In a civil legal case, the plaintiff usually has to prove their case at a “more likely than not” standard (>50%) in order to win their claim. Legal claims typically involve multiple individual elements that also need to be proven to the same standard. In a civil claim with two elements relying on independent evidence, if the likelihood of each element was 60%, it is debatable whether the claim should win or lose. Each element has met the threshold and intuition suggests that the claim should win. However, probability theory would suggest multiplying these probabilities to reach a 36% likelihood of the overall case, which would dictate a loss for the plaintiff. Some state legal instructions contain a “conjunction problem” by stating both that a decision-maker should decide the claim based on “all of the evidence” as well as a statement that the claim is required to win if “all of the elements” are found to be true. Across four studies we examine how lay participants choose to combine elements into overall case-level decisions. We find very few people naturally follow a multiplicative combination rule across various contexts. In Studies 1 and 2 participants read detailed case descriptions and estimate their personal element probabilities. In Study 1 we find participants are sensitive to case strength details, but provide equivalent answers to two-element versus four-element cases, which sharply violates the multiplication rule. In Study 2 we do not observe differences based on whether or not the conjunction problem is included in the legal instructions. In Studies 3 and 4 participants read abstracted case descriptions and we provide stipulated element probabilities across multiple cases, allowing categorization of participants into combination strategies. In Study 3 we stipulate numeric probabilities and find that an introductory tutorial about probabilities reduces randomness and increases the use of conjunctive multiplication. In Study 4 we stipulate verbal probabilities and find that translating these verbal probabilities into numeric ones on each case page reduces randomness. Across these studies we find that most people are averaging the strength of evidence for the elements to reach an overall conclusion about the case, and that the conjunction-multiplying rule is followed by less than 10% of the respondents.