So what is the base of the method? All of know that playing games of chess masters (and grandmasters) is a good way to learn more about chess.
Much better way is consulting games with chess masters.
But really - how many times does an average chess player this chance in his/her lifetime? I do not count on those lucky who have chess master learning them in a chess club for example.
What Purdy suggests is this.
Chose a game between two strong players.
Play first 7 moves of the game and them play instead of the player who won the game (in case the game was a draw, you can chose any player to play instead of him).
Hide all the next moves of the game (by a card for example) and start to play like the winner.
Think out of all your moves as you would be the real player of the game.
Be precise in creating your plan and calculating all variants.
Play your moves without looking at "real" moves of your side from the game.
Once you made your move on the board then slide the card over till the game move is exposed.
If the move is different from your one, try to find why.
Did you make a blunder? Selected another/wrong plan? But NEVER look at an annotation beforehand - it would destroy the learning purpose! Look at the opponent's reply immediately.
It could help you to find where could be a problem with your move.
Repeat this until you "win" the game.
And repeat it with many many games.
Of course, you can use clocks like in a real game.
It would move your experience and training closer to reality.
You can apply this method also in cases when you want to learn more about specific style.
Personally, I am going to play Karpov's games according Purdy's method as I need to improve in positional play.
And Karpov is one of the best positional players of all times.