2
$\begingroup$

I'm very new at this and I don't actually understand the differences between the plotting methods, but loess seems to be giving me the most informative graphs, considering I have a small-ish data set (n=~300). I'm trying to split my data by gender using facet_wrap, and loess is working fine for men, but not for women.

Here's the code I'm using to plot the graph:

ggplot(data = df, aes(x = STM, y = ATTRACTcomp, color=Harasser_Attractiveness)) +
     geom_point(position="jitter", size=0.5) +
     facet_wrap( ~Participant_Gender, 
                 labeller = as_labeller(c("Female" = "Female Participants", "Male" = "Male Participants"))) +
     geom_smooth(method = "loess") +
     labs(title = paste(strwrap("Interaction of Harasser Attractiveness, Participant Gender 
                                and SOI on Attraction/Flattery", 50), collapse="\n"),
          x = "Participant Short-term Mating Orientation", y = "Participant Attraction/Flattery", 
          color="Harasser:") +
     theme(plot.title = element_text(hjust = 0.5), 
           plot.caption = element_text(hjust=0, margin=margin(t=15,0,0,0)), 
           legend.position="top", legend.margin = margin(1,0,0,0), legend.title = element_text(size=10), 
           legend.text = element_text(size=9), legend.key.size=unit(c(12), "pt")) +
     scale_color_grey(start = .6, end = .1)

Here's the plot I'm getting:

enter image description here

And here are my error messages:

Warning messages:
1: In simpleLoess(y, x, w, span, degree = degree, parametric = parametric,  :
  at  0.97
2: In simpleLoess(y, x, w, span, degree = degree, parametric = parametric,  :
  radius  0.0009
3: In simpleLoess(y, x, w, span, degree = degree, parametric = parametric,  :
  all data on boundary of neighborhood. make span bigger
4: In simpleLoess(y, x, w, span, degree = degree, parametric = parametric,  :
  pseudoinverse used at 0.97
5: In simpleLoess(y, x, w, span, degree = degree, parametric = parametric,  :
  neighborhood radius 0.03
6: In simpleLoess(y, x, w, span, degree = degree, parametric = parametric,  :
  reciprocal condition number  1
7: In simpleLoess(y, x, w, span, degree = degree, parametric = parametric,  :
  zero-width neighborhood. make span bigger
8: In simpleLoess(y, x, w, span, degree = degree, parametric = parametric,  :
  There are other near singularities as well. 1
9: Computation failed in `stat_smooth()`:
NA/NaN/Inf in foreign function call (arg 5)

The interesting thing is this happens for multiple y variables: the female graph is always missing the lines and I always get similar errors.

From what I understand reading threads about similar error messages, some computation within geom_smooth (or stat_smooth as I think it's called under the hood) is returning infinite values. (I am fairly certain there are no NAs/NaNs in the relevant variables here.) The problem is, all the threads about this error assume that you have access to the process producing the infinite values, and I don't.

Some people have been saying this can occur when you have values equal to exactly 1. I do have quite a few values of ATTRACTcomp (my y variable) equal to exactly 1, but they are both men and women, so I don't know why I'm able to get the correct lines for men but not women.

Alternative plotting methods that would be equally informative would also be helpful.

I'm not sure what the minimal amount of data necessary to reproduce this error is, so I'm just going to include a dataframe with only the variables used in the graph:

