Maybe you can visualize your data. You might communicate your findings better using graphs instead of showing four separate frequency tables. Here are some ideas:
Using two tables on the left (e.g., hellbender bite tables), I recreate part of your data. So, you have three variables, namely, level of education (ordinal variable with 7 categories, but the first one is empty), (likelihood of) Hellbender bite (again ordinal, likert-like item, 5 categories), and familiarity (binary variable). First, you want to know whether familiarity with the species makes a difference for the belief in the likelihood of hellbender bite.
You can create a bar plot for each category of the familiarity using Hellbender bite. I share the R code just in case:
library(tidyverse)
library(ggplot2)
dataset %>%
ggplot(aes(Familiarity, fill=Hellbender)) +
geom_bar(aes(y = (..count..)/sum(..count..)), position="dodge") +
scale_fill_brewer(type="seq", palette = "Oranges") +
labs(x=" ", y=" ") +
guides(fill=guide_legend(title="Hellbender Bite")) +
scale_y_continuous(labels = scales::percent) +
theme_minimal()
Please note that this graph uses cell percentages. You can also use counts, or as in the following example, the percentage distributions within the categories of familiarity (changing the code to position="fill"):
Now, these graphs give us some idea. For instance, the percentage of respondents reporting hellbender bite is unlikely is higher among those who are familiar with the species compared those who are unfamiliar. But instead of just eyeballing the graphs, you might want to test (e.g., using $\chi^2$ test of association) whether there is an association between these two variables.
What about the influence of education? We can incorporate information from the education variable into our graph. Here is one way to do that:
dataset %>%
ggplot(aes(Education, fill=Hellbender)) +
geom_bar(aes(y = (..count..)/sum(..count..))) +
scale_y_continuous(labels = scales::percent) +
facet_grid(~Familiarity) +
scale_fill_brewer(type="seq", palette = "OrRd") +
labs(x=" ", y=" ") +
theme_minimal() +
scale_x_discrete(labels = c("High\nSchool", "Some\nCollege",
"Associate\nDegree","Bachelor's\nDegree",
"Master's\nDegree", "Doctorate")) +
guides(fill=guide_legend(title="Hellbender Bite"))
On the x axis, we have categories of the education level now, but the graph has somewhat limited usefulness. We can tweak it a little further:
dataset %>%
ggplot(aes(y = reorder(Education, desc(Education)))) +
geom_bar(aes(fill = Hellbender),
position = position_fill(reverse = TRUE)) +
scale_fill_manual(values=c("#FEF0D9", "#FDCC8A",
"#FC8D59","#E34A33","#B30000"),
name="Hellbender Bite",
labels=c("Very unlikely", "Somewhat unlikely",
"Average risk","Somewhat likely",
"Very likely"))+
scale_x_continuous(labels = scales::percent) +
theme_minimal() +
theme(legend.position = "bottom") +
labs(x=" ", y=" ")+
facet_grid(~Familiarity)
It does not tell us a detailed story, but across the categories of education (except high school), only a small portion of the respondents familiar with the species report average and higher risk of Hellbender bite (you can tell a little more looking at these graphs, but one should be wary of over-interpretation).
However, it seems like you ask about something more: How does the relationship between familiarity and belief in the likelihood of hellbender bite vary by education? Now, it is hard to show whether such a variation exists using graphs and tables. So, you might want to model these relationships by including an interaction effect. But that also means going beyond reporting descriptive statistics.
