// includes, etc

class WordGenerator {
  public:
    // Constructor, etc;
    void setRandWord();
    void setBlankWord();
    void newWord();
    // Accessors, etc
  private:
    // Other member variables, etc
    std::string m_randWord;
    std::string m_blankWord;;
};

WordGenerator::WordGenerator( // parameters ) {
  std::srand( std::time(NULL) );
  // Code to initialize variables and vectors, etc
  WordGenerator::setRandWord();
  WordGenerator::setBlankWord();
}

void WordGenerator::setRandWord() {
  m_randWord = m_dictionaryVector[ std::rand() % m_dictionaryVector.size() ];
}

void WordGenerator::setBlankWord() {
  std::string blankStr( m_randWord.length(), '_' );
  m_blankWord = blankStr;
}

void WordGenerator::newWord() {
  setRandWord();
  setBlankWord();
}

// Other declarations, etc
/*
  I'm asking which is best? When I call newWord(), is it possible that something happens between
  the execution of setRandWord() and setBlankWord() that would cause a mismatch? It is better to do something like this:
  void WordGenerator::newWord() {
    m_randWord = m_dictionaryVector[ std::rand() % m_dictionaryVector.size() ];

    std::string blankStr( m_randWord.length(), '_' );
    m_blankWord = blankStr;
  }
  Then just call newWord() in the constructor? I can't imagine a situation where I would call setRandWord() without needing to call setBlankWord().
*/