Hello Everyone,
I created a dataset a while ago by merging several cross-sectional datasets into panel data. After receiving feedback, I redid everything from scratch, following the same steps but writing a new, more organized do-file to ensure reproducibility. I’m using the PSID datasets, just for context. However, now that I’m working with this new dataset, the results I’m getting are completely different from those I obtained previously—many of my significant findings have become insignificant.
I’ve tried every possible method to pinpoint what might be causing the difference, but I haven’t been able to figure out why I can't replicate my original results. The only time I get the same results back is when I use the older merged dataset, not the newly created one. Even after using commands like summarize, the statistics for both datasets seem similar.
Does anyone have any suggestions on how I can identify the differences between the two datasets? I might be overlooking something, and I would appreciate any ideas or insights you may have!
Thank you!
I created a dataset a while ago by merging several cross-sectional datasets into panel data. After receiving feedback, I redid everything from scratch, following the same steps but writing a new, more organized do-file to ensure reproducibility. I’m using the PSID datasets, just for context. However, now that I’m working with this new dataset, the results I’m getting are completely different from those I obtained previously—many of my significant findings have become insignificant.
I’ve tried every possible method to pinpoint what might be causing the difference, but I haven’t been able to figure out why I can't replicate my original results. The only time I get the same results back is when I use the older merged dataset, not the newly created one. Even after using commands like summarize, the statistics for both datasets seem similar.
Does anyone have any suggestions on how I can identify the differences between the two datasets? I might be overlooking something, and I would appreciate any ideas or insights you may have!
Thank you!
Comment