Yes, it makes a substantial difference. You used the key word yourself: diversification.
Each of these has a different risk/reward profile, moves in different ways in response to world news, etc. For the same reason you want to diversify rather than putting it all on a single stock, you may want to diversify across categories.
A good fee-only advisor can recommend a mix that suits your timeframe and risk tolerance, and test that mix against both historical data and simulations of the market to validate its behavior. Putting it all in one place is generally not the one that shows best results under these circumstances. And relatively minor shifts in allocation percentages can have significant effects on the expected behavior.
The following ONLY AN EXAMPLE, NOT A RECOMMENDATION, since what's right for me may not be for you and I don't even claim that it's perfect for me: I am currently using a mix of index funds (traditional, not ETF) with the target allocation being 27.5% domestic bonds
37% large/medium cap stocks
6.5% small cap stocks
22.82% international (much higher than I would have used in the past)
6% in income-producing investments (I'm using REITs for that purpose)
I rebalance once a month at most, usually more like once a quarter when my actual distribution is more than a full percentage point away from those targets.
I don't go anywhere near individual stocks, option, derivitives, or anything beyond the index funds. I don't want to work that hard or take higher risk; I'm already considered a relatively aggressive investor in being willing and prepared to sit out downturns of a full year.
With that mix, my projected typical 5-year return was estimated as -1.4% to +10.2%. An even more aggressive mix (25,40,9,21,5) was estimated as -10.4 to +27.3 -- and you can see that this isn't a hugely different mix of percentages.
Of course the past 5 years have not been anything like typical. Indeed the recent downturn may have qualified as a "Black Swan" year, which they say I should expect to see 1 or 2 of per lifetime.