9
Method
We analyzed the content of the main evening newscasts by NOS en RTL before the
‘Battle of the Newscasts’ started (spring 1995), as well as the main evening newscasts by
NOS, RTL and SBS at a moment when the developments in the Dutch television
landscape seemed to have settled for a while (spring 2001). For obvious reasons, the
design does not allow definite conclusions concerning ‘effects’ of the increased
competition. A comparison between merely two time points can establish no more than
an indication of simultaneous trends, of which the increase in competition between news
programs and the expected increase in sensationalism are only two. For this reason, the
theoretical influence of concurrent developments on the degree of sensationalism in the
news cannot be ruled out empirically. For instance, technological developments or
changing preferences amongst the public might play their part.
Sampling method
For both 1995 and 2001, we analyzed newscasts of the month March, because this month
can be regarded as month that is reasonably representative of an entire year. For instance,
in most years (and also in 1995 and 2001), no large holidays, big sports events or
elections occur. Of the year 2001, we selected March 9, followed by March 11, 13 and so
on, until we arrived at a sample of 14 days, with each day of the week being selected
twice. Given the relatively small sample of days, such a procedure is superior to random
sampling, because it guarantees a large variance of days over the period (March 9 – April
4), as well as an equal number of weekdays in the sample (cf. Riffe, Aust & Lacy, 1993).
For each selected day, we analyzed all news items from the main evening newscasts by
NOS, RTL and SBS, which amounted to a total of 423 items.
For the year 1995 we followed a similar procedure. However, because of a limited
availability of RTL newscasts (the RTL archive appeared to be incomplete, due to a
random loss of tapes), we could not sample as systematically as we did for the year 2001.
Nevertheless, we arrived at 13 days over the period March 1 to March 28, with each day
of the week except Thursday being selected twice. For each selected day, we analyzed all
Convention