Why is this else statement being considered first than one of the if values above?

This function loops through an array of strings. If the string doesn't start with an # it's wrapped with a p tag. If the previous string of the string is a # or a * * * it's wrapped with a p tag with the class ni

function replaceParagraphs(data) {
  data = data.split('\n\n')
  for (var i = 0; i < data.length; ++i) {
    if (data[i].match(/^[^#]/g)) {
      var content = data[i].replace(/\n(?=\w)/g, ' ').replace(/\n(?=$)/g, '')
      if (data[i - 1].match(/^#/g) || data[i - 1] === "* * *") {
        data[i] = '<p class="ni">' + content + '</p>'
      } else {
        data[i] = '<p>' + content + '</p>'
      }
    }
  }
  data = data.join('\n\n')

  return data
}

But it seems like data[i - 1] === "* * *" is not being considered, since I end up with an output like this:

## Test

<p class="ni">Lorem “ipsum?” dolor ‘sit!’ amet</p>

<p>Consetetur eirmod</p>

<p>* * *</p>

<p>“Ipsum?” “‘dolor’” sit! amet, consetetur eirmod tempor—invidunt ut labore</p>

What could be the problem?

EDIT:

Oh, based on the answers and comments I realized that it's because the code produces <p>* * *</p>

What can I do to prevent this?