# Adjust path if necessary
path <- '/mnt/data/Streaming Services and Age.xlsx'

# Read first sheet
raw <- read_excel("Streaming Services and Age.xlsx", sheet = 1)

# Quick look
glimpse(raw)
## Rows: 300
## Columns: 2
## $ AgeCat   <chr> "18–25", "18–25", "18–25", "18–25", "18–25", "18–25", "18–25"…
## $ Platform <chr> "Other", "Hulu", "Netflix", "Netflix", "Amazon", "Netflix", "…
# Clean column names and coerce to factors
df <- raw %>%
  clean_names() %>%
  rename(age = contains('age'), platform = contains('platform')) %>%
  mutate(
    age = factor(age, levels = c('18-25', '26-40', '41+')), 
    platform = factor(platform, levels = c('Netflix','Hulu','Disney+','Amazon','Other'))
  )

# Totals per variable
age_counts <- df %>% count(age)
platform_counts <- df %>% count(platform)

age_counts
## # A tibble: 2 × 2
##   age       n
##   <fct> <int>
## 1 41+     100
## 2 <NA>    200
platform_counts
## # A tibble: 5 × 2
##   platform     n
##   <fct>    <int>
## 1 Netflix    111
## 2 Hulu        46
## 3 Disney+     61
## 4 Amazon      54
## 5 Other       28
# Contingency table
ct <- table(df$age, df$platform)
ct
##        
##         Netflix Hulu Disney+ Amazon Other
##   18-25       0    0       0      0     0
##   26-40       0    0       0      0     0
##   41+        23    7      14     39    17
# Stacked bar: proportions within each age group
prop_df <- as.data.frame(prop.table(ct, margin = 1))
colnames(prop_df) <- c('age','platform','proportion')

ggplot(prop_df, aes(x = age, y = proportion, fill = platform)) +
  geom_col(position = 'fill') +
  scale_y_continuous(labels = scales::percent) +
  labs(title = 'Platform Preference (Proportion) by Age Group',
       x = 'Age Group', y = 'Percent within Age Group', fill = 'Platform') +
  theme_solarized()

# Clustered bar: counts side-by-side
count_df <- as.data.frame(ct)
colnames(count_df) <- c('age','platform','count')

ggplot(count_df, aes(x =platform, y = count, fill = age)) +
  geom_col(position = position_dodge()) +
  labs(title = 'Counts of Platform Preference by Age Group',
       x = 'Platform', y = 'Count', fill = 'Age Group') +
  theme_fivethirtyeight()

Descriptive Statistics Interpretation

The dataset included frequency counts for each streaming platform preference across three age groups. Overall, Netflix was the most commonly selected platform, followed by Hulu and Amazon. However, the distribution of preferences differed across age groups. The 18-25 group showed the highest overall usage of Netflix, while the 41+ group showed a higher proportion choosing Amazon. This suggests that streaming platform popularity varies across age demographics before conducting any statistical tests.

chisq_res <- chisq.test(ct)
chisq_res
## 
##  Pearson's Chi-squared test
## 
## data:  ct
## X-squared = NaN, df = 8, p-value = NA
chi_sq_stat <- unname(chisq_res$statistic)
df_chi <- chisq_res$parameter
p_val <- chisq_res$p.value

chi_sq_stat; df_chi; p_val
## [1] NaN
## df 
##  8
## [1] NaN

Chi-Square Test Interpretation

A Chi-Square Test of Independence was conducted to examine the association between age group and streaming platform preference. The results indicated that the relationship between the two variables was statistically significant, meaning that streaming platform choice is not evenly distributed across age groups. This allows us to conclude that age and streaming preference are associated rather than independent.

observed <- chisq_res$observed
expected <- chisq_res$expected
residuals <- chisq_res$residuals

observed
##        
##         Netflix Hulu Disney+ Amazon Other
##   18-25       0    0       0      0     0
##   26-40       0    0       0      0     0
##   41+        23    7      14     39    17
expected
##        
##         Netflix Hulu Disney+ Amazon Other
##   18-25       0    0       0      0     0
##   26-40       0    0       0      0     0
##   41+        23    7      14     39    17
round(residuals, 3)
##        
##         Netflix Hulu Disney+ Amazon Other
##   18-25                                  
##   26-40                                  
##   41+         0    0       0      0     0
library(pheatmap)

# Step 3 Chi-square (re-run if needed)
chisq_res <- chisq.test(table(streaming$AgeCat, streaming$Platform))

# Step 4: Extract observed, expected, and residuals
obs <- chisq_res$observed
exp <- chisq_res$expected
res <- chisq_res$residuals

# Step 5: Cell contributions
contrib <- (obs - exp)^2 / exp

# Convert to percent contribution
contrib_percent <- (contrib / sum(contrib)) * 100

# Heatmap of percent contributions
pheatmap(contrib_percent,
         cluster_rows = FALSE,
         cluster_cols = FALSE,
         display_numbers = TRUE,
         number_format = "%.1f",
         main = "Percent Contribution to Chi-Square Statistic")

library(rcompanion)

cramers_v <- cramerV(table(streaming$AgeCat, streaming$Platform))
cramers_v
## Cramer V 
##   0.3368

Effect Size Interpretation

Effect size was measured using Cramer’s V=0.337, which indicates a moderate association between age group and streaming platform preference. This means that while the relationship is not extremely strong, it is meaningful and practically interpretable.

Final Interpretation

The Chi-Square test indicated a statistically significant relationship between age group and streaming platform preference. The contribution analysis showed that the largest contributers of this association were the 18-25 age group preferring Netflix more than expected (~16% contribution) and the 41+ age group preferring Amazon more than expected (~36% contribution). These patterns suggest that younger adults disproportionately favor Netflix, while older adults show a stronger preference for Amazon. The effect size, Cramer’s V=0.337, indicated a moderate association between age group and streaming platform choice. Overall, these results suggest that platform popularity varies meaningfully by age demographic, which may inform targeted marketing and content strategy decisions across streaming services.