manuscript.tex

% define document type (i.e., template. Here: A4 APA manuscript with 12pt font)
\documentclass[man, 12pt, a4paper, mask]{apa7}

% change margins (e.g., for margin comments):
%\usepackage{geometry}
% \geometry{
% a4paper,
% marginparwidth=30mm,
% right=50mm,
%}

% add packages
\usepackage[american]{babel}
\usepackage[utf8]{inputenc}
\usepackage{csquotes}
\usepackage{hyperref}
\usepackage[style=apa, sortcites=true, sorting=nyt, backend=biber, natbib=true, uniquename=false, uniquelist=false, useprefix=true]{biblatex}
\usepackage{authblk}
\usepackage{graphicx}
\usepackage{setspace,caption}
\usepackage{subcaption}
\usepackage{enumitem}
\usepackage{lipsum}
\usepackage{soul}
\usepackage{xcolor}
\usepackage{fourier}
\usepackage{stackengine}
\usepackage{scalerel}
\usepackage{fontawesome5}
\usepackage[normalem]{ulem}
% \usepackage{longtable}
\usepackage{amsmath, nccmath}
\usepackage{mdframed}
\usepackage{ntheorem}
\usepackage{afterpage}
\usepackage{float}
\usepackage{array}
\usepackage{censor}
\usepackage{pdflscape}
\usepackage{lscape}
\usepackage{pdfpages}
\usepackage{enumitem}
\usepackage{caption}
\usepackage{adjustbox}
\usepackage{makecell}
\usepackage{tabu}
% Path Diagrams
\usepackage{tikz}
\usepackage{pgfplots}
\pgfplotsset{compat=1.17}
\tikzset{mynode/.style={draw,text width=1in,align=center} }
\usetikzlibrary{positioning}

% make warning with red triangle
\newcommand\Warning[1][2ex]{%
  \renewcommand\stacktype{L}%
  \scaleto{\stackon[1.3pt]{\color{red}$\triangle$}{\tiny\bfseries !}}{#1}}%

% make question with red triangle
\newcommand\Question[1][2ex]{%
  \renewcommand\stacktype{L}%
  \scaleto{\stackon[1.3pt]{\color{red}$\triangle$}{\tiny\bfseries ?}}{#1}}%
  
% add definition sections
\theoremstyle{break}
\newtheorem{definition}{Definition}
\newcommand{\defref}[2][]{\hyperref[#2]{Definition \ref*{#2}#1}}

% add hypothesis sections
\theoremstyle{plain}
\theoremseparator{:}
\newtheorem{hyp}{Hypothesis}

\newtheorem{subhyp}{Hypothesis}
   \renewcommand\thesubhyp{\thehyp\alph{subhyp}}

% Analysis Plan sections (probably better as Definitions or Remarks instead of Theorems)
\newtheorem{ap}{Hypothesis}

\newtheorem{subap}{Hypothesis}
   \renewcommand\thesubap{\theap\alph{subap}}

% add quote section
\usepackage{csquotes}

% framed box section
\usepackage{framed}
\emergencystretch=1em

% formatting links in the PDF file
\hypersetup{
pdfpagemode={UseOutlines},
bookmarksopen=true,
bookmarksopenlevel=0,
hypertexnames=false,
colorlinks   = true, %Colours links instead of ugly boxes
urlcolor     = blue, %Colour for external hyperlinks
linkcolor    = black, %Colour of internal links
citecolor   = black, %Colour of citations
pdfstartview={FitV},
unicode,
breaklinks=true,
}

% ref labels
\newcommand{\fgrref}[2][]{\hyperref[#2]{Figure \ref*{#2}#1}}
\newcommand{\tblref}[2][]{\hyperref[#2]{Table \ref*{#2}#1}}
\newcommand{\appref}[2][]{\hyperref[#2]{Appendix \ref*{#2}#1}}

