Documentation

... OCT a9 '~1 03:57PM i:rG.S REID loFG MB ~.2 ~ CT-80/l -.·•.•• • • ·• • ·-.·•.•• ••·•···.·a••• COMPETITION TRIBUNAL Jll TBB IA.T'l"Elt o~ an applicatio'n F TRIBUNAL OE LA CONCURRENCE P br th• Director of Investigation I R and Research for or4•r• pursuant ~ j ~ to •ection 12 of the Competition OCT l 0 1991 \)) ac.t 1.s.c. 1915, a. c-3~, •• amende~; t T REGISmAR- REGISTRAIRE MID Ill !BB llAftER of the direct and ! inc1£roct aotui•i~ion1 by South•~ Inc. OTTAWA, Ot-rr. ;J.t IS /p (µ) of eQulty interasts in the businesses of publi•hin9 '!'he vangouyer courter, tbe lortb Shore Bewa and the Beal 11t1t• Jlaekly a B T W • K •: TD DIDC"1'01t 01' IllVBS'lIGATIOB AID> aRSKUCll, COMPETITION TRIBUNAL Applicant, TRIBUNAL OE ~ CONURRENCE

File No. (];I ~ tfV I ... •n.4 ... ~'~ Bout:Jlua Inc., Lower X.inl&JM! ··.,ibll No. tt tf; Jr, : J 6 : O o l'Uhli1hing Lt4., am Pu.blisbiDO ,.:::• pi/A1J:t;:" %nc., Yellow Ceclar Propertie• Lta., 3' I q ~ I llortla Sbore rre Press Lt4 •• Speeialty Publisher Inc. , De~;6e la ~' ..Q.u Elt7 Pub11cat1oas Ltd •• RegiJtrar / ..,.. -Gretlitr I Respondents. AFFIDAVIT 8LAICB, CASSBL& ~ CRAYOOH aaz 25, ~•tree Court West ~oronto, ODt:azio llSL 1A9 COWUl•l t:o the :Respondent• G.F. Leslie J.3. Qu:lnn •. l'inkel.Jlt:eiii Telepbona llo. (41&) 8'3-2672 Facsillile ao. ('16) 163-2653 OCT 9 '91 13: 58 204 942 5669 PAGE.002 .. '.',• .. 4 /-~b .• ., • • ,;;..."'','}~, '~ ·.!.-. •· ::~· -4

-·------oc---r-09-,.. .9.. -1 · 03:57P11 i=tGJS REID 1oFG MB P.3 CT-90/l IB '.l'llS D.ft'BR of an appliaatioa by the Dlreetor of InYestit•tion •!Mi ...e arcll for orders pursuant to section 12 of the CQ'ID•titipp Act. a.s.c. 1915, c.C-34. •• .-eadedJ AID IB 'ml llATD!R of tbe 4lrect U4 iatliract acpisitt.011.& hy lont.llm Ina. of ecJUlt:lF interests in tile buaine•ses ot •ublishing %DA Y•nctmY•r Courier, tbQ Marth &bore lltllrl an4 the lto•l Bst•t• MIA'~ B 1!! T W B B 9: TSE DIREC'l'OR 01' l:llRSTIGA'l"J:O• AJID RESEARCll .&Dplic~t - and -SOtmu.11 J:JIC • , 10'CR laDII.iUID lfUBLISllDIC L'l'D • , 2I11 PUBLISHIJIG nrc., nr.r.cw CBDNt l'llOPDTIBS LTD., BOJml IBOP l'.IBB nus LTD. , SPBCIAL'lY PUBLISllBlt DIC. , BLT!' PUBLICATIOllS L'l'D • ..a DGIMlea.ta A!'FIDAV%T OP ~ um I, Angua Reid, of the City of Winnipeg, in the Province of Manitoba, MAD OATH AHD SA'Y:

1. Since 1980 I he~e been the Pre11dant of the An;us Reid Group Ino., in Wi:Anip•g, Kanltoba. Formed in late 19?9, Angus Reid A&sociataa has grown to become one of Canada's largest marketing/social r••••rch firms,

2. On August 15 of thi• year I •wore aD Aff14av1t in this proceeding to which was attached a tru• copy of a ~•port on the results of a retail adve~ti1ar •urvey conducted for Southam Inc. accompanied by eopies cf the •urvey questionnaire.

