr. Cox. Is that binding on people who use it?

Mr. HYNES. It should be, sir, and that is the only piece of material Trendex releases. Any other ratings you see in the trade press are not released by Trendex. The only thing we release to the press each month is the Top Ten for the first week of the month, but every one that gets a rating says-you know.

Mr. Cox. How about my discussion with Mr. Seiler on this matter of the margin of error in respect to this. Do you have, and do you publish with your reports, a table which shows-I haven't had a chance to examine the report-a table which shows this error factor for your ratings?

Mr. HYNES. No, sir; we do not.

Mr. Cox. Do you have one?

Mr. HYNES. Oh, yes.

Mr. Cox. What is it; is it on the same order as Mr. Seiler's?

Mr. HYNES. Actually, the report we sell-and the reports most of these gentlemen that appeared before you today_sell is sold to research people in the networks and the agencies. Before they buy our reports, they are completely conversant with our sample, our method, and the margin of error, if it exists in any one of the services that they buy; so that it isn't really necessary to publish it.

Mr. Cox. On your sample size, and assuming ratings within 4 percentage points in a range of 30, would you agree that if they are within 4 points it is not clearly demonstrated that one really led the

other?

Mr. HYNES. Not necessarily. I would like Dr. Deming to answer the question, though.

Mr. Cox. Dr. Deming, did you get the question? I asked Mr. Seiler, and I understood he agreed that if you had a rating of 32 and a rating of 28 that because of this error factor it is not clearly demonstrated that the one really exceeded the other at all, much less by 4 points.

Dr. DEMING. Well, it could be, I suppose. I am not just sure what survey you are speaking about, nor what the standard error of that

survey is.

Mr. Cox. Well, let's take yours. Suppose you had in your Top Ten a show that is given a rating of 32 and another with a rating of 28. I don't know whether this will make your Top Ten or not. But based on your sample and on your table for standard error, you had Show A rated at 32 and Show B rated at 28. Now, is that within a range such that actually Show B might have been viewed by more viewers than Show A?

Dr. DEMING. I don't believe anybody can answer that question. I don't know that the standard error is known. I do not know it. There has been a lot of talk about standard errors, but I am not sure what they are.

Mr. Cox. Your organization does have

Mr. HYNES. Dr. Deming isn't with our organization; he is here as a consultant, sir.

Mr. Cox. Could you supply us with a copy of your standard

Mr. HYNES. It is a standard standard error. It would be the same. Mr. ROGERS. It would be the same as anybody that uses approximately 1,000 homes used. It is a random sample, and it would be the same. If what they say is correct, it is correct for us, too.

Dr. DEMING. I am not sure that it is correct. I might point out that the size of the sample has very little to do with the precision or accuracy thereof. It depends on how it is drawn. That is what counts. Hence, I can't answer the question about any of them, because I do not know how they were drawn.

89567-59-pt. 7-9

Mr. Cox. Is there, in this field, a likelihood that this allowance for error increases as the rating in question is reduced?

Dr. DEMING. I think Mr. Nielsen gave the correct answer on that. That is if the percentage, or whatever you call it, rating goes up, the absolute error will also go up, but the relative error will go down. I think that would be true regardless of the method of sampling. I think that is a general statement that anyone is safe in making. I can't imagine any sampling plan that wouldn't work out that way.

Mr. Cox. I wonder if you could supply this information for us, to be sent in for the record. Could you take your survey for this month, and indicate the set count, the penetration used for each of the 20 cities; the total calls assigned to each of the 20 cities; the calls actually made; those completed and uncompleted; the number of interviewers employed in each of those cities; and the calculations, then, that are made to produce the rating which you finally come up with. Is that possible?

Mr. ROGERS. For all the programs?

Mr. HYNES. That is a lot of work. I don't know that it is necessary. Mr. Cox. Take one night.

Mr. ROGERS. We would be glad to.

Mr. Cox. Take Sunday night of this month's week.

Mr. HYNES. What would you get out of that, may I ask? Maybe I am out of order asking.

Mr. Cox. We would, I think, simply have a better understanding of how your system is run in relation to the others.

Mr. HYNES. The reason I asked the question was that I think there are other things that we could give you which would be of more interest to you; or would demonstrate, perhaps better, that the survey technique that we are using is a good survey technique and can reproduce itself, and so forth-which is an important feature of the survey, not whether it correlates with somebody else.

Mr. Cox. Well, supply that in your own way. We simply wanted a little more information about how the thing is put together, what the actual base in these different cities is.3

I referred in the discussion with Mr. Seiler to this TPI rating for Syracuse. That is not one of your cities; you don't operate there. But in terms of the comparison of methods, where they used personal coincidental alongside telephone coincidental in 50 percent of the samples selected, they got in some instances considerable parallelism between the two, but in other cases up to 5 or 6 percentage points difference. Do you, as the others have indicated, use methods other than your own to check the results you get?

Mr. HYNES. We do not.

Mr. Cox. You rely entirely on the telephone coincidental method? Mr. HYNES. That is correct. I think, and Dr. Deming could perhaps express this better than I, that the comparison between two methods is not as important as whether we can produce, or reproduce the same results using the same method in different samples.

Mr. Cox. You mean from month to month?

Mr. HYNES. Any time.

3 Certain materials were furnished later and inserted in the record on July 1, 1958. These are set forth below in an appendix at the end of the testimony for this day. However, certain of the information requested was not included.

Subsequently, on February 20, 1959, the tabulation was received and set forth at the end of the testimony for this day.