This is an anonymized version of the pre-registration. It was created by the author(s) to use during peer-review.

A non-anonymized version (containing author names) should be made available by the authors when the work it supports is made public.

No, no data have been collected for this study yet.

The study examines habit formation in the context of fruit and vegetable consumption, whereas the habit to develop will be to fill half the plate with vegetables at dinnertime. The main question is whether context stability and reward value influence the automaticity of the target behavior.

1. Habit formation over time: between-level effects of behavioral consistency, context stability and reward value

a. There is a positive main effect of time (number of days since intervention start) on the automaticity of the target behavior, that is, automaticity increases over the course of the intervention.

b. There is a positive two-way interaction effect of time on the within level and consistency on the between level on the dependent variable automaticity, that is, the higher the consistency, the greater the effect of time on automaticity.

c. There is a positive two-way cross-level interaction effect of time on the within level and the person-specific mean reward value on the between level on the dependent variable automaticity, that is, the higher the mean reward value, the stronger the effect of time on automaticity.

d. There is a positive two-way cross-level interaction effect of time on the within level and the person-specific mean context stability level on the between level on the dependent variable automaticity, that is, the higher the context stability, the stronger the effect of time on automaticity.

Exploratory analysis:

Exploratorily, we will examine the three-way interaction effect between time on the within level, person-specific mean reward value and person-specific mean context stability on the between-level.

2. Habit formation on the daily level: within-level lagged effects

a. At the daily level, there is a positive lagged cross-level interaction effect of the person-centered reward value and habit execution at day x on the person-centered automaticity at day x+1, that is, the higher the reward value, the stronger the effect of habit execution on automaticity.

b. At the daily level, there is a positive lagged cross-level interaction effect of the person-centered context stability and habit execution at day x on the person-centered automaticity at day x+1, that is, the higher the context stability, the stronger the effect of habit execution on automaticity.

Exploratory analysis:

Exploratorily, we will examine the three-way lagged interaction effect of behavioral execution, person-centered context stability and person-centered reward value on day x on automaticity at day x + 1.

Behavioral automaticity will be measured with the four items from the Self-Report Behavioural Automaticity Index (SRBAI, Gardner, Abraham, Lally & de Bujin 2012). The prefix in front of the four items will be adapted to our study purposes (“Filling half of the plate with vegetables at dinnertime is something …”). Behavioral automaticity will be measured every day for six weeks at 8:30 p.m.

The mean of the four items will serve as the value for the dependent variable.

There will be only one condition: For the following six weeks, all participants receive the goal of building the habit of “filling half of the plate with vegetables at dinnertime” (Philipps et al., 2019). Additionally, all participants will learn about the benefits of vegetable consumption and the concept of habit. Subsequently, they will identify a cue for prompting the desired habit and develop implementation intentions (Gollwitzer et al., 1999) for the habit execution. We will use ecological momentary assessment (EMA) and examine intra-individual and inter-individual effects of the reward value of the behavior and the context stability of the behavior and influences in the habit formation process.

The two-way and three-way interaction effects as well as the cross-level lagged interaction effects will be examined via generalized linear mixed modeling (glmm) in R.

Therefore, we will specify two separate two-level random-intercepts-random-slope models, where measurement points (level 1) will be nested within persons (level 2).

In the first model, we will examine hypothesis 1 a-d, i.e. main and interaction effects of time (level 1), person-aggregated habit consistency, context stability, and reward value (level 2) on habit automaticity (level 1).

In the second model, we will examine hypothesis 2a-b, i.e. time-lagged main and interaction effects of person-centered habit execution, reward value, and context stability at day x (level 1) on habit automaticity at day x+1 (level 1).

Inclusion criteria: participants must be aged over 18.

Exclusion criteria: no access to a smartphone or computer in order to answer daily assessments.

Outliers: We will conduct the analyses with all participants who pass our inclusion and exclusion criteria (see above). Afterwards, we will conduct outlier analyses based on the median absolute deviation (|values|3 * median absolute deviation; Leys et al., 2013) and recode outliers utilizing Winsorization (Leys et al., 2019). The results with and without outlier handling will be compared to reach informed conclusions. We will further conduct two separate analyses: first with all participants who participated in the study and second with participants with a good adherence (> 50% or 28 questionnaires answered, respectively) to examine the influence of adherence on the results.

No need to justify decision, but be precise about

Unfortunately, the size of the expected effects is not known since there is no comparable study available, which complicates sample size calculation. However, we aim to recruit about 80-90 participants which will result in up to 3570 data points (85 participants on level 2 * 56 surveys on level 1), which should be enough to reliably estimate fixed effects (Moos & Hox, 2005; McNeish & Stapleton, 2016).

(e.g., secondary analyses, variables collected for exploratory purposes, unusual analyses planned?)

Participants will take pictures of their dinner plates during the 6-week examination period.

We collect them for double checking their statements on behavioral execution.