Smith conducted a poll that was published by Globes on Jan 31 2014.
Current Knesset seats in [brackets]
36 [31] Likud Beitenu
17 [15] Labor
12 [19] Yesh Atid
11 [12] Bayit Yehudi
11 [11] Shas
10 [06] Meretz
08 [07] Yahadut Hatorah/UTJ
04 [06] Movement
11 [11] Hadash, Balad and Ra’am-Ta’al
00 [02] Kadima
66 [61] Right-Religious
54 [59] Center-Left-Arab
Knesset Jeremy Analysis: The Smith poll is different than the Panels poll. Smith’s poll was conducted before the “apology” and the Panels poll was conducted afterwards. Smith shows a larger right-religious block than the Panels poll, with higher numbers for Likud Beitenu, Shas & UTJ. Panels poll gives Bayit Yehudi 17 seats and Smith gives Bayit Yehudi 11 seats.
Jeremy's Knesset Insider 
As the results are quite different between different polling companies; one question: do they have own voter databases?
The answer is that it depends on the polling company.
So do you mean that the polling company
a) use the “usual” database and then ask the questions in some certain way to get some certain results (i.e. the company’s partisan position impacts the question setup
b) use even a database “of their own” to make certain that the results really follow the company’s partisan position.
As a comparison for this question, look at how polling is done in the USA. Certain companies like PPP, Quinnipiac, ABC news, CNN etc. always gives democrats an advantage for example because of the sample; often a too “overweighted” democratic part in the sample. On the other hand one can see (at least that is what some people claim) that i.e. Rasmussen reports is too pro-GOP and oversamples (if that even is true; not in mine opinion).