ggplot(wisqars.suicide, aes(Year, Crude.Rate, color=Region)) +
stat_summary(fun.data=mean_se, geom="pointrange") +
stat_summary(aes(y=fitted(m.1)), fun.y=mean, geom="line") +
scale_x_continuous(breaks=1999:2007)
ggplot(NamingRecovery, aes(TestTime, Correct, color=Diagnosis)) +
stat_summary(fun.data=mean_se, geom="pointrange") +
stat_summary(aes(y=fitted(nr.1)), fun.y=mean, geom="line") 
CP (d’ peak at category boundary): Compare categorical perception along spectral vs. temporal dimensions using second-order orthogonal polynomials.
summary(CP)## Participant Type Stimulus d.prime
## 1 : 8 Temporal:128 Min. :1.00 Min. :-1.41
## 2 : 8 Spectral:128 1st Qu.:2.75 1st Qu.: 0.40
## 3 : 8 Median :4.50 Median : 1.18
## 4 : 8 Mean :4.50 Mean : 1.34
## 5 : 8 3rd Qu.:6.25 3rd Qu.: 2.22
## 6 : 8 Max. :8.00 Max. : 4.77
## (Other):208ggplot(CP, aes(Stimulus, d.prime, color=Type)) +
stat_summary(fun.data=mean_se, geom="pointrange")
ggplot(CP, aes(Stimulus, d.prime, color=Type)) +
stat_summary(fun.data=mean_se, geom="pointrange") +
stat_summary(aes(y=fitted(m.2)), fun.y=mean, geom="line")
Az: Use natural (not orthogonal) polynomials to analyze decline in performance of 30 individuals with probable Alzheimer’s disease on three different kinds of tasks.
summary(Az)## Subject Time Task Performance
## 1 : 30 Min. : 1.0 cADL :300 Min. : 2.0
## 2 : 30 1st Qu.: 3.0 sADL :300 1st Qu.:40.0
## 3 : 30 Median : 5.5 Memory:300 Median :52.0
## 4 : 30 Mean : 5.5 Mean :49.3
## 5 : 30 3rd Qu.: 8.0 3rd Qu.:61.0
## 6 : 30 Max. :10.0 Max. :85.0
## (Other):720ggplot(Az, aes(Time, Performance, color=Task, fill=Task)) +
stat_summary(fun.data=mean_se, geom="ribbon", color=NA, alpha=0.5) +
stat_summary(fun.y=mean, geom="line")
ggplot(Az, aes(Time, Performance, color=Task)) +
stat_summary(fun.data=mean_se, geom="pointrange") +
stat_summary(fun.y=mean, geom="line", aes(y=fitted(m.Az.full)))
Try GCA on your own data, ask for help if you need it.
If you don’t have any appropriate data on hand, try FunctTheme: Use orthogonal polynomials to compare “Function” (broom - vacuum cleaner) and “Thematic” (broom - dustpan) semantic competition.
ggplot(FunctTheme, aes(Time, meanFix, color=Object, fill=Object)) +
facet_wrap(~ Condition) + stat_summary(fun.y=mean, geom="line") +
stat_summary(fun.data=mean_se, geom="ribbon", color=NA, alpha=0.5) +
theme_bw() + labs(y = "Fixation Proportion")
Make a graph
# term labels
trm <- c("Intercept", "poly1", "poly2", "poly3", "poly4", "Object", "Condition",
"poly1:Object", "poly2:Object", "poly3:Object", "poly4:Object",
"poly1:Condition", "poly2:Condition", "poly3:Condition", "poly4:Condition",
"Object:Condition", "poly1:Object:Condition", "poly2:Object:Condition",
"poly3:Object:Condition", "poly4:Object:Condition")
# assemble terms labels and parameter estimates from both models into one data frame
coefs <- data.frame(Term=ordered(rep(trm, 2), levels=rev(trm)),
Model = rep(c("Initial", "Revised"), each=20),
Estimate = c(coef(summary(m.ft))[, 1], coef(summary(m.ft2))[, 1]),
SE = c(coef(summary(m.ft))[, 2], coef(summary(m.ft2))[, 2]))
# plot the parameter estimates
ggplot(coefs, aes(Term, Estimate, color=Model)) +
geom_pointrange(aes(ymin=Estimate-2*SE, ymax = Estimate+2*SE),
position=position_dodge(width=0.4)) +
geom_hline(yintercept=0) + coord_flip()
This figure shows all of the coefficient estimates (+/- 2SE) for the two models. There are two key differences:
Object coefficient refers to the main effect of competition (competitor vs. unrelated) on the intercept across both conditions and is statistically significant (SE range does not include 0). In the initial model, the Object coefficient is the simple effect in the Function condition and does not reach statistical significance (SE range includes 0).Condition coefficients refer to the simple effect of condition (function vs. thematic) on the unrelated object and none of them are statistically significant, so there were no systematic differences in the time course of unrelated object fixations between conditions (as expected). In the initial model, the Condition coefficients refer to the simple effect of condition on the competitor object and some of the coefficients are statistically significant (effects on the cubic and quartic term) reflecting differences in competitor fixation between the two conditions, although the Object-by-Condition interaction coefficients better capture that effect.ggplot(WordLearnEx, aes(Block, Accuracy, color=TP)) +
stat_summary(fun.data = mean_se, geom="pointrange") +
stat_summary(aes(y=fitted(wl.logit)), fun.y = mean, geom="line") +
stat_summary(aes(y=fitted(wl.lin)), fun.y = mean, geom="line", linetype="dashed") +
theme_bw() + expand_limits(y=c(0.5, 1.0)) + scale_x_continuous(breaks=1:10) +
scale_color_manual(values=c("red", "blue")) + labs(y="Accuracy")
Make a multi-panel plot showing each kind of competition for each group
ggplot(CohortRhyme, aes(Time, FixProp, color=Object)) +
facet_grid(Type ~ Group) + stat_summary(fun.y=mean, geom="line") +
stat_summary(fun.y=mean, geom="point") + theme_bw()
Make a multi-panel scatterplot showing these correlations.
ES.m <- melt(ES, id=c("subjID", "Group"))
ES.m <- cbind(ES.m, colsplit(ES.m$variable, "\\.", c("Type", "Term")))
ES.c <- dcast(ES.m, subjID + Group + Term ~ Type)
ggplot(ES.c, aes(Cohort, Rhyme, color=Group)) +
facet_wrap(~ Term, scales="free") + geom_point() +
scale_color_manual(values=c("grey", "red", "blue")) +
theme_bw()