1:Accept with final revisons

(Reviewer 1)
1. In the process of calculating the moderating variable (CentralityC), is the network constructed based on the time when each fundraiser launches a project? What is the frequency of network updates?
2. Given the small values of the coefficients for GapAbove and GapBelow, it is necessary to discuss the implications of the results, which would help in understanding the empirical results. In Table 3, should “Extreme value removal: Project level” be changed to "Extreme value removal: Fundraiser level?"
(Reviewer 2)
1. Thank you for incorporating the exploration versus exploitation dilemma to support your arguments. However, in my opinion, the logic of exploration versus exploitation doesn't align well with your hypotheses. This dilemma emphasizes choosing between exploring new possibilities and exploiting existing resources. However, crowdfunding creators have the option to opt out of creating new projects regardless of their previous performance.
2. Additionally, while your hypotheses focus on the exploration aspect and why creators explore new categories (a novel aspect of your paper), you've overlooked the exploitation aspect. As you mentioned in your data, only 20% of observations (6972 out of 34883) exhibit exploratory behavior. What about the remaining 80% who chose to exploit existing categories? Since they constitute the majority of your observations, it's essential to include your analysis of exploitation behavior if you intend to use the exploration versus exploitation dilemma as your theoretical foundation.
3. Since you mentioned making a contribution to performance feedback theory in your introduction, why is there no literature review concerning this theory? Why are you choosing the exploration versus exploitation dilemma instead of performance feedback theory as your theoretical framework?
(Reviewer 3)
1. I am still not convinced that a category switch for a new project is purely based on the performance above or below the target level of the creator's previous project. It may be due to many other factors, as I mentioned, such as the creator's social capital and expertise. I understand the constraints of secondary data, but I believe you are able to collect some important variables via Kickstarter, such as social capital (measured by the creator's number of Facebook friends, and how many collaborators are on this project), etc.
2. Additionally, I highly suggest the authors conduct another study to confirm the reasons for the creator's category switch. This could be done through interviews or surveys with the creators, or via Mturk or other survey collection platforms. In order to draw conclusions about findings, the authors need to obtain survey results to confirm whether the creator's actual thoughts are consistent with the secondary data and their main assumptions.
3. You need further justification for your current data cleaning process. Do you have control over the target value? What if the target value was set too high or too low? What threshold is supported by the literature?
4. Delete the following paragraph: "Our study relates to three streams of literature: (1) organizational learning, (2) entrepreneurial exploration, and (3) crowdfunding platforms. To place our contributions in perspective, we review the relevant work in these areas." Change it into subtitles instead.