2B Lab 7 Week 9

This is the pair coding activity related to Chapter 13.

Task 1: Open the R project for the lab

Task 2: Create a new .Rmd file

Task 3: Load in the library and read in the data

The data should already be in your project folder. If you want a fresh copy, you can download the data again here: data_pair_coding.

We are using the packages afex, tidyverse, and performance today, and need to read in dog_data_clean_long.

library(afex)
library(tidyverse)
library(performance)

dog_data_long <- read_csv("dog_data_clean_long.csv")

Task 4: Tidy data & Selecting variables of interest

Let’s define a potential research question:

How does the Year of Study (1st, 2nd, 3rd, 4th, 5th or above) and Time point (pre- and post-intervention scores) affect perceived Happiness?

This is a 2 x 5 mixed factorial ANOVA, with one (Stage: pre vs. post) and one (year of study: 1st, 2nd, 3rd, 4th, 5th or above). The variable is self-reported Happiness Score.

Not much tidying to do today. All the variables and average scores are already tidied up in dog_data_long. We just need to check for missing values in our variables of interest. Best to do this on a reduced dataframe.

  • select the variables of interest and store them in a new data object called dog_factorial_anova
  • check for missing values and remove participants who did not have pre- and post-intervention Happiness recorded
  • You need 4 variables: Participant ID, Year of Study, Time point of recording (i.e., pre- or post-intervention), and the self-reported Happiness scores.
  • You should have identified 1 participant who did not complete both ratings. There are 2 options on how to achieve that:
    • Option 1: identify the missing value visually and filter out their participant ID (works well with smaller dataframes, but not very reproducible)
    • Option 2: remove the missing value line first and then add a new columns to count how often each participant is now represented in the data object and then only include Participants with a count of 2 ratings in this case (which is the more reproducible option and more appropriate with larger dataframes or when there are loads of missing values).
## Option 1
dog_factorial_anova <- dog_data_long %>%
  select(RID, Year_of_Study, Stage, SHS) %>% 
  filter(RID != 12)

## Option 2
dog_factorial_anova <- dog_data_long %>%
  select(RID, Year_of_Study, Stage, SHS) %>% 
  drop_na() %>% 
  group_by(RID) %>% 
  mutate(RID_count = n()) %>% 
  filter(RID_count == 2)

Task 5: Model creating & Assumption checks

Now, let’s create our ANOVA model.

According to our research question we have the following model variables (use the spelling of the variable names in dog_factorial_anova):

  • Dependent Variable (DV): , assessed pre- and post-intervention
  • Independent Variable 1 (IV 1): (1st, 2nd, 3rd, 4th, 5th or above)
  • Independent Variable 2 (IV 2): (pre- and post-intervention scores)

As a reminder, the ANOVA model has the following structure:

mod <- aov_ez(id = "NULL",
       data = NULL, 
       between = "NULL", 
       within = "NULL",
       dv = "NULL", 
       type = 3,
       anova_table = list(es = "NULL"))

Let’s use this template to fill in our variables of interest and store the model in a separate object called mod. The effect size should be partial eta squared ("pes"):

mod <- aov_ez(id = "RID",
       data = dog_factorial_anova, 
       between = "Year_of_Study", 
       within = "Stage",
       dv = "SHS", 
       type = 3,
       anova_table = list(es = "pes"))

Let’s check some assumptions.

check_model(mod)

check_homogeneity(mod, method = "levene")
OK: There is not clear evidence for different variances across groups (Levene's Test, p = 0.882).

Are the following assumptions met or violated?

  • Assumption 1: Continuous DV?
  • Assumption 2: Data are independent?
  • Assumption 3: Normality?
  • Assumption 4: Homoscedasticity?

Task 6: Interpreting the output

Call the model object to view the ANOVA results:

mod
Anova Table (Type 3 tests)

Response: SHS
               Effect     df  MSE         F  pes p.value
1       Year_of_Study 4, 278 1.34      1.58 .022    .179
2               Stage 1, 278 0.09 19.86 *** .067   <.001
3 Year_of_Study:Stage 4, 278 0.09      0.45 .006    .772
---
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '+' 0.1 ' ' 1

How do you interpret the results?