**Differences from the original study.** The protocol for the replication multisite study is improved in several ways and was agreed upon by the first author of the original study to be relevant for replication: - The current study collected multinational data, while the original study collected data in one lab in one city. - The sample size at each site was increased (minimum N=50 per site with at least N=500 for the present project; original project N=41). - The major methodological improvement was the measurement of the peripheral skin temperature via four sensors, compared to one sensor in the original study. Four ISP131001 sensors (Sarda et al., 2021) are fitted in total. One to the index and one to the pinky finger of the non-dominant hand, one on the wrist of the non-dominant hand, and one on the supraclavicular area on the non-dominant side. - Although the ISP131001 sensors are less accurate than the original study (0.1 degree accuracy as opposed to 0.0315 degrees Celsius), they measure more frequently (every second as opposed to every 15.6 seconds). We did not integrate this further into our power analysis, thus taking a very conservative approach to power for the new study. - We also measured several potential moderators in the study: attachment style (anxiety and avoidance via the ECR-R, Fraley et al., 2011), social integration and network size (via the social network index, Cohen et al., 1997), the Social Thermoregulation and Risk Avoidance Questionnaire (Vergara et al., 2019), and a variety of demographics (including, but not limited to height, weight, age). - To achieve better control of the experimental environment, the room temperature was continuously measured using Tempo DicsTM (Bluetooth Temperature, Humidity, and Dew Point Sensor Beacon and Data Logger). - For the day data was collected, the day before data was collected, and two days before data was collected, we scraped the minimum and maximum temperature and the minimum and maximum humidity through a data scraping script from Aerisweather (Paris & IJzerman, 2021a, b, & c), which relied on study location, start and end date, and start and end time for each session to retrieve its data. **Procedure** Data collection was organized either in one or, preferably, in two sessions. If data collection was organized in two sessions, survey data was collected in a group-testing session, in individual lab cubicles, or via an online survey, and experimental data were collected individually. For peripheral skin temperature measurement, upon entering the lab, participants were fitted with sensors on the aforementioned body parts. The sensors were then connected to the smartphone application on a Wiki Android phone. They were told, like in the original study, a cover story that the sensors record electrodermal activity during their performance on different cognitive tasks. Peripheral skin temperature procedure consisted of baseline skin temperature measurement (before the experimental manipulation) and skin temperature measurement during the experimental manipulation. At the start of the study, participants were offered questionnaires in a random order via the survey platform QualtricsTM, which was afterwards used to present the instructions, the reading task, and to administer the Cyberball game. The reading task was restricted to 10 minutes and its purpose was to provide cognitive activity for the participant during which his/her baseline skin temperatures were recorded. The measurement of the baseline skin temperature was conducted during the last five minutes of the reading task; the first five minutes of the reading task enabled the participant to adapt to the room ambience and to the task itself). Afterwards, participants were instructed to play the Cyberball game which was presented on a computer screen. They were assigned randomly to either the experimental or control condition. Participants were informed that there were other live players in the game that were represented by other avatars (players in the game). The control condition in the experiment was the social inclusion condition (where they had equal chances to be included in the game), while the experimental condition was the social exclusion condition (where participants were excluded from the ball-tossing game after two throws). In both conditions, the number of “players” (including the participant) in a Cyberball game randomly was fixed to three. Participants were thus randomly assigned to one of the two experimental conditions. Skin temperatures were recorded in both conditions. In this study, we determined the Cyberball game to last for 40 throws, taking approximately 3 minutes. After the game was over, the experimental measurement was completed. Participants were only debriefed via email after all participants were tested at a given location. **Materials, instruments, and equipment** Peripheral temperature measurement. For the measurement of peripheral temperature, we relied on the following materials: - Four skin temperature sensors ISP131001 (Sarda et al., 2021), - Room air temperature