There's an increasing trend I've noticed recently among people looking at stories - to take a brief look at stories to find the ones that "typify" the organisation. And, because of the perception of storywork being highly resource-intensive, it's generally a very small number of stories that are used.
Somewhat akin to the periodic iPod randomizer memes that crop up (the latest being here) - fun to see what comes up, but from such a small sample it's impossible to tell what's actually the dominant taste.
For example, my five random songs this morning were:
Beachcomber Voodoo by Julee Cruise (The Art of Being a Girl)
Haunted House by Leon Redbone (On the Track)
Better Be Home Soon by Crowded House (Farewell to the World)
Somebody Told Me by The Killers (Hot Fuss)
Losing My Religion by REM (Precious Rarities)
And it would be very easy to see patterns and make assumptions about my musical taste from that. Even more so if, by chance, an artist were repeated.
Yet a better way of seeing the bigger picture is to look at Last.fm (my profile is here) which builds up a cumulative picture from thousands of pieces.
Similarly, it's more accurate to look at larger volumes of stories and pick out patterns than it is from taking a small sample of stories from within the organisation. And, despite the assumptions, it doesn't have to be either difficult or expensive - it's relatively straightforward to build systems that gather and collect story material in volume and to be able to see the patterns that emerge. Software like SenseMakerTM make it very simple to look at thousands of story fragments quickly.
From those patterns, it might be possible to see specific stories that dominate a culture, but they'll be far fewer and not necessarily the ones that come out of a simple sample of a few hundred stories...