Previous node types are inferior to current generation in terms of performance and fee. Unless there is a special reason, you should avoid to use these ones. It is recommended to select a current generation node type according to the upgrade paths.
resource "aws_elasticache_cluster" "redis" {
cluster_id = "app"
engine = "redis"
engine_version = "3.2.4"
maintenance_window = "sun:00:00-sun:06:00"
node_type = "cache.t1.micro" // previous node type!
num_cache_nodes = 1
port = 6379
parameter_group_name = "default.redis3.2"
subnet_group_name = "app-subnet-group"
security_group_ids = ["${aws_security_group.redis.id}"]
}