sensor - Tempo DicsTM (Bluetooth Temperature, Humidity and Dew Point Sensor Beacon and Data Logger), - Smartphone for collecting data from the sensors (via Bluetooth Low Energy). - The reading task consisted of a brief text about the history of psychology and three multiple-choice questions that were related to the text. The purpose of the reading task was to provide cognitive activity for the participant during which his/her baseline skin temperature was recorded. We only conducted a forward translation with a check by a second reader for this text. - A PC or Mac connected to the internet. - The Cyberball game to assess social ostracism. In Cyberball, the respondents play an online ball-tossing game with virtual players. The game enables manipulation between social inclusion and social exclusion. During this computer game, the ball is thrown from one player to the other. Each time the participant’s avatar captures the ball, they can throw it to any other player of their choice by clicking the mouse pointer to that player. The game can be programmed to determine the number of players, the number of throws and the way the throws are distributed among players. In the inclusion condition, the participants played as a “co-player” and had an equal chance of receiving the ball as the other players in the game, while in the exclusion condition they did not receive the ball from the “co-players” after the first two throws. - A minimum and maximum temperature and minimum and maximum humidity of the day, the day before, and the two days before for that location. - Start and end date and time for each session. **Social Thermoregulation protocol.** In the survey part of the study, the standard set of measures recommended for assessment in all social thermoregulation studies was administered. Social thermoregulation protocol includes the following instruments: - Social Network Index (SNI, Cohen et al., 1997) - measures participation in 12 types of social relationships, including those with a spouse/partner, parents, parents-in-law, children, close family members, close neighbors, friends, workmates, schoolmates, fellow volunteers, members of different religious affiliations and members of religious groups. We took two scores from the Social Network Index: one for the total network size and one for the diversity of one’s social network. For the former, it simply constituted a total number of people they communicate with at least one’s every two weeks. For the latter, the respondent assigns one point to each type of relationship indicating that they communicate to someone in that relationship role (e.g., spouse/partner or friends) at least once every 2 weeks. - Experiences in Close Relationships Scale-Revised (ECR-R, Fraley et al., 2011) assesses individual differences in attachment defined by two dimensions: attachment-anxiety and attachment-avoidance. It assesses attachment in different contexts: romantic, parental, and platonic relationships. ECR-R contains 36 items, and responses are given on a 7-point Likert-type scale, where 1 = strongly disagree and 7 = strongly agree. The scale allows for a calculation of avoidance and anxiety for romantic, parental, and platonic relationships. For all of these subscales ECR-R, we reported the McDonald’s Omega as a measure of reliability (McDonald, 1999). - Social Thermoregulation and Risk Avoidance Questionnaire (STRAQ-1; Vergara et al., 2019) assesses individual differences in social thermoregulation, solitary thermoregulation, risk avoidance, and high-temperature sensitivity. It aims to measure how attachment styles are reliant on biological drives. It contains 23 items, and responses are given on a 7-point Likert-type scale, where 1 = strongly disagree, and 7 = strongly agree. For the STRAQ-1, we will only ask participants questions from the social and solitary thermoregulation and high-temperature sensitivity subscales, due to their obvious relevance. We again reported the McDonald’s Omega for these three subscales. Demographics. Participants provided information on their height, weight, native language, self-declared gender, age, (romantic) relationship status, smoking status (and, if yes, how many cigarettes per day), and whether they use medication (and, if so, which kinds of medication). For women, a backward counting question to assess the last menstrual onset is included, measured, ideally, at the time of recruitment and during the study (Gangestad et al., 2016). - Study location (city, region, country) or participant latitude and longitude. - Language adaptations. Questionnaires and items from the Social Thermoregulation Protocol were already translated to many languages as part of the STRAEQ-2 crowdsourcing project (Dujols et al., 2022). If social thermoregulation protocol questionnaires were not already available in the target language, the participating labs would translate the survey items into the language of the participant pool following the translation process of the Psychological Science Accelerator, using a forward-backward-local adaptation procedure (Moshontz et al., 2018).