https://github.com/JulieBousquet/CDD_Step_Peer_Review_2018/blob/0b0d1cd9545f5ff92ab6de62ded606e0e483e281/CDD2018_DescStat.do#L284-L290
In the graph, you say that respondents are asked to rank the issue but apparently, from looking at the variable label, they are asked for each issue to rank it from Most important to less important, which is slightly different. Because of that, here it does not work I think as one respondent can select more than one issue as most important (ex: obs 2). You only select the last issue which was seen as most important and not the previous one.
I think a better way to treat it is to take for each issue the percentage of people which rated it as most important or very important out of either total respondents or number of people who rated the issue.
Also, I am not sure why you compute cleavage_n2 and n3
https://github.com/JulieBousquet/CDD_Step_Peer_Review_2018/blob/0b0d1cd9545f5ff92ab6de62ded606e0e483e281/CDD2018_DescStat.do#L284-L290 In the graph, you say that respondents are asked to rank the issue but apparently, from looking at the variable label, they are asked for each issue to rank it from Most important to less important, which is slightly different. Because of that, here it does not work I think as one respondent can select more than one issue as most important (ex: obs 2). You only select the last issue which was seen as most important and not the previous one.
I think a better way to treat it is to take for each issue the percentage of people which rated it as most important or very important out of either total respondents or number of people who rated the issue.
Also, I am not sure why you compute cleavage_n2 and n3