Day 257 - 06 Jun 96 - Page 16

     
     1        we analysed a slightly bigger database, totalling 8,400,
     2        from the Taylor Nelson data.
     3
     4        My own analysis of the comparison between the first
     5        document, which is the 'Frequency of Visit' document, and
     6        the second document 'Number of People Eating at
     7        McDonald's', is that actually against the main breaks, and
     8        the main breaks which we, McDonald's, pay attention to,
     9        which is visiting weekly, which are heavy users, visiting
    10        monthly, lighter users, and less than monthly, from a
    11        research perspective in adding the percentages together the
    12        broadly 15 per cent of people who you would call heavy
    13        users from the first sheet correlates fairly well with
    14        broadly 10 per cent of the cumulative weekly usage on the
    15        second sheet, and also the 37 per cent working up to
    16        monthly usage on the first sheet correlates fairly well to
    17        the 32 per cent working up to monthly usage on the second
    18        sheet.
    19
    20   MR. JUSTICE BELL:  Sorry, do that again.  What correlates fairly
         well to what?
    22        A.  The cumulative percentage of users visiting, so my only
    23        analysis was that the comparability of the relative
    24        percentages of people visiting weekly from the first
    25        document is, in my opinion, encouraging.
    26
    27   Q.   What do you mean 'cumulative'?
    28        A.  By adding.  On the first sheet by adding the 4 per cent
    29        of the several times a week and the once a week together,
    30        I get--
    31
    32   Q.   15 per cent?
    33        A.  Yes, more frequently, or equivalent to weekly, is 15
    34        per cent.  If you do the same with the second sheet, you
    35        get approximately 10 per cent.  Also, then, if you build
    36        the two sheets up to, on the same basis, usage on a monthly
    37        basis, you get 37 per cent on the first sheet, and 32 per
    38        cent on the second sheet.  Again, building up again to less
    39        frequent usage, 48 per cent on the first sheet and about 40
    40        per cent on the second sheet, given that within samples of
         this size we normally accept a variance of plus or minus 4
    42        per cent points, either way, to me I was actually
    43        surprised, encouragingly surprised, that the two different
    44        sources of information gave what I would call broadly
    45        similar and therefore bore each other out in terms of
    46        relative accuracy.  It was encouraging for me to see two of
    47        my research sources correlating in this way when
    48        information is collected for different reasons and by a
    49        different user base.
    50 
    51   Q.   Just pause a minute.  So what you are saying is, if you 
    52        look at the two surveys, if you look at the individual 
    53        divisions or components, the percentages or number of
    54        people dining at McDonald's with that frequency, or using
    55        McDonald's with that frequency, may well be very different,
    56        it does not matter whether you do it by numbers or
    57        percentages, you are going to get very different answers.
    58        You do, in fact, get very different answers but you take it
    59        in brooder sectors, then you say they are not so far out?
    60        A.  That is correct.