% custom open science badge height
\newlength{\badgeheight}
\setlength{\badgeheight}{1em}

% language settings
\DeclareLanguageMapping{american}{american-apa}

% add reference library file
\addbibresource{references.bib}

% Title and header
\title{Need Fulfillment During Intergroup Contact: Three Experience Sampling Studies}
\shorttitle{Need Fulfillment in Intergroup Contact}

% Authors
\author[*,1]{Jannis Kreienkamp}
\author[1]{Maximilian Agostini}
\author[1]{Laura F. Bringmann}
\author[1]{Peter de Jonge}
\author[1]{Kai Epstude}
\affiliation{\hfill}

\affil[1]{University of Groningen, Department of Psychology}


\authornote{
   \addORCIDlink{* Jannis Kreienkamp}{0000-0002-1831-5604}\\
   \addORCIDlink{Maximilian Agostini}{0000-0001-6435-7621}\\
   \addORCIDlink{Laura F. Bringmann}{0000-0002-8091-9935}\\
   \addORCIDlink{Peter de Jonge}{0000-0002-0866-6929}\\
   \addORCIDlink{Kai Epstude}{0000-0001-9817-3847}

We have no known conflict of interest to declare. The authors received no specific funding for this work. Materials and software are available at \url{https://janniscodes.github.io/intergroup-contact-needs/}  \citep{Kreienkamp2022}. Protocols, materials, data, and code are available at \url{https://osf.io/pr9zs/?view_only=208a53a1f0ff48dda1c17357328fa578} \citep{Kreienkamp2022a}. The preregistration of Study 3 can be accessed as part of our Open Science Framework repository \citep{Kreienkamp2021f}

Correspondence concerning this article should be addressed to Jannis Kreienkamp, Department of Psychology, University of Groningen, Grote Kruisstraat 2/1, 9712 TS Groningen (The Netherlands).  E-mail: j.kreienkamp@rug.nl}

\leftheader{Kreienkamp}

% Abstract
\abstract{
One challenge of modern intergroup contact research has been the question of when and why an interaction is perceived as positive and improves intergroup relations. We propose to consider the perceived fulfillment of the situationally most relevant need. We conducted three intensive longitudinal studies with recent migrants, to capture their interactions with the majority outgroup ($N_{measurements}$ = 10,297; $N_{participants}$ = 207). The core need fulfillment mechanism is consistently a strong predictor of perceived interaction quality and positive outgroup attitudes following intergroup interactions. The model is specific to outgroup contact, robust to various need types, and works at least as well as Allport's contact conditions. As one of the first studies to test intergroup contact theory using intensive longitudinal data, we offer insight into the mechanisms of positive intergroup contact during real-life interactions and find situational motivations to be a key building block for understanding and addressing positive intergroup interactions.

\noindent\textbf{Public significance statement}: In this paper, we provide evidence that the fulfillment of situational needs during real-life intergroup contacts meaningfully predicts perceived interaction quality and positive outgroup attitudes. Methodologically, this offers testament to the emerging practice of capturing real-life interactions using intensive longitudinal data. Theoretically, our results give weight to motivational fulfillment as a flexible and effective mechanism for understanding positive intergroup contact.
}

\keywords{
    Intergroup Contact, Need Fulfillment, Outgroup Attitudes, Interaction Quality, Intensive Longitudinal Data\\
    \textit{Open Science Practices:}
    \noindent \href{https://osf.io/r5e8c?view_only=bfaed2196d49420385ce5fa6506a7765}{\includegraphics[height=\badgeheight]{assets/open-badges-small/registrationplus-color.png}} Preregistration+, 
    \href{https://osf.io/pr9zs/?view_only=1ea47bb646694632a764dead807ef970}{\includegraphics[height=\badgeheight]{assets/open-badges-small/material-color.png}} Open Materials, 
    \href{https://osf.io/pr9zs/?view_only=1ea47bb646694632a764dead807ef970}{\includegraphics[height=\badgeheight]{assets/open-badges-small/data-color.png}} Open Data,\break
    \href{https://osf.io/pr9zs/?view_only=1ea47bb646694632a764dead807ef970}{\includegraphics[height=\badgeheight]{assets/open-badges-small/code-color.png}} Open Code, 
    \href{https://osf.io/pr9zs/?view_only=1ea47bb646694632a764dead807ef970}{\includegraphics[height=\badgeheight]{assets/open-badges-small/supplements-color.png}} Open Supplements
    }