OCT 9 '91 13: 58

204 942 5669 PAGE.003

OCT 09 '91 03:5E1PM Fr'G..15 REID 1oFG MB • ; : 11 ••• P.4 ·. - 2 3, I have since prepared a mupplernental report to clarify the methodoloe17 used to conduct the retail advertiser survey, to foe~• IDJ' discu11ion of the implications of • price increase in either the daily or conaunity newspaper and to correct errors in the data preaented. Attached as Ezhi~it •A• to thi• lllY Af !iaavit is a true copy of lf'l';f supplemental report.

Sworn before a&e· at the ) City of Winnipe~ in the ) Provi~~ cf Manitoba ) this ::t:!Jt.9ay of October 1991.) llGUS DID AC mm A co~ FOR OATHS tMJr/llJ9f.ff~ IBe flt\~~~ MV COM~ EY.PmES dl~-""'lj.J.,..._f....,~~--OCT 9 '91 13:58 204 942 5669 PAGE.004

-----oc~r....,0=9-·=9~1--=:.03i 5EPM FrG.J5 . REID loFG MB P.5. :'I I ~' ~ ' ' Thia ia .Bzhihit •A• to the Affidavit ot Angu1 ae1~, Sworn before • on t:he ~ day Of October 1 19 9 l.

AlllODS UID October 9f/_, 1991

~D/23 OCT 9 '91 13:59

204 942 5669 PAGE.005

-----O-CT -09 -'91- 03:58PM i:lGS REID loFG MB P.6 SUPPLEMENT TO 'IHE RETAIL ADVERTISERS STUDY

PREPARED FOR: PREPARED BY:

OCT 9 '91 13: 59

Blake, Cassels & Graydon Angus Reid Group, Inc. October 9, 1991

204 942 5669 PAGE.006

OCT 09 '91 03: 59PM i:rQJ5 REID loFG MB P.7 INTRODUCTION This is a supplement to the report of survey findings tabled on August 15, 1991 ("the August 15 reportj.

PARTl This part contains a detailed review of the methodology employed to complete the survey of retail advertisers in the Greater Vancouver area commissioned by Blake, Cassels & Graydon and undertaken by the Angus Reid Group.

PART II. This part contains a further analysis of the switching behaviour of advertisers.

PARTIU This part contains corrections to the repon due to an error in the tabulation of Tables 3 through 9 of the August 15 report.

&718-01 OCT 9 '91 13: 59

204 942 5669 PAGE.007

OCT 09 '91 03:59PM i:H:;lJ5 REID Ir.PG MB P.8 PART L DETAILED METHODOLOGY 1.0 SAMPLING FRAME For every survey, it is important to have a sampling frame which is as representative as possible of the overall target population for the study. In constructing the sampling frame for this study, a conscious effort was made to avoid some of the common and less eomplete list sources which might be considered candidates for the sampling frame, such as Yellow Pages lists and direct marketing lists which can be purchased from commercial list vendors. Instead, a multi--pbased approach was used to amstruct the most complete and representative sampling frame possible. The first phase in constructing the frame was to build a list based on the names of retailers who advertised in the Pacific Press newspapers and the two community newspapers targeted in the study - The Nortb Shore News and .Ih£ Courier. The second phase of this approach to constructing a 5alllpling frame consisted of a "random walk" enumeration of retail establishments within the circulation areas of Tu Courier and The North Shore News, which approximate their trade areas. The Angus Reid Group viewed the random walk enumeration as an essential component of constructing the sampling frame for two rea.IDns. First, this enumeration was necessary to provide a source of retail advertisers which was representative of advertisers who had not advertised in the newspapers targeted in the study. Second, the enumeration was nee~ to determine the relative sample sizes for four distinct sub-strata. The sub-strata are bued on the newspaper advertising purchasing behaviour of retail advertisers and were defined as follows:

1. Retailers who have advertised onty in daily newspapers, 2. Retailers who have advertised only in community newspapers. 3. Those who advertised in both community and daily nms.papers. and 4. Those who have advertised in neither comrounfty nor daily newspmers. &7iiPJi hid OCT 9 '91 14: 00 204 942 5669 PAGE.008

OCT 09 '91 03:59PM R'G..l5 REID i.FG.MB P.9 It was important to measure the relative size of each of these sub-6trata and to employ sample weighting in order to give appropriate weight to each sub-strata in the calculation the survey results.