> dput(df)
structure(list(STM = c(6L, 4L, 7L, 3L, 6L, 7L, 3L, 1L, 4L, 6L, 
1L, 1L, 6L, 4L, 6L, 3L, 5L, 2L, 5L, 5L, 4L, 1L, 1L, 4L, 4L, 1L, 
1L, 2L, 3L, 4L, 3L, 4L, 6L, 6L, 1L, 1L, 1L, 5L, 1L, 1L, 2L, 4L, 
2L, 1L, 1L, 1L, 1L, 1L, 2L, 4L, 7L, 2L, 1L, 6L, 4L, 1L, 1L, 1L, 
1L, 1L, 4L, 1L, 4L, 5L, 1L, 1L, 7L, 4L, 1L, 1L, 1L, 1L, 2L, 1L, 
1L, 1L, 2L, 1L, 1L, 1L, 4L, 1L, 1L, 2L, 1L, 1L, 2L, 4L, 5L, 1L, 
1L, 1L, 1L, 4L, 1L, 2L, 1L, 7L, 5L, 4L, 1L, 1L, 1L, 1L, 1L, 4L, 
1L, 1L, 1L, 1L, 1L, 1L, 2L, 1L, 1L, 1L, 1L, 1L, 2L, 1L, 1L, 1L, 
7L, 3L, 1L, 1L, 1L, 1L, 7L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 
1L, 2L, 5L, 1L, 1L, 1L, 1L, 1L, 1L, 3L, 1L, 1L, 1L, 1L, 1L, 1L, 
1L, 1L, 1L, 1L, 1L, 1L, 1L, 7L, 1L, 2L, 1L, 1L, 1L, 1L, 2L, 1L, 
1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 2L, 5L, 1L, 1L, 1L, 1L, 1L, 1L, 
1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 2L, 5L, 5L, 4L, 1L, 1L, 
1L, 1L, 1L, 2L, 1L, 7L, 1L, 2L, 2L, 1L, 1L, 1L, 1L, 1L, 5L, 2L, 
1L, 1L, 6L, 2L, 1L, 1L, 1L, 1L, 5L, 2L, 1L, 1L, 1L, 1L, 4L, 1L, 
1L, 1L, 1L, 1L, 2L, 4L, 1L, 1L, 1L, 6L, 1L, 1L, 1L, 3L, 1L, 1L, 
1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 3L, 1L, 1L, 1L, 1L, 1L, 
4L, 5L, 5L, 1L, 1L, 4L, 4L, 1L, 7L, 1L, 1L, 4L, 3L, 1L, 1L, 1L, 
1L, 1L, 1L, 2L, 2L, 5L, 1L, 1L, 1L, 1L, 5L, 2L, 1L, 4L, 7L, 1L, 
1L, 2L, 1L, 1L, 4L, 5L, 5L, 2L, 1L, 4L, 7L, 3L, 5L, 4L, 5L, 4L, 
5L, 7L, 7L, 3L), ATTRACTcomp = c(6.53125, 4.25, 5.84375, 4.21875, 
5.4375, 2.15625, 3.96875, 4.71875, 3.875, 5.875, 2, 1.87096774193548, 
5.65625, 4.5625, 5.65625, 4.53125, 5.375, 1, 5.125, 3.5625, 4.71875, 
3.96875, 4.03125, 4.15625, 4.28125, 4.6875, 3.53125, 2.40625, 
4.15625, 2.8125, 4.54838709677419, 3.40625, 4.09677419354839, 
4.625, 4.53125, 1.90625, 2.32258064516129, 3.53125, 1.90625, 
3.46666666666667, 2.2258064516129, 3.625, 4.40625, 4.625, 2.125, 
4.3125, 1.9375, 2.4375, 3.96875, 4.875, 5.16129032258065, 2.1875, 
1.0625, 3.34375, 3.40625, 1.90625, 1, 3.75, 3.45161290322581, 
1.93548387096774, 3.53125, 1.84375, 2.71875, 3.40625, 2.59375, 
4.09375, 4.125, 3.96875, 4.34375, 1, 2.6875, 3.6875, 1.09375, 
1.0625, 1.375, 1.96875, 2.25, 1.28125, 1.03125, 3.8125, 4.0625, 
2.09375, 1.25, 2.34375, 2.90625, 1, 1.5625, 1.25, 1.5625, 1.34375, 
2.46875, 1.96875, 1.15625, 1.59375, 1.09375, 2.03125, 1, 5.40625, 
3.59375, 1.1875, 1.90625, 1.8125, 1.56666666666667, 1.0625, 3.58064516129032, 
4.90625, 6.28125, 1.0625, 2.9375, 1.09375, 1.78125, 1, 2.09375, 
1.03125, 4.75, 2.71875, 1, 5.96875, 1.42307692307692, 1, 1.0625, 
1.0625, 1.03125, 1.90625, 1.28125, 1.15625, 1.03125, 1.09375, 
6.53125, 2.15625, 1.03125, 1.59375, 2, 1.1875, 1.1875, 1.34375, 
2.25, 1.03125, 1.0625, 1.3125, 1, 1.5, 1, 2.375, 1.1875, 1.0625, 
1.35483870967742, 1, 1.09375, 1.15625, 1, 1, 1.5625, 2, 1, 1.03125, 
1.03125, 1, 1.125, 1, 6.6875, 1.1875, 1.51612903225806, 1.0625, 
1.125, 1, 1.15625, 1.4375, 1.25, 1.0625, 1.03125, 1.41935483870968, 
1, 1, 2.09375, 1.15625, 1, 1, 1, 3.06451612903226, 1, 1, 1, 1, 
1, 1, 1, 1.03125, 1.1875, 1.875, 1, 1, 1.5625, 3.25, 1.3125, 
1.46875, 2.375, 3.78125, 3.25, 1.21875, 1.25, 1, 1.65625, 1, 
1, 6.0625, 1.90625, 6.80645161290323, 1.21875, 1.65625, 1, 1.28125, 
1.26666666666667, 1.03125, 1, 2.3125, 4.125, 3.59375, 2.40625, 
5.34375, 4.84375, 3.65625, 1.28125, 1.5625, 3.6875, 1.53125, 
1.09375, 1.21875, 2.15625, 1.25, 1, 1.375, 1.3125, 1.125, 1.5625, 
