李克特图显示百分比值

AFAIK没有实现此目的的任何参数，因此您需要以这种方式定义自定义面板功能:

AFAIK there isn't any parameter to achieve that, so you need to define a custom panel function, in this way :

### just reproduce your input
library(HH)
sgbar.likert <- data.frame(X1 = c(34L, 7L, 13L),X2 = c(1L, 4L, 13L),
                           X3 = c(7L, 84L, 24L), X4 = c(7L, 2L, 27L), X5 = c(51L, 3L, 23L))
yLabels = c(expression(a[1*x]),expression(b[2*x]),expression(c[3*x]))
### 

# store the original col names used in custom panel function
origNames <- colnames(sgbar.likert)

# define a custom panel function
myPanelFunc <- function(...){
  panel.likert(...)
  vals <- list(...)
  DF <- data.frame(x=vals$x, y=vals$y, groups=vals$groups)

  ### some convoluted calculations here...
  grps <- as.character(DF$groups)
  for(i in 1:length(origNames)){
    grps <- sub(paste0('^',origNames[i]),i,grps)
  }

  DF <- DF[order(DF$y,grps),]

  DF$correctX <- ave(DF$x,DF$y,FUN=function(x){
    x[x < 0] <- rev(cumsum(rev(x[x < 0]))) - x[x < 0]/2
    x[x > 0] <- cumsum(x[x > 0]) - x[x > 0]/2
    return(x)
  })

  subs <- sub(' Positive$','',DF$groups)
  collapse <- subs[-1] == subs[-length(subs)] & DF$y[-1] == DF$y[-length(DF$y)]
  DF$abs <- abs(DF$x)
  DF$abs[c(collapse,FALSE)] <- DF$abs[c(collapse,FALSE)] + DF$abs[c(FALSE,collapse)]
  DF$correctX[c(collapse,FALSE)] <- 0
  DF <- DF[c(TRUE,!collapse),]

  DF$perc <- ave(DF$abs,DF$y,FUN=function(x){x/sum(x) * 100})
  ###

  panel.text(x=DF$correctX, y=DF$y, label=paste0(DF$perc,'%'), cex=0.7)
}

# plot passing our custom panel function
likert(sgbar.likert,
       scales = list(y = list(labels = yLabels)),
       xlab="Percentage",
       main="Example Diverging Stacked Bar Chart for Likert Scale",
       BrewerPaletteName="Blues",
       panel=myPanelFunc,
       sub="Likert Scale")

代码非常复杂，但是关键是面板函数接收(在省略号...参数中)每个条形的x，y坐标对和每个条形的组因子(组是列原始的Likert输入).默认面板功能是panel.likert；默认设置是panel.likert.因此，在调用该命令之后，我们可以将更改添加到绘制的面板(在这种情况下，根据条形坐标的标签).

The code is pretty convoluted, but the key is that panel function receives, (in the ellipsis ... parameter), an x,y pair of coordinates for each bar and group factor for each of them (groups are the columns of the original likert input). The default panel function is panel.likert; so, after calling that, we can add our changes to the plotted panel (in this case the labels according to the bars coordinates).

看起来很简单，但是有两个问题:

Seems easy, but there are two problems :

1. 组在它们为偶数时会重新定义，因此中心列(在这种情况下为"X3")分为两组:"X3"和"X3 Positive".
2. 绘制的条形图是堆叠的"，因此要正确计算它们的中心(以放置标签)，您需要使用原始列名的顺序来计算坐标的累积总和.

1. groups are redefined when they are even, so the central column, in this case "X3" it's split in two groups: "X3" and "X3 Positive".
2. plotted bars are "stacked", so to correctly compute the centers of them (in order to put a label) you need to calculate the cumulated sum of the coordinates, using the original column names ordering.

以上代码希望以一种通用的方式进行所有这些计算(阅读:您可以更改输入，它应该可以工作...).

The above code does all those calculations, hopefully in a quite generic manner (read: you can change the input and it should work...).