# Analyze and visualize participants response towards particular condition

Method: I presented 15 participants with audiovisual clips. There were six different clips I presented, and for each clip there was emotional incongruence between visual and auditory information. The combinations were:

• visual negative with auditory positive
• visual negative with auditory neutral
• visual positive with auditory negative
• visual positive with auditory neutral
• visual neutral with auditory negative
• visual neutral with auditory positive

I asked participants to make an emotional judgements, whether the clip was happy, angry or neutral. There were five repetitions.

Research question: Are participants driven by visual or auditory information in their emotional choices, when they watch displays with incongruent emotional information?

Basic analysis: Because there were no “correct” responses per se for the incongruent stimuli, I calculated the tendency to respond correctly when emotion was presented auditorily or visually. The tendency was estimated by subtracting the proportion of “auditory correct” responses from the proportion of “visual correct” responses for the six incongruent conditions. For example if the incongruent display combined visual negative with auditory positive information, and participants responded that it was a negative interaction, then it was counted as visual response. The estimated indices varied from 1 (participants always responded correctly to the visual information) to −1 (participants always responded correctly to the auditory information). Below is a figure produced in R that shows averaged tendency scores, with standard error bars. The data is available here - columns represent conditions, rows - tendency scores for each participant.

Now, getting to the point I have two questions:

Question 1: What would be the best way to check for significant differences between those different clips?

I know I could simply run a series of 30 paired t test's but it doesn't seem as a pretty solution. I don't think I can use ANOVA here - I have no clue how I could define levels for all those incongruent conditions. Maybe I could organize this data in some different way, but I really not sure what it would be.

Question 2: Is there a better way you could use to visualize this?

I don't love the above plot - in fact I really don't like it. The oversized legend is very heavy to read, and so are the patterns differentiating each condition. I could use color instead, but I am a bit color blind to be honest, so I would prefer to avoid it.

EDIT

Question 1 I decided to just compare the 6 pairs which had the same type of incongreunt information using paired t test. Works kind-of-ok to get the differences.

Question 2 has been answered well by @AndyW and @gung - I decided to use pure SE bars with vertical orientation of x axis. Done in R using segplot() - I need to tweak the details, but it's roughly the idea.

For two, avoid dynamite plots (see Drummond & Vowler, 2011), and utilize dot plots since you only have 15 participants. You can super-impose confidence lines over the dot plots, and you can create a category axis label to label the dots/bars/lines, foregoing the need to differentiate between categories using color, point/line symbols or other hashings.

I will post back an example using your data later, but for now the paper cited above has several examples perfectly applicable to your situation, and one is inserted below.

Since you tagged the question r, this previous question has applicable code snippets to generate similar charts, Alternative graphics to “handle bar” plots.

# Citation

Drummond, Gordon B. & Sarah L. Vowler. 2011. Show the data, don't conceal them. The Journal of Physiology 598(8): 1861-1863. PDF available from publisher.

Below is an example extended to your data. I have posted full examples of generating similar plots in R using ggplot2 and in SPSS on my blog in this post, Avoid Dynamite Plots! Visualizing dot plots with super-imposed confidence intervals in SPSS and R.

• Another note, a recent question on stack overflow demonstrates how these types of dot plots can even be effectively extended when the number of data elements is much larger using jittering and making the points semi-transparent, ggplot2: Multiple color scales or shift colors systematically on different layers?. – Andy W Feb 13 '12 at 15:15
• Thank you, this is excellent solution for the visualization problem. I am a bit worried only in the sense that studies I am quoting heavily are all using standard dynamite plots, but I guess one has to push better solutions in order to break the bad habits in reporting effects... – Geek On Acid Feb 13 '12 at 16:05
• Did you used stripchart() or something else to produce this in R? And what function did you used to plot error bars? – Geek On Acid Feb 15 '12 at 1:04
• @GeekOnAcid, the above chart is produced in SPSS. I did experiment and have a similar chart using ggplot2 in R, but I'm currently stuck in producing a legend in either! I will have the opportunity to update with code I used to produce said charts tomorrow. – Andy W Feb 15 '12 at 1:40
• @GeekOnAcid, here I have posted some code to generate a similar plot in ggplot2 in R, I will I will let you know in an update when I figure out how to generate a legend. – Andy W Feb 15 '12 at 13:35

@AndyW has a good answer. I think dot plots, or even box plots are good approaches, although I think bar graphs are OK. One thing I would recommend is that you rotate your figure 90 degrees. Then, more visual would go to the right, and more auditory would go to the left. The advantage of this is that you could drop the legend and list your conditions on the left hand side. They would be easy to read, because they would be aligned with their corresponding graph elements (whatever you end up going with), and because people read horizontally from left to right. You could probably wrap the text, but you could also abbreviate somewhat, perhaps "Vis - & Aud +" (you'd need to come up with something for neutral). Thus, the figure needn't take up any more space than it does now.