Sponsor (Apr-June 1964)

e s.iiiu' person s rcpl) com. mis :ui inaccuracies, too. uls nuisi also be kepi on the ^position ol .ill such erroneous plies (except where the specific rating procedure is well known ml the records are accessible r stud) b\ MK( oi Us auditors, i . i .1 Dls( I osi Rl PROCI Dl RES I he second pari ol the adopted minium standards applies to disosuxe tiou each rating service lould report its surveys. l I ach rating report should in"concise desci iption" of the ethodologies used. I his should indefinition ol the sample, the chnique used to covei it. the area ivolvcd, the time slot and a statetent whether or not "weighting" en applied. I ach report must also menpa all known omissions, errors hd biases that might affect results. ; I urthei. each report must so cite an) deviations from standd procedures that might ei>lor te results tor example, that 20 inrviewers involved were working on leii first survey. 4 I he rate-of-c o o p e r a t i o n night (and won I must also be notI ", example, each report should te the number of households inall) selected -plus the number btuall) providing usable informaffl that was incorporated into the port i Hut it some usable informami was not used. that. too. should reported. ) 5. In a prominent place, each re >rt must compare its sample data ith comparable primarv -source such as households or indi duals) to show the degree to which i sample reall) does represent the rse" it is said to be measur -' i I hese are to be broken down L counties or reasonable count) OUpings.) Services that use the UDC sample over and over again in eir regularly-issued reports must •t the same data in each report, but \latc it onl) semi-annually. 6 (u (graphic areas surveyed iould be clearl) defined, with the on criteria given. Thus, if rea surveyed is Metropolitan 1 ork as defined by the I S ensus, it should be s(> recorded the report. Survevs executed fol a cific client shall clcailv show the repot t is special, not p.u t of a n lai syndicated service in fact the client must be named and the poll's loim.it be made clcailv dis tinguishable from that ol th< lai report B1 II I -IN I RROR I he BR( statement took | care to acknowledge thai audience measurement is subject "to main kinds oi error." Some, ol course, aie noil sim pling errors. I hese ma) result from the methodology used, the maniiei m which (he survey's conducted, oi — even more unpredictably sun pie non-cooperation or non re spouse. "However." the MR( announce menl explained, "even a true probability sample is likely, to include errors due to the operation ol chance in the selection ol (he sain pie.'" I he rize ol (his chance depends, among other things, upon the Size ol the sample. ( I he sampling research is subject to such "sampling error.") The sample variation that is due onl) to the si/e ol the sample m.i\ be expressed as "statistical tolerance" or "standard error." S. thus, each rating report should list, preferabl) on its front page, several keys i 1 i the standard error: (2) the formula used -to select the specific sample in the first place. (3) a chart or table that lists the statistical tolerances fot one and or two standard errors — in other words, a chart that shows just what these variations are (and what the) mean i when applied to typical items included in the report. It must also be pointed out that. JUS1 because estimates of sampling error have been shown, thai doesn't necessaril) mean that a piobahihtv sample design has been achieved ( ()\\ l-RIIM, I(> \ Kl POM 9. When a rating service converts basic raw data into a rating report, it must show all the "weighting" or data adjustments thai have been applied, along with the I sons for so doing I his information must be available to all users of said report. Ki I ach rat indicate the nun turns thai ate acqu to its stand. nd poM Such a minimum ma) diffei from seivk. to i I i methodolog the nunilvi ..| stations b ilk numbci ol horn dio m tv. I I w here i isucd on a regulai basis each rai must indicate the normal s.impL turn fot eacfa sin .. Vnd when the return is below normal i but not | low the required minimum), this. too. must be |*>mtcd out pr. abl) in a prominent pla it Rim R (I \R|| |( \||<>\> 12. Cross tabulations demographic and consumer information must b up. lined bv the minimum sample base requii W lien the sample tor one pel iod is inadequate lor reporting such information accurately, it ma) necessar) to combine the samples ol two. three oi more successive periods 13. It an) station has to "special, non-regular promotional techniques" that might hypo distort — its ratings, the rating service must also poinl thai out. 14. I he rating service must also publish an) other distorting influences that it is aware ol 1 1 might include unusual weather. tastrophes, political oi social .vents, or preemptions such as world ries, elections. Congressional h« ings even transmission failures In addition to the ab standards." whu pplicable to all rating servk ecific stand aids" will also be established HR( Since these will have to tailored individual!) eh spc eitu technique in they*H evolve onl) over a p. time. S that are expected to most helpful in developii C standards, h include questionnaires that some rating services have already filled out submitted to BR< participating I |N I \\l and COLI VM met! stud ■ pril 20 1964 35