One could probably write several research papers on the topic. It would be more likely appropriate to have a discussion format, but I'll touch (not in depth) on a few of the major points:
1. Multiple popular clients implementing the same consensus algorithm. In theory, this means that bugs show up faster and you end up with something closer to the designed specification. In practice, if you ever find one tiny little difference (even a case where one has a bug and the other doesn't) you can fork the network and take advantage of the nodes that will be on the minority chain.
Eth defends this by suggesting you run multiple clients. Well, now you've doubled the resource requirements, and you force users to come up with some contingency plan in the event that the clients disagree. And if the two clients you pick are different from the two clients of much of the rest of the network, you can still fork the network.
2. Super complex contract vm. The thing so far has been a massive magnet for bugs. Bitcoin's relatively simple contact engine still has vulnerabilities being discovered in 2017 (mostly minor performance issues at this point), eth is likely to be finding bugs in their much more complicated (both theoretically and in practice) system.
3. Asic resistant PoW. No PoW is asic resistant in theory, you can always make hardware specifically specialized to your task. All it means is that eventually someone will figure out how to do it, and when they do the up-front cost may be billions of dollars - a cost that only one player will ever be able to afford, especially if they keep all their optimizations secret. Bad choice unless you think you can switch to PoS in time, but that's a bad idea for a completely separate set of reasons.
4. Centralized dev team with obvious conflicts of interest and no scruples about interfering with the networks operation to get stuff done. If a government decides to compromise when Ethereum, you could get pretty far by compromising just three people (Lubin, Wood, Buterin). Bitcoin has no such group, the devs are an amorphous blob who all suspect eachother of being CIA and would never accept a proposal merely because of the name of the submitter.
5. General support for hard forks. If you wanted to get a bad change into bitcoin, you'd mostly need to hard fork. The community actively resists hardforks. Change is very hard, but at least that means bad change is also very hard.
6. Really sketchy blockchain download. You trust a hash from a miner and then start downloading blocks after that. This means the miner can easily cheat you. Also means you can get started a lot faster, but it's a big security problem.
7. Really bad scalability. Much worse than Bitcoin's. The EVM is very expensive, and a simple transaction is going to tax you computer a lot more than on Bitcoin. Doing 1M txns per day will probably not be possible for most nodes, and Eth is within an order of magnitude of hitting. Eth also doesn't have a flagship application yet, but all of the current ones under construction do not play nice with the scaling constraints. BTC is struggling to do payments. How will will eth cope when it's doing:
Payments
Prediction markets
Name resolution
Content attribution
And who knows what else
Scalability is a security concern, because when the shit hits the fan you are going to have a bunch of incumbents pulling every which way to make scaling work. Eth has already made massive compromises to keep up with their current load. It's going to get worse. My guess is that eth will continue to favor scaling over security, and they will end up in a 12 full nodes worldwide type of situation. That's very very very much worse than Bitcoin's situation of 12 miners.
8. In general, eth culture just doesn't consider security. They think blockchains are largely magic shields that do all the hard work for you, and despite being bitten a bunch of times already they continue prioritizing innovation. Fine, but you asked why I think bitcoin is more secure, and a culture of mistrust and slow adoption goes a long way to achieving security.
I could probably find another 10 things to talk about but I'll stop here. Ethereum has selling points over Bitcoin, but they are not security or decentralization by a long shot.