This study examines word order variation in negative word + "más" constructions in Caracas Spanish, with "más" pre-posed or post-posed in relation to the negative word. We empirically analyze the effect of formal syntactic and semantic constraints, the contribution of priming and frequency, as well as several social factors on variable positioning of "más" in "NW" + "má"s constructions in Spanish. Data for this study comes from the "Estudio sociolingüístico de Caracas" (Bentivoglio and Sedano 1993). This corpus contains half-hour interviews, conducted between 1987 and 1988 with 160 speakers of Caracas Spanish. Regarding the role of syntactic and semantic formal constraints in the variation of "NW" + "más" constructions, the findings reveal that only polarity and the position of the "NW" + "más" construction relative to the verb significantly constrained variation between pre-posed and post-posed "NW" + "más" constructions. García Cornejo (2008) and Gutiérrez-Rexach and González-Rivera (2012, 2014) argued that pre-posed "más" only occurs with purely negative uses of these constructions. While our results are consistent with this observation, we also found that this pattern is not categorical, since pre-posed "más" is used in 34.3% of affirmative sentences. In order to address variation not explained by formal analyses, we explore the role of frequency to account for the variable position of "más" + "NW." Frequency appears to be a factor in this variation: the most frequent "NW" + "más" construction, with "nada," showed the highest rate of post-posing and lowest rate of pre-posing. Priming did not have a significant effect, although it trends in the expected direction: in contexts with a pre-posed prime, the rate of pre-posed "más" was higher (58.3%). We argue that analogical levelling with other pre-posed "más" constructions has led to an increased use of pre-posed "NW" + "más" constructions in Caracas Spanish and that the most frequent construction, "nada más," is resisting this analogical change due to its higher frequency and consequential entrenchment.