SELECT t.value, count(*) as users_count
FROM users u
JOIN (
SELECT d1.text as value, d1.user_id FROM data1 d1
UNION
SELECT d2.name as value, d2.user_id FROM data2 d2
) t on t.user_id = u.id
GROUP BY t.value
ORDER BY users_count DESC
select d.value,
count(*) users_count
from user u
left join (select user_id, data1.text value from data1
union all
select user_id, data2.name value from data2
) d on d.user_id = u.id
group by d.value
SELECT `experience`.`role` AS role, COUNT(*) AS count_users
FROM `experience`
GROUP BY `role`
UNION
SELECT 'Not role',
( SELECT COUNT(*) FROM `users` ) - ( SELECT COUNT( DISTINCT( `user_id` )) FROM `experience` )
ORDER BY 2 DESC
