When I create a factor X using summarise() and then use that factor to derive another variable Z, I realize that summarise() will not recognize the levels of factor X.

For example

# create fake dataset
tmp <- data.frame(id = 1:3,
                  x1 = c(1, 0, 0),
                  x2 = factor(c("Yes", "Yes", "No"), levels = c("Yes", "No")))
# test summarise()
tmp2 <- tmp %>%
  group_by(id) %>%
  summarise(y1 = ifelse(x1 > 0, "Yes", "No"),
            y2 = factor(ifelse(x1 > 0, "Yes", "No"), levels = c("Yes", "No")),
            y3 = ifelse(x2 == "Yes" & y1 == "Yes", "Yes", "No"),
            y4 = ifelse(x2 == "Yes" & y2 == "Yes", "Yes", "No"),
            y5 = ifelse(x2 == "Yes" & y2 == 1, "Yes", "No")
            )
tmp2

give

# A tibble: 3 × 6
     id    y1     y2    y3    y4    y5
  <int> <chr> <fctr> <chr> <chr> <chr>
1     1   Yes    Yes   Yes    No   Yes
2     2    No     No    No    No    No
3     3    No     No    No    No    No

y4 is clearly wrong, and somehow summarise() only recognizes that y2 has 2 values: 1, 2 and does not recognize its levels

My sessionInfo()

R version 3.3.1 (2016-06-21)
Platform: x86_64-pc-linux-gnu (64-bit)
Running under: Ubuntu 16.04.1 LTS
locale:
 [1] LC_CTYPE=en_US.UTF-8       LC_NUMERIC=C               LC_TIME=en_US.UTF-8        LC_COLLATE=en_US.UTF-8     LC_MONETARY=en_US.UTF-8    LC_MESSAGES=en_US.UTF-8    LC_PAPER=en_US.UTF-8
 [8] LC_NAME=C                  LC_ADDRESS=C               LC_TELEPHONE=C             LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C
attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base
other attached packages:
[1] dplyr_0.5.0