You can point out ‘ethnic block voting’ without making bogus use of graphs.
Putting two measures on one graph line like that and then claiming that the GOP and the population are most similar—when you’re comparing graphs of two different things—is idiotic and/or deceitful.
You’re reading the chart wrong. They are both percentages of something, share of the population and share of the electorate for each of the racial categories. Even if you used multiple y axes the graph would look exactly the same. Think of it like stock charts measuring rate of return on two different stocks over time both measured in percentages. A perfectly legitimate comparison.
One more thing this works best if we believe that the voting patterns of the electorate are similar to the party preferences of the population as a whole which is probably mostly true.