% set indentation size
\setlength\parindent{1.27cm}

% Start of the main document:
\begin{document}

% add title information (incl. title page and abstract)
\maketitle

% **CHEAT SHEET / LEGEND**
%
% Comments:
% '%' starts a comment in LaTeX (not printed)
% '\todo[inline]{} makes orange boxes in PDF
% '\marginpar{}' notes in margins
% '\footnote{}' footnote
% '\Warning' important note indicator in PDF (triangle with exclamation mark)
% '\Question' question note indicator in PDF (triangle with question mark)
%
% Citation (with Natbib citation style):
% '\citep[e.g.][p. 15]{CitationKey}' citation in parentheses "(e.g., Berry, 2003, p. 15)"
% '\citet{CitationKey}' citation in text "Berry (2003)"
% '\citealt' and '\citealp' alternate citation without parentheses
% '\citeauthor' and '\citeyear' only year or author
% 
% Headings:
% '\part{}' and '\chapter{}' only relevant for multi-part or multi-chapter documents
% '\section{}' heading level 1
% '\subsection{}' heading level 2
% '\subsubsection{}' heading level 3
% '\paragraph{}' heading level 4
% '\subparagraph{}' heading level 5
%
% formatting:
% '\textbf{}' text bold font
% '\textit{}' text italic font
% '\underline{}' text underline
% '\sout{}' text strike out
% '\textsc{}' text small caps
% '\vspace{1em}' add vertical space
% '\hspace{1em}' add horizontal space
% '\\' new line (i.e., line break)
% '\pagebreak' start new page (i.e., page break)
% '\noindent' do not indent current line (e.g., current paragraph)
% 'begin{center}...end{center}' center text or object
%
% Math mode:
% '$\alpha = .8$' mathematical equation inline
% '$$\hat{y} = b_0 + b_1x$$' mathematical equation in its own line
% '\begin{equation}...\end{equation}' multi-line equation
% '\approx' approximate symbol
% '\neq' not equal
% '\bar' mean bar over letter
% '\pm' plus minus sign 
% '^{}' superscript
% '_{}' subscript
% '\fraq{numerator}{denominator}' fraction
% '\sqrt[n]{}' square root
% '\sum_{k=1}^n' sum for 1 through n
%
% Insert things from elsewhere:
% '\input{filename}' inputs the raw (tex) file as a command (e.g., tables and R-Markdown imports)
% '\include{filename}' includes section on new page (incl. possible auxiliary info)
% '\includegraphics[settings]{filename}' add a figure or graph
% '\caption{}' adds a caption to a table or figure
% '\label{}' labels sections, tables, figures, etc. so that they can be referred to.
% '\ref{}' refer to a labelled sections, tables, figures, etc.
% '\begin{enumerate}...\end{enumerate}' numbered list
% '\begin{itemize}...\end{itemize}' bullet-ed list
% '\item' item in list section 
%
% Symbols:
% '\&' and sign
% '\%' percent sign
% '\_' three dotes
% '\#' hash symbol
% ------------------------------------------------------------------

% Migrant Example Pathways
% Relevance (Version 1 of 2): Migrant Example Version:
One of the main intergroup societal issues to date, are the struggles of many migrants across the world, hoping to build a new life that includes a positive relationship with the majority group. The intergroup contact hypothesis postulates that prejudice can be reduced and favorable attitudes be increased if members of two groups have frequent and positive contact \citep[e.g.,][]{Allport1954b, Hewstone1996, Pettigrew1998}. Over the past 70 years, a plethora of studies and interventions have shown the general effectiveness of positive intergroup contact \citep[e.g.,][]{Pettigrew2006}. However, even though a central assumption of intergroup contact theory is that the contact should be positive, relatively little research has thus far explained when and why people perceive their everyday intergroup interactions as positive. 

