Carol

Looking at data

First get rid of unusable data where age is 99. There are length(filter(data, AGE>98)$AGE) such points.

Complain about conflating not wanting to reveal gender and declaring oneself as non-binary.

Complain about gender imbalance in data.

Age is an independent variable, but there are not many people older than 25: length(filter(data, AGE>25)$AGE).

There are now length(filter(data, GENDER == 1)$AGE) men. And length(filter(data, GENDER == 2)$AGE) women. Third gender category: length(filter(data, GENDER == 3)$AGE). See below for values.

length(filter(data, GENDER == 1)$AGE)

## [1] 17

length(filter(data, GENDER == 2)$AGE)

## [1] 75

length(filter(data, GENDER == 3)$AGE)

## [1] 3

length(filter(data, AGE>25)$AGE)

## [1] 8

ggplot(data, aes(AGE, CONGRUENTrt)) + 
  geom_point(aes(color=GENDER))+ scale_colour_brewer(palette = "Set2")

Density Plots

Normality looks bad

## [1] 40  2

## [1] 81  2

## [1] 1 3

## [1] 93 46

And so says shapiro test, p values are lower than 5%, this data is not normal

shapiro.test(data$CONGRUENTrt)

## 
##  Shapiro-Wilk normality test
## 
## data:  data$CONGRUENTrt
## W = 0.88135, p-value = 3.643e-07

shapiro.test(data$CONGRUENTacc)

## 
##  Shapiro-Wilk normality test
## 
## data:  data$CONGRUENTacc
## W = 0.83023, p-value = 4.567e-09

shapiro.test(data$INCONGRUENTrt)

## 
##  Shapiro-Wilk normality test
## 
## data:  data$INCONGRUENTrt
## W = 0.88477, p-value = 5.072e-07

shapiro.test(data$INCONGRUENTacc)

## 
##  Shapiro-Wilk normality test
## 
## data:  data$INCONGRUENTacc
## W = 0.87488, p-value = 1.978e-07

Best one can do is look at non-param smoothers and say data does suggest some relationship with age however reliability of these results is highly questionable due to too few data points among older adults.