1.25, 1.5, 1.28125, 2.21875, 3.09375, 3.15625, 1, 1.15625, 4.75, 
1, 1.61290322580645, 1.90322580645161, 1.74193548387097, 1.46875, 
1, 1.1875, 1.1875, 1.03125, 1.34375, 1.78125, 1, 1.8125, 1, 1, 
1.2258064516129, 1.0625, 1.25, 1.59375, 1.09375, 1, 1.03125, 
3.9375, 1.46875, 2.71875, 7, 3.875, 3.40625, 2.4375, 2.53125, 
2.09677419354839, 1.28125, 1, 1.8125, 1, 1.78125, 1.0625, 1, 
1, 1.03125, 1.09375, 1.4375, 1, 1.625, 1.03125, 1.03125, 1.40625, 
1.84375, 3.40625, 3.21875, 1, 1, 6.6875, 2.71875, 2.5625, 3.96875, 
2.8125, 2.125, 4.21875, 3.65625, 3.25, 1.53125, 5.8125, 3.5625, 
4.78125, 1.625, 5.875, 3.21875, 3.41935483870968, 3.21875, 6, 
6.34375, 6, 1.40625), Harasser_Attractiveness = structure(c(1L, 
1L, 1L, 2L, 1L, 1L, 2L, 2L, 2L, 1L, 1L, 1L, 2L, 1L, 2L, 2L, 2L, 
2L, 2L, 1L, 2L, 2L, 2L, 2L, 1L, 2L, 1L, 2L, 2L, 1L, 1L, 2L, 2L, 
2L, 1L, 1L, 2L, 2L, 2L, 1L, 2L, 1L, 1L, 1L, 2L, 2L, 2L, 1L, 1L, 
2L, 1L, 2L, 2L, 1L, 2L, 1L, 2L, 2L, 1L, 1L, 1L, 1L, 2L, 1L, 2L, 
2L, 1L, 1L, 1L, 1L, 2L, 1L, 2L, 2L, 2L, 1L, 1L, 2L, 1L, 1L, 2L, 
1L, 2L, 1L, 2L, 2L, 2L, 1L, 2L, 1L, 1L, 1L, 2L, 1L, 2L, 1L, 2L, 
1L, 1L, 1L, 1L, 2L, 2L, 2L, 1L, 1L, 1L, 1L, 2L, 2L, 1L, 1L, 1L, 
2L, 2L, 1L, 1L, 1L, 1L, 2L, 2L, 1L, 2L, 2L, 2L, 1L, 2L, 1L, 1L, 
2L, 2L, 2L, 1L, 1L, 1L, 1L, 1L, 2L, 1L, 2L, 2L, 1L, 1L, 2L, 1L, 
2L, 2L, 2L, 2L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 2L, 2L, 2L, 2L, 1L, 
2L, 1L, 2L, 2L, 1L, 1L, 1L, 2L, 1L, 1L, 2L, 1L, 2L, 1L, 2L, 2L, 
1L, 1L, 1L, 1L, 1L, 1L, 2L, 2L, 2L, 1L, 2L, 2L, 2L, 1L, 2L, 2L, 
1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 
1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 
1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 
1L, 1L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 
2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 
2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 
2L, 2L, 2L, 2L, 2L, 2L, 2L, 1L, 1L, 1L, 1L, 1L, 2L, 2L, 2L, 2L, 
2L, 1L, 1L, 1L, 1L, 2L, 2L, 2L, 1L, 2L, 2L, 2L, 2L), .Label = c("Attractive", 
"Unattractive"), class = "factor"), Participant_Gender = structure(c(2L, 
2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 1L, 2L, 1L, 2L, 
2L, 1L, 2L, 2L, 2L, 2L, 1L, 2L, 2L, 1L, 2L, 2L, 2L, 2L, 2L, 1L, 
2L, 2L, 2L, 1L, 2L, 2L, 2L, 2L, 1L, 1L, 2L, 2L, 2L, 1L, 2L, 2L, 
2L, 1L, 2L, 1L, 2L, 2L, 1L, 2L, 2L, 1L, 1L, 2L, 1L, 2L, 1L, 1L, 
1L, 2L, 1L, 2L, 2L, 2L, 2L, 1L, 1L, 1L, 2L, 1L, 2L, 2L, 2L, 2L, 
2L, 1L, 2L, 1L, 1L, 1L, 1L, 1L, 2L, 1L, 1L, 2L, 1L, 2L, 2L, 1L, 
2L, 1L, 1L, 1L, 1L, 2L, 1L, 2L, 2L, 2L, 1L, 1L, 1L, 1L, 1L, 1L, 
1L, 2L, 2L, 1L, 2L, 1L, 1L, 1L, 2L, 1L, 1L, 1L, 2L, 1L, 1L, 2L, 
1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 2L, 1L, 1L, 1L, 2L, 2L, 
1L, 1L, 1L, 1L, 1L, 1L, 2L, 1L, 1L, 1L, 1L, 1L, 2L, 1L, 1L, 2L, 
1L, 1L, 2L, 1L, 1L, 1L, 2L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 2L, 2L, 
1L, 1L, 2L, 2L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 2L, 2L, 1L, 1L, 1L, 
1L, 1L, 1L, 2L, 2L, 2L, 2L, 1L, 1L, 1L, 1L, 1L, 2L, 2L, 2L, 2L, 
2L, 1L, 1L, 2L, 1L, 1L, 1L, 1L, 1L, 1L, 2L, 2L, 2L, 1L, 1L, 1L, 
1L, 2L, 2L, 2L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 2L, 2L, 2L, 
2L, 1L, 2L, 1L, 1L, 2L, 2L, 2L, 1L, 1L, 1L, 1L, 1L, 1L, 2L, 2L, 
1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 2L, 2L, 1L, 2L, 1L, 1L, 2L, 1L, 
2L, 2L, 2L, 1L, 2L, 1L, 2L, 1L, 2L, 2L, 2L, 1L, 1L, 1L, 2L, 1L, 
1L, 1L, 1L, 2L, 2L, 1L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 1L, 2L, 2L, 
