在R中的分组箱图上显示p值

Question

I want to show p-values on above my data (possibly with arcs). My data is below:

ID  Blog    Region  Dimension   Score
1   Blog1   PK  Info. vs. P. Focus  -4.75
2   Blog1   PK  Info. vs. P. Focus  -5.69
3   Blog1   PK  Info. vs. P. Focus  -0.27
4   Blog1   PK  Info. vs. P. Focus  -2.76
5   Blog1   PK  Info. vs. P. Focus  -8.24
6   Blog1   PK  Addressee Focus -12.51
7   Blog1   PK  Addressee Focus -1.28
8   Blog1   PK  Addressee Focus 0.95
9   Blog1   PK  Addressee Focus -5.96
10  Blog1   PK  Addressee Focus -8.81
11  Blog1   PK  Thematic Variation  -8.46
12  Blog1   PK  Thematic Variation  -6.15
13  Blog1   PK  Thematic Variation  -13.98
14  Blog1   PK  Thematic Variation  -16.43
15  Blog1   PK  Narrative Style -4.09
16  Blog1   PK  Narrative Style -11.06
17  Blog1   PK  Narrative Style -9.04
18  Blog1   PK  Narrative Style -8.56
19  Blog1   PK  Narrative Style -8.13
20  Blog1   PK  Narrative Style -14.46
21  Blog1   PK  Info. vs. P. Focus  -4.21
22  Blog1   PK  Info. vs. P. Focus  -4.96
23  Blog1   PK  Info. vs. P. Focus  -5.48
24  Blog1   PK  Info. vs. P. Focus  -4.53
25  Blog1   PK  Info. vs. P. Focus  6.31
26  Blog1   PK  Addressee Focus -11.16
27  Blog1   PK  Addressee Focus -1.27
28  Blog1   PK  Addressee Focus -11.49
29  Blog1   PK  Addressee Focus -0.9
30  Blog1   PK  Addressee Focus -12.27
31  Blog1   PK  Thematic Variation  6.85
32  Blog1   PK  Thematic Variation  -5.21
33  Blog1   PK  Thematic Variation  -1.06
34  Blog1   PK  Thematic Variation  -2.6
35  Blog1   PK  Narrative Style -0.95
36  Blog1   PK  Narrative Style -0.82
37  Blog1   PK  Narrative Style -7.65
38  Blog1   PK  Narrative Style 0.64
39  Blog1   PK  Narrative Style -2.25
40  Blog1   PK  Narrative Style -1.58
41  Blog1   PK  Info. vs. P. Focus  -5.73
42  Blog1   PK  Info. vs. P. Focus  0.37
43  Blog1   PK  Info. vs. P. Focus  -5.46
44  Blog1   PK  Info. vs. P. Focus  -3.48
45  Blog1   PK  Info. vs. P. Focus  0.88
46  Blog1   PK  Addressee Focus -2.11
47  Blog1   PK  Addressee Focus -10.13
48  Blog1   PK  Addressee Focus -2.08
49  Blog1   PK  Addressee Focus -4.33
50  Blog1   PK  Addressee Focus 1.09
51  Blog1   US  Thematic Variation  -4.23
52  Blog1   US  Thematic Variation  -1.46
53  Blog1   US  Thematic Variation  9.37
54  Blog1   US  Thematic Variation  5.84
55  Blog1   US  Narrative Style 8.21
56  Blog1   US  Narrative Style 7.34
57  Blog1   US  Narrative Style 1.83
58  Blog1   US  Narrative Style 14.39
59  Blog1   US  Narrative Style 22.02
60  Blog1   US  Narrative Style 4.83

The code is below:

get_wraper <- function(width) {
  function(x) {
    lapply(strwrap(x, width = width, simplify = FALSE), paste, collapse="\n")
  }
}
plotgraph <- function(x, y, colour, min, max, incr, p_values)
{
  plot1 <- ggplot(dims_Blog, aes_string(x = x, y = y, fill = colour)) +
    geom_boxplot()+
    labs(color=colour) +
    labs(x="Dimensions", y="Score") +
    scale_fill_grey(start = 0.3, end = 0.6) +
    theme_grey()+
    theme(legend.justification = c(1, 1), legend.position = c(1, 1)) +
    scale_x_discrete(labels = get_wraper(10))+
    scale_y_continuous(breaks=c(seq(min,max,incr)), limits = c(min, max))+
    theme(panel.grid.minor.y = element_blank(), panel.grid.major.x = element_blank())+
    geom_text(data = dims_Blog %>% group_by_(x, colour) %>% summarise_(mean=paste("mean(",y,", na.rm=TRUE)")), aes_string(x=x, y="mean", label="round(mean,3)"), position=position_dodge(width=0.8), size = 3, vjust = -0.5, colour="white")+
    geom_text(data = p_values, aes_string(x="Dimension", y="height", label="val"))
  return(plot1)
}

plot graph

plot1 <- plotgraph("Dimension", "Blog1", "Region", -30, 50, 10, p_val1)
plot1

data frame for p-values

Dimensions <- c("Info. vs. P. Focus", "Addressee Focus", "Thematic Variation", "Narrative Style")
val <- c("0.184", "0.079", "0.044", "\u003C.0001")
height <- c(48, 48, 48, 48)
p_val1 <-data.frame(Dimensions, val, height)

Unfortunately I am not sure how to define geom_text for showing p-values.

Error: Aesthetics must be either length 1 or the same as the data (8): label, x, y, fill

I have tried go through a few similar questions but my limited knowledge did not let me solve the problem. Any ideas?

Answer 1

It seems that you were very close in the original post: the error message says that you need to provide label, x, y, fill attributes for every layer. (That is because you defined these attributes in the main ggplot call.) The layer that you use for p-values includes three of these aesthetics in the aes_string(x="Dimensions", y="height", label="val") . Try adding a constant fill, like:

+ geom_text(data = p_values, aes_string(x="Dimensions", y="height", label="val"), fill="black")

or you can move the aesthetics definitions out of the main call, if you are not using multiple layers anyway:

ggplot(dims_Blog) +
    geom_boxplot(aes_string(x = x, y = y, fill = colour)) +
    ... +
    geom_text(data = p_values, aes_string(x="Dimensions", y="height", label="val"))

Secondly, there's a typo - you refer to Dimension in the plotting call, but Dimensions in creating the p-val dataframe.

I haven't tested this without having the full dataset, though, so something additional might come up.

Answer 2

I have used annotate instead of geom_text which takes three separate vectors (?) instead of a data frame. The code is shown below:

annotate("text",x=Dimension,y=height,label=val)
Dimension <- c("Info. vs. P. Focus", "Addressee Focus", "Thematic Variation", "Narrative Style")
height <- c(48, 48, 48, 48)
val <- c("p=0.184", "p=0.079", "p=0.044", "p\u003C.0001")

This is not a very good solution, but at least it printed the values I wanted it to print. I don't know how to expand these vectors to 8 (as my previous data frames are of this size). That was the problem as well.