I have to give a quick response without time to refresh myself on the book that I first looked at over a decade ago. But Doug's question perplexes me.
Shaikh&Tonak, Dumenil&Levy, Mosely, and all the others who do empirical work HAVE to try to recalculate the conventional National Accounts since they are using different definitions of key items (capital, surplus value, profit rate etc).
I just quickly picked up the book, turned to the "Summary & Conclusions" and then to the section "Empirical Results" (pp 221-224). Shaikh&Tonak start by saying "...Marxian "total product" is roughly... about 1.5 times GNP...Marxian "gross final product" is about 15% smaller than GNP..."Surplus value" is almost double the most inclusive measure of profit type income..." etc etc. They show the different measures between the philosophies for productivity data, employment categories, and the rest. They then show different statistical picture you get from the two methods when you look at the trends for the major components of the economy that interest the analysts. It seems to be all summarized in 4 pages. The full version is about 60 pages (i.e. Chap 5).
IMO, most of the book should be thought of as basic research -- statistical methods that others can then apply (and some have since then). But S&T also apply their statistical methods to what they believe is a critical issue. The rate of profit for the period they examine (1948-89, they published the book in '94) was in a long run decline, bottoming out in the late 70s. Twelve years later we take this for granted, using similar methods others have shown the same, and today it is much less hard to convince people that something structural had happened to the economy in the post war decades that can not be explained by neo-classical theory (there have also been some new developments in the economy since 1989). But when S&T wrote this it was both fundamental and highly disputed, even on the left, and certainly from liberals over to conservatives (most of whom had tended to see post-war profit rates as cyclical and without a long run trend).
S&T go on to assert what they believe to be a principal cause of the decline in the profit rate: the rise in "unproductive" activities. Since Adam Smith and Ricardo this has been an important point among Classical economists: activities like the war, most advertising, most finance, health insurance companies, prisons, sales cashiers, supervisors devoted to monitoring the labor force, police, etc don't help get people the goods and services they want -- however "necessary" under capitalism and competition they are a cost and sap the growth of the economy.
Would you get the same results using the conventional National Accounts? As S&T (and everyone else who has done such research) show the conventional sources give you very different numbers on most of the important questions (pp 146-151 summarize the different results between their marxian national accounts and NIPA). Sometimes marxian and conventional methodologies do give you the same results, usually either by coincidence or because a trend is so strong that it overcomes the differences in methods and categories. This is less true as you try to parse the broad trends for explanations. But in any case you won't know until you have properly done the tedious research.
Hope this helps.
Paul