A researcher conducted a survey on perceptions of the strength of the economy and the favourability rating of the current Prime Minister. Although she expected to find a positive relationship, to her surprise, she found no relationship as the Prime Minister's average favourability rating was 60% irrespective of whether respondents said the economy was Worse, the Same, or Better than last year. However, the researcher suspects that once she controls for the political leanings of the survey respondent, her results may change. Data on political leanings was collected by having survey respondents listing which political party they most closely identify with: the NDP, Liberals, and Conservatives. Why might controlling for party identification change the initial results of the study.