**Analytic Plan** General models To begin, we will conduct a 3-level multilevel model (MLM) with days nested within individuals nested within couples. As all couples are heterosexual (and therefore individuals nested in couples can be identifed by their gender), we have distinguishable dyads. We plan to allow for heterogeneity of variance across levels of the distinguishing variable (gender) by using a heterogeneous compound symmetry (CSH) residual structure. Additionally, we will use a standard dyadic reciprocal design, as we have data from both partners in each dyad on the same variables, and each partner is a member of only one dyad. Dyads do not have enough lower level units (partners), so the slopes are not allowed to vary randomly from dyad to dyad. However, as we have 21 lower level units (days) per partner, the slopes are allowed to vary randomly from partner to partner. We treat the scores from the two dyad members as repeated measures such that each dyad member has an error and the errors are correlated. All non-dummy coded variables will be grand mean centered. We will create 4 dummy coded variables for sexual advance status, with “no advance was made” as the reference category: 1) Accepted (the participant made asexual advance that was accepted by their partner), 2) Rejected (the participant made a sexual advance that was rejected by their partner), 3) Accepter (the participant accepted a sexual advance from their partner), and 4) Rejecter (the participant rejected a sexual advance from their partner). These dummy variables will then be used to predict daily sexual and relationship satisfaction, accounting for individual level satisfaction (average for that person across all days) and couple level satisfaction (average for that couple across all days). We will also run additional models, with our hypothesized moderators included at level 1 (day), including daily security, daily trust, and gender, predicting daily sexual and relationship satisfaction. Finally, we will run a full model with the dummy variables, hypothesized moderators, and control variables (gender, level 2; relationship length, level 3) included, with the control variables predicting aggregates of the dummy variables as well as satisfaction at levels 2 and 3. We will use a model building approach, where we look at the relationship between sexual advance status and satisfaction without moderators, the effects of each moderator independently, the effects of the moderators in combination, and finally all previous models with the control variables included. Lag analysis We will also conduct lag analyses to determine how long sexual rejection may be associated with sexual and relationship satisfaction. These analyses are based on the methods of Meltzer, Makhanova, Hicks, French, McNulty, and Bradbury (2017), but with slight variations due to differences in the predictor variables. That is, their predictors involved variables that represented similar experiences across both partners (whether or not sexual activity occurred that day, which should be the same for both partners) but the current study does not (the experiences of sexual rejection is likely different for the rejecter and the rejected, therefore it is not enough to simply say rejection occurred that day). Therefore, we will again incorporate the 4 dummy coded variables for sexual advances status but lag them to see if previous day’s status impacts today’s sexual satisfaction, while accounting for today’s sexual advance status. First, we will examine the extent to which sexual rejection on a given day predicts sexual satisfaction 1 day later by estimating the following lagged Level 1 equation in a three-level model: Yti = π0i + π1i (day of diary entry) + π2i (accepted 1 day prior) + π3i (rejected 1 day prior) + π4i (accepter 1 day prior) + π5i (rejecter 1 day prior) + π6i (accepted the same day) + π7i (rejected the same day) + π8i (accepter the same day) + π9i (rejecter the same day) + eti where Yti refers to individual i’s sexual satisfaction at time t, the day of the diary entry, and π0i is the intercept. Day of diary entry (a covariate) will be grand-mean-centered, and the sexual advance status variables will be uncentered. All Level 2 estimates will be allowed to vary across individuals, and the shared variance between partners’ data will be controlled in Level 3; all Level 3 estimates will be allowed to vary randomly across couples. In the second model we will examine the extent to which sex on a given day predicts sexual satisfaction 48 hr later by estimating the following lagged Level 1 equation in a three-level model with the same specifcations: Yti = π0i + π1i (day of diary entry) + π2i (accepted 2 days prior) + π3i (rejected 2 days prior) + π4i (accepter 2 days prior) + π5i (rejecter 2 days prior) + π6i (accepted 1 day prior) + π7i (rejected 1 day prior) + π8i (accepter 1 day prior) + π9i(rejecter 1 day prior) + π10i (accepted the same day) + π11i (rejected the same day) + π12i (accepter the same day) + π13i (rejecter the same day) + eti Some modifcations to this plan may be necessary depending on whether the models converge or not. Therefore, although we will proceed with this plan to the best of our ability, there may be some differences between this plan and the final models run. **Revised plan:** We based our original analytic plan on the methods of Meltzer, Makhanova, Hicks, French, McNulty, and Bradbury (2017), but with slight variations due to differences in the predictor variables. However, a further review of the literature has brought to our attention that the use of a three-level model with days nested within individuals nested within couples has a number of statistical problems and is not recommended by prominent statisticians in our field (e.g. Kenny, Kashy, & Cook, 2006). Therefore, we now amend our previous analytic plan based on the recommended method. That is, we will run a series of two-level linear mixed models predicting either sexual or relationship satisfaction. We will still include the 4 dummy coded variables for sexual advance status as predictors in the basic model, and the lagged predictors in the additional models. In the basic model we will estimate the residual matrix by nesting partner x time within couple to control for differences in couple-level satisfaction with a heterogeneous first-order autoregressive (ARH1) structure. In the lagged models, we will estimate the residual matrix by nesting partner within couple x time satisfaction with a heterogeneous first-order autoregressive (ARH1) structure. Note: We have already run the 3-level SEM models included in our previous preregistration. Although we will likely include the recommended analyses in our final manuscript, but the results of the 3-level model will be posted on the OSF in our Results component for comparison. Additional update: We will also run a series of models that conceptually replicate the findings of Meltzer and colleagues (2017), but with the updated statistical methods. That is, we will run a series of two-level linear mixed models predicting sexual satisfaction, including whether sexual activity occurred that day (Model 1, Model 2, and Model 3), whether sexual activity occurred the previous day (Model 2 and Model 3), whether sexual activity occurred 2 days prior (Model 3), and grand mean centered day of diary entry as predictors. If the effect of sexual activity 2 days prior is significant, we will run an additional model (Model 4) examining the effects of sexual activity 3 days prior, and so on until we no longer have evidence of an effect. In the basic model (Model 1) we will estimate the residual matrix by nesting partner x time within couple to control for differences in couple-level satisfaction with a heterogeneous first-order autoregressive (ARH1) structure. In the lagged models (Models 2, 3 and 4), we will estimate the residual matrix by nesting partner within couple x time satisfaction with a heterogeneous first-order autoregressive (ARH1) structure. We anticipate that the results of these analyses will replicate those of Meltzer and colleagues (2017). That is, we expect to find that sexual activity is associated with elevated sexual satisfaction for up to 48 hours after the sexual activity occurred.