Which came first: the readiness or the change? Longitudinal relationships between readiness to change and drinking among college drinkers

Abstract

Aims: Although readiness to change (RTC) is cited as a key mechanism underlying drinking behavior change, empirical evidence supporting RTC as a predictor of college drinking has been mixed. Considering methodological limitations of previous studies, the current aim was to conduct a more comprehensive test of longitudinal relationships between readiness to change and college drinking.

Design: In this correlational, longitudinal study, we used a series of cross-lagged path analyses to test associations between RTC and college drinking outcomes over a 2-year period.

Setting: Data collection was conducted via online surveys on a university campus in the US Pacific Northwest.

Participants: Participants (n = 818; 58% women) were college students who reported at least one heavy-drinking episode in the past month and participated in a randomized controlled trial of personalized normative feedback interventions.

Measurements: Drinking quantity-frequency items and the Rutgers Alcohol Problem Index assessed drinking outcomes. The Readiness to Change Questionnaire assessed RTC.

Findings: For drinking-related problems, the best-fitting model included cross-lagged paths between RTC and subsequent drinking-related problems. For drinking quantity-frequency, best-fitting models also included the cross-lagged paths between drinking quantity-frequency and subsequent RTC. Higher RTC almost uniformly predicted higher subsequent levels of drinking and greater experience of drinking-related problems, and drinking quantity-frequency variables were primarily positive predictors of subsequent RTC.

Conclusions: Contrary to previous assumptions, 'the Readiness to Change Questionnaire' does not appear to be predictive of lower levels of subsequent drinking.

Figure 1

The reduced model showing autoregressive and temporal stability effects of readiness to change (RTC) and drinking variables at each time-point provided the base model for all analyses. Statistically significant, positive paths are denoted with a ‘+’ sign. Time-point 1: baseline; time-point 2: 6-month follow-up; time-point 3: 12-month follow-up; time-point 4: 18-month follow-up; time-point 5: 24-month follow-up; E: measurement error

Figure 2

Significant hypothesized model adding the cross-lagged paths from readiness to change (RTC) to subsequent drinking outcome variables to the autoregressive model. Because the model included both count and continuous outcomes, path coefficients in the white boxes represent standardized beta coefficients, whereas path coefficients in the gray boxes represent incident rate ratios (IRRs). Time-point 1: baseline; time-point 2: 6-month follow-up; time-point 3: 12-month follow-up; time-point 4: 18-month follow-up; time-point 5: 24-month follow-up; E: measurement error. PEAK: drinking quantity on peak drinking occasion in past month; TYP: drinking quantity on typical drinking occasion in past month; RAPI: Rutger’s Alcohol Problem Index; HDE: frequency of heavy drinking episodes in past month. There were no significant zero-inflated effects for the RAPI model (all Ps > 0.05). *P < 0.05; **P < 0.01

Figure 3

Significant hypothesized model adding reciprocal paths [i.e. from drinking outcome variables to subsequent readiness to change (RTC)] to the previous hypothesized models. Because the model included both count and continuous outcomes, path coefficients in the white boxes represent standardized beta coefficients, whereas path coefficients in the gray boxes represent incident rate ratios (IRRs). Time-point 1: baseline; time-point 2: 6-month follow-up; time-point 3: 12-month follow-up; time-point 4: 18-month follow-up; time-point 5: 24-month follow-up; E: measurement error. PEAK: drinking quantity on peak drinking occasion in past month; TYP: drinking quantity on typical drinking occasion in past month; HDE: frequency of heavy drinking episodes in past month; FREQ: drinking frequency in past month. *P < 0.05; **P < 0.01. Significant zero-inflated coefficients were found for the autoregressive RAPI effects between time-points 2 and 3 (IRR = 0.75, P < 0.001), time-points 3 and 4 (IRR = 0.80, P = 0.01) and time-points 4 and 5 (IRR = 0.81, P = 0.002), such that greater experience of previous alcohol problems predicted lower likelihood of a self-report of zero problems at the later time-points