1.1 NEWSPAPER ADVERTISERS UST Source lists for the first three sample sub-strata were obtained from lists of advertisers from the files of both Pacific Press and the two community newspapers. A full list of all companies who have advertised in the two dallies and the two community newspapers was compiled and a special matching program was written to determine which retail advertisers advertise only in the daily, which advertise only in the community newspapers and which advertise in both.

A computer matching program was used to unduplicate the list of advertisers and, in so doing, identify which of the three sample sub.strata they belong to. This unduplication process identified a total of 559 retailers who advertised both in the dai1y and the community newspapers, 2.100 advertisers who advertise only in the community newspapers, and 7,000 advertisers who advertise only in the daily newspapers.

The unduplication process proceeded as follows. A comprehensive list of newspaper advertisers was compiled by merging the advertiser lists from the two community papers and Pacific Press. The file was sorted alphabetically by advertiser name so that duplicate names would tend to be placed adjacent to each other on the list. The list was scanned and any apparent duplicates were eliminated. Next, the file was sorted alphabetically on the largest word in the advertisers name so that again duplicate advertiser names would tend to sort adjacently on the list Again, the list was scanned and &DY apparent duplicates were removed. This procedure of sorting alphabetically on the largest word was repeated for the second largest word and then the third largest word on the file. Once again, these soned files were scanned and apparent duplicates were removed. At each step, the number of duplicates diminished, until on the last sort only 1 percent duplication was detected. This provided good evidence that very few duplicates remain on the list.

OCT 9 '91 14: 00

hp3 204 942 5669 PAGE.009

OCT 09 >91 04:121aPM f:IG..E REID i.FG MB P.10 12 RANDOM WAJ.X FIElD ENUMERATION A "random wane.- field enumeration was conducted within the circulation areas of .!bl Courier and The Nortb Shore New$, This was undertaken primarily to supplement tlle newspapers' advertiser lists with retailers who advertise in vehicles other than the newspapers. This enumeration also served the purpose of determining the correct proportions in the advertisers marketplace for each of the four sub-strata defined above. We provide here the details of the sample selection and field enumeration procedures.

First. a complete and up-ttrdate set of detailed zoning maps and by-laws was obtained from municipal authorities for the Oty of Vancouver, the Cty of North Vancouver, the District of North Vancouver, and West Vancouver. On the basis of the set of all maps and by-laws J which corresponded to the circulation areas of ]'be Courier and The North Shore News. a ),f two-stage cluster sample was selected as follows:

1. We identified all blocks which had at least some commercial retail wning. / 2. Primary sampling units were defined which consisted of a geographically contiguous grouping of blocks such that each primary sampling unit bad roughly the same number of blocks zoned for commercial retail purposes.~ These primary sampling units typically consisted of one or more zoning maps. 3. A simple random sample of primary sampling units was selected using a computer random number generator. 4. Within each selected primary sampling unit, the secondary sampling units were defined as the blocks which had at least some commercial retail zoning. 5. With the assistance of random number tables, a simple random sample of secondary sampling units {that is, bloclc:s) was selected within each prinwy sampling unit 6. Angus Reid Group interviewers were provided with copies of the zoning maps with the selected blocks highlighted and were sent on foot to cnwnerate each selected block on their map. ..... . OCT 9 '91 14:01 204 942 5669 PAGE.010

OCT 09 '91 04:3FM i:rG.E REID loFG MB P.11 7. Angus Reid Group interviewers visited these bloclcs during normal business hours and canvassed the entire block, conducting a brief enumerative interview at each commercial establishment to determine whether the establishment bad any retail operations, whether the business bought any advertising during the previous two years for the purposes or promoting retail sales, and if so, what types of advertising had been purchased.

8. This sampling procedure produced a total of 500 enumerative interview~ with retailers in the circulation areas of the two community newspapers.

Training for the field enumeration consisted of a one hour formal training session of the field enumeration staff. The training session covered enumeration procedures, and stressed the importance of avoiding potential enumeration undercoverage due to missing units of the target population of retail establishments. Interviewers were instructed to enter every doorway or opening in building facades on the selected blocks, in order to avoid missing potential retail businesses.

In fact, undercoverage was virtually non-existent in the field enumeration since interviewers reported there was little difficulty in locating retail business estabHshments because they are so highly visible from the street (it is in their best interest to be so), and because the enumeration questionnaire was very short and unambiguous. There was no instance of a business which was unable to readily identify whether or not they had a product or service for sale. In instances where respondents were unable to respond to one or more questions on the questionnaire concerning the advertising behaviour of their business, a telephone callback was completed to obtain the missing information. The refusal rate for the random walk field enumeration was 11.S percent.