2L, 1L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 1L, 2L, 2L, 2L), .Label = c("Female", 
"Male"), class = "factor")), .Names = c("STM", "ATTRACTcomp", 
"Harasser_Attractiveness", "Participant_Gender"), row.names = c(NA, 
-318L), class = "data.frame")
Improve this question
$\endgroup$
6
  • 3
    $\begingroup$ You have several options: (1) include , span=1 in the geom_smooth call. (2) include method="gam" in the geom_smooth call. (3) Manually jitter the data by changing the aes call to aes(x = STM + runif(nrow(df), -0.1, 0.1), .... The last is the most informative and useful, because the way the confidence bands explode reveals where the problem really lies. Running this several times shows you how sensitive the smooth might be to the jittering. It should convince you of the merits of solution (1) (using a large span). I hope these comments help show this is a statistical question! $\endgroup$
    – whuber ♦
    Commented Apr 16, 2018 at 19:46
  • $\begingroup$ There doesn't seem much information here in most parts of the graph. Have you considered smoothing on some transformed scale? $\endgroup$
    – Nick Cox
    Commented Apr 16, 2018 at 19:53
  • $\begingroup$ @Nick That's a good idea, but it doesn't succeed because the x-coordinate takes on only integral values $1,2,3,4,5,6,7$ (which isn't immediately apparent due to the overly heavy default jittering applied by ggplot), with frequencies $199,30,12,31,22,10,14$ respectively for female participants, and the problem occurs (as speculated in the question) because of the pile of data with $x=0.$ It's an interesting example for learning more about how Loess behaves. $\endgroup$
    – whuber ♦
    Commented Apr 16, 2018 at 20:07
  • $\begingroup$ @whuber You're right. I am currently looking at the data. I didn't appreciate that until I'd translated from R to my different software. $\endgroup$
    – Nick Cox
    Commented Apr 16, 2018 at 20:08
  • 1
    $\begingroup$ I ran all of the solutions I suggested and they worked :-). However, I began by applying some basic debugging operations: I restricted the dataframe to females and removed all superfluous graphical elements such as facet_wrap, the labels, theme, and titles. BTW, comments aren't terribly important around here, but they can be upvoted by hovering the mouse just to the left: you should see a triangular icon for upvoting and a flag icon beneath it for flagging objectionable comments for moderator attention. $\endgroup$
    – whuber ♦
    Commented Apr 16, 2018 at 22:14

1 Answer 1

Reset to default
1
$\begingroup$

This isn't a complete answer, but its one contribution -- a graph to show a different method and to make clear an underlying problem -- can't fit in a comment.

enter image description here

The idea is just a combined dot and box plot (boxes show median and quartiles) with extra lines for means. So, that is a fairly conservative graph with "smoothing" applied only mentally. It underlines how few data points are for STM $> 1$ and how far they behave systematically, or otherwise.

I can't comment on any problems with the OP's R code. FWIW, I didn't use R for this, but Stata. (When there are singleton values, the program draws boxes of zero length. They are there but hard to see.)

Improve this answer
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.