すべてのエントリを組み合わせたスタックおよびグループ化されたグラフggplot2

の設計はggplot、プロットに埋め込む必要のある変数の数についてかなり意見が分かれています。したがって、これはあなたが求めた方法では実行できません。

A few things to consider, none of them exactly what you asked for, none of them are necessarily good visualizations.

Here's your data (I changed the names so you'll need to adjust them to those on your local setup). Also, "Don't know" was codified as "Don't".

df <- read.delim(text=
                   "residence   age sex agree   persons percent
Berlin  18-39   Woman   Agree   1.689550205 0.009283243
Berlin  18-39   Woman   Disagree    3.942283812 0.0216609
Berlin  18-39   Woman   Don't   0.563183402 0.003094414
Berlin  40-70   Woman   Agree   1.689550205 0.009283243
Berlin  40-70   Woman   Disagree    3.942283812 0.0216609
Berlin  40-70   Woman   Don't   0.563183402 0.003094414
London  18-39   Woman   Agree   1.689550205 0.009283243
London  18-39   Woman   Disagree    3.942283812 0.0216609
London  18-39   Woman   Don't   0.563183402 0.003094414
London  40-70   Woman   Agree   1.689550205 0.009283243
London  40-70   Woman   Disagree    3.942283812 0.0216609
London  40-70   Woman   Don't   0.563183402 0.003094414
Berlin  18-39   Man Agree   1.689550205 0.009283243
Berlin  18-39   Man Disagree    3.942283812 0.0216609
Berlin  18-39   Man Don't   0.563183402 0.003094414
Berlin  40-70   Man Agree   1.689550205 0.009283243
Berlin  40-70   Man Disagree    3.942283812 0.0216609
Berlin  40-70   Man Don't   0.563183402 0.003094414
London  18-39   Man Agree   1.689550205 0.009283243
London  18-39   Man Disagree    3.942283812 0.0216609
London  18-39   Man Don't   0.563183402 0.003094414
London  40-70   Man Agree   1.689550205 0.009283243
London  40-70   Man Disagree    3.942283812 0.0216609
London  40-70   Man Don't   0.563183402 0.003094414
")

You could try to use color to codify this. The plot is quite bad and I bet you could potentially find a combination of colors and fills that is less disgusting. I wouldn't recommend this to my worst enemy but be my guest.

ggplot(df,
       aes(x=agree,
           y=round(percent, digits = 2),
           fill = age,
           color=sex)) +
  scale_fill_manual(values=c("gray50", "white"))+
  scale_color_manual(values=c("red", "black"))+
  geom_bar(stat = "identity",
           position=position_dodge(width = 1)) +
  facet_grid(.~residence)

Another option is creating an interaction variable. But you don't get to dodge and stack.

ggplot(df,
       aes(x=agree,
           y=round(percent, digits = 2),
           fill = interaction(sex, age))) +
  scale_fill_manual(values=c("gray50", "lightblue", "black", "blue"))+
  geom_bar(stat = "identity", color="white") +
  facet_grid(.~residence)

Finally, consider using facet_wrap(). This can allow you to stack as you wanted, but you will not have the age comparison side by side

ggplot(df,
       aes(x=agree,
           y=round(percent, digits = 2),
           fill = interaction(sex, age))) +
  scale_fill_manual(values=c("gray50", "lightblue", "black", "blue"))+
  geom_bar(stat = "identity", color="white") +
  facet_grid(age~residence)
  

注目すべきことに、パーセントを2桁に丸めると、それは消えます（値は0.0に丸められます）。アップロードした画像が実際に丸められているのか、データにゼロが追加されているのかわからない。これが最後のプロットですround(...)