6"71&-01 OCT 9 '91 14:01

hje3 204 942 5669 PRGE.011

0 (") -I

TABLE 1 CD -CD - SURVEY POPULATION SAMPLE PROPORTION :.. cg Daily and Community NeW8))8per N Advertisers (sub-strata 1,2,3) 7j~ Neither Daily Nor Community Advertisers (sub-stratum 4) 25~ T.ABL:El SAMPLE SAMPLE SUB-sTRATUM COUNT PROPORTION 1. Daily Only 153 26"1 2. Community Only 137 23% 3. Both Daily end Community 153 26% 4. Neitller DaJly Nor Conununity Ul 25'. "cg' :.. CD • Note: This population count was estimated based on the relatiw proportions of newspaper advertisen versus non-neWIPIPfJr advertisen :.. measured in the random walk enume1'tio~ and the known population counts for newspaper advertistn. "' UI m m CD ""O D Ci) rn ­cg N

POPULATION PROPORTION STEP ONE WEIGHr s~ 0.710 ..... 47% l.165 ~ ~ ~ I::! ~ POPULATION POPULAllON SIEPTWO COUNT PROPORTION WEIGHT ffi 7049 38' 1.4802 2105 11% 0.493S SS9 3• 0.1 J74 8613• "'"" 1.86l0 ,, ..... N

OCT 09 '91 04:01PM AGE REID liFG MB P.13 2.0 TELEPHONE SURVEY OF RETAIL ADVERTISERS A telephone survey was conducted with 594 retail advertisers from the sample frame. Telephone interviews were conducted between June 26th and July 22nd, 1991.

2.1 SAMPLE DESIGN AND SAMPLE WEIGHTING A random sample was drawn from each of the four sub-strata on the sampling frame. This sample was pursued exhaustively in order to obtain interviews with the selected retail advertisers. In attempting to reach the person responsible for making decisions about purchasing advertising and promotions, Angus Reid Group interviewers had to make frequent aillbacks and were often referred to another telephone number or business location (such as a head office) where th ts person could be reached. In some cases, this pursuit extended over several weeks, but every possible effort was made to interview as many of the selected advertisers as possible.

The sample plan was designed to yield ISO retail advertisers in each sub-stratum. The actual sample yields are displayed in Table 2. Sampling weights were applied to the results of the survey in order that each sub-stratum be represented in proportion to its actual representation in the overall population of retail advertisers. This was achieved using a two-step weighting scheme.

Step One of the weighting scheme consisted of applying a weighting factor to ensure the appropriate relative weight was given to non-newspaper advertisers (i.e. advertisers in the Neither sub-stratum) relative to advertisers in the three newspaper adVcrtfser sub-strata (Daily only. Community only, and Both). As discussed in TabJe l, the Step One weights were simply calculated as the ratio of the population proportion to the sample proportion for each of these advertiser sub-populations. These population proportions were estimated from the random walk enumeration. The population of newspaper advertisers on the

6"718"0i OCT 9 '91 14: 02

204 942 5669 PRGE.018

OCT 09 '91 04:02PM ~ REID loFG MB- P.14 sample frame consists of 9, 713 advertisers. Based on these proportions, the estimated relative population count for non-newspaper advertisers is 8,613.

Step l'Wo of the weighting scheme consisted of weighting each of the three newspaper advertisers sub-strata (Daily only, C.Ommunity only, and Both) in proportion to their known representation on the sampling frame. A5 shown in Table 2, the Step Two weights again were simply calculated as the ratio of the populatioo proportion to the sample proportion for each sub-stratum.

All data and tables produced by the SPSS computer analysis were calculated using the Step Two weights. In this manner, each sub-stratum contributes the appropriate relative weight to the final estimates produced from the data.

22 SURVEY QUESTIONNAIRE A copy of the survey questionnaire for the telephone interviewing is found in Appendix •A". The questionnaire was administered by fully trained and supervised interviewers from the Angus Reid Group.

Question 1 wu used to obtain a basic profile of the retail advertiser being interviewed, including the number of retail outlets, the number of outlets in different locations within Greater Vancouver, and the number of outlets for which the respondent is responsible for buying advertising.

Question 2 was used to measure the full range of advertising vehicles purchased by Vancouver retail advertisers, the frequency of advertising purchases. and the amount spent on each vehicle over the last year.

Question 3 was used to determine the advertising objectives of retail advertisers. and for each of these objectives, Question 4 was used to determine their perceptions of the cost effectiveness of the different advertising vehicles available in Greater Vancouver. Question

&:'m1:c1 OCT 9 '91 14: 03

204 942 5669 PAGE.014

OCT 09 '91 04:02PM FtG.E REID loFG MB P.15 S provided an assessment of the cost effectiveness of the various advertising and promotional vehicles specifically for the business of the respondent. Questions six and seven provided an assessment of the size of the core trading area of both the individual stores as well as the combination of all stores of the respondent in Vancouver and Lower Mainland.

Question 8 was used to measure price sensitivity and switching behaviour with respect to price increases in each of the advertising vehicles. Two questions were used here, the first to determine what would be the likely response of advertisers when faced with a price increase (would they stay with that vehicle or switch?), and the second question was used to assess which alternative advertising vehicles they would switch to.

2.3 TELEPHONE INTERVIEWING The telephone interviews for this study were carried out by trained interviewers at the Angus Reid Group. The interviewers received two types of training. The first is general telephone interviewing which incorporates proper interviewing techniques and an overview of the entire research process. The second type of training received is project specific training. The supervisor reviews the questionnaire with the interviewers to ensure that each question is fully understood. Each question is read aloud by the interviewing staff and, at this time, supervisors and staff discuss the procedures and quality controls which apply to each question.

A ratio of approximately one supervision hour was spent for every six interviewer hours. This ratio has been proven to ensure required levels of monitoring and editing and supervision of workshifts. 10 to 15 percent of all calls on each shift were monitored or validated. All interviewers were monitored as equally as possible. Momtorlne consists of listening in on the interviews with telephone monitoring equipment. Validation consists of calling back selected respondents and verifying the responses which the interviewers have recorded on the questionnaires.

OCT 9 '91 14: 03

204 942 5669 PAGE.015

------OCT- 0 -9 ' -91- 0 -4:0-3PM- ~ REID i.FG MB P.16 All completed questionnaires were edited for completeness and accuracy. Any information which was found missing or incomplete was obtained by calling back respondents for clarification.

Appended to this report is a copy of some information on swvey response rates published by the Canadian Association of Market Research Organi7.ations (CAMRO) to help provide a context for interpreting the same information from this study. Comparing the two, it is quite evident that the response rates for the study are quite comparable to those reported in most other surveys in the industry, and better than those typically found in media-related research.

Completed interviews as a proportion of total contacts for this study were 25%, (as compared to 18% reported in media-related research by the CAMRO report). Refusals as a proportion of total contacts were 45% (as compared to the 43% reported in the CAMRO study for telephone interviews).

2.4 QUESTIONNAIRE PROCES.5ING AND DATA ANALYSIS Once all questionnaires were edited by the interviewer staff and field supervisor to ensure completeness and consistency, they were coded and keypunched in our Winnipeg office. All coding was 100% verified by independent coder, and all keypunching was 100% verified as well

The data was analyzed by computer using the SPSS software package. This software was used to produce the detailed tables of results which are appended to the report of findings for the study. All results were produced using the Sample Weights u calculated in Section 2.1.

For analytical purposes, the survey results were reported not only for the population of retail advertisers as a whole, but also for four categories of advertisers which are of special interest in the analysis of findings for this study. The four categories are defined as follows:

OCT 9 I 91 14: 04

204 942 5669 PAGE.016

OCT 09 '91 04:~ AG..E REID loFG MB P.17 Retail advertisers who have mn advertised using either community or dally newspapers in the last two years. It Js estimated that 52 percent of all retail advertisers belong to this category.

Malaly Dally: Advertisers who spend 70% or more of their newspaper advertising budget on ROP or insert advertising ln daily newspapers. For advertisers where detailed expenditure data was not available a second rule was used. The advertisers bad to advertise either weekly or monthly in the dailies Jlllil rarely or never in the community papers. Only under this circumstance were they considered "Mainly Daily". Fourteen percent of the smvey population are Mainly Daily advertisers.

Mainly Community: Advertisers who spend 70 percent or more of their newspaper advertising budget on ROP or insert advertising in community newspapers. For advertisers where detailed cxpenc:liture data was not available a second rule was used. The advertisers had to advertise either weekly or monthly in the community paper ..aru1 rarely or never Jn the daily papers(s). Only under this circumstance were they considered "Mainly Community", and these advertisers constitute 19 percent of the population.