% Full Migrant Pathway
% Relevance (Version 2 of 2): Full Migrant Focus Version:
%The adaptation of migrants in new cultural contexts has become an important issue for many societies around the world. A major aspect of such migrant adaptation arguably unfolds during the daily interactions migrants have with the cultural majority members \citep{Maxwell2017, Sam2010}. One of the main social psychological theories aimed at understanding contact between social groups is ’intergroup contact hypothesis’. In its most essential interpretation, the intergroup contact hypothesis postulates that prejudice can be reduced and favorable attitudes increased if members of two groups have frequent and positive contact \citep[e.g.,][]{Allport1954b, Hewstone1996, Pettigrew1998}. Over the past 70 years, a plethora of studies and interventions have shown the general effectiveness of positive intergroup contact \citep[e.g.,][]{Pettigrew2006}. However, even though a central assumption of intergroup contact theory has been that the contact should be positive, relatively little research has thus far explained when and why people perceive their everyday inter-group interactions as positive.

% Old Relevance: Conflict is a problem, positive contact as solution, but little research on when daily contact is perceived positive
%Conflict between social groups and their individual members remains a prevalent feature of the modern human condition. Experiences of prejudices, discrimination, and animosities with other groups continue to plague the everyday lives of many people around the world. One of the main ameliorations proposed by social psychologists has been the intergroup contact hypothesis. In its most essential interpretation, the intergroup contact hypothesis postulates that frequent and positive contact with an out-group reduces prejudice and increases favorable attitudes towards the other group \citep[e.g.,][]{Allport1954b, Hewstone1996, Pettigrew1998}. And even though over the past 70 years, a plethora of studies and interventions have shown the general effectiveness of positive intergroup contact \citep[e.g.,][]{Pettigrew2006}, relatively little research has thus far explained when and why people's everyday inter-group interactions are perceived as positive.

% Problem v.02: theoretical interaction quality central to understanding outcomes, practical many impactful negative interactions in everyday life
% Problem Illustration Paragraph:       
Importantly, as we still fail to understand when and why an interaction is perceived as positive, substantial theoretical and practical challenges remain. There is now consistent evidence that negative intergroup contacts lead to worse attitudes, prejudice, and reduced future interaction motivation \citep[e.g.,][]{Barlow2012, Prati2021, Graf2014}. In light of these findings, understanding interaction quality thus sits at the heart of understanding when an intergroup contact is successful \citep[e.g.,][]{Allport1954b, Brown2007, Tropp2016}. But also in applied settings, policymakers and practitioners are thus far often under-prepared to deal with the occurrences of negative interactions, especially in everyday life contexts. Understanding the psychological mechanisms of when and why interactions are perceived as positive is, thus, an important issue for understanding whether an interaction leads to better intergroup perceptions, especially during everyday interactions.

% Aim / Solution v.02: look at need fulfillment as a mechanism in daily interactions of migrants
% Proposal Paragraph:
We propose that one key to understanding how an interaction is perceived is to examine the level of need fulfillment it provided to an individual. As an example, if someone seeks acceptance by their interaction partner, and this need is fulfilled during the interaction, the person should rate the interaction and the group of the interaction partner more favorably. To test this idea, we collected three sets of real-life data from recent immigrants, assessing their daily interactions with majority group members, tracking situational needs, interaction quality, and outgroup attitudes. 

\section{Need Mechanism in Intergroup Contact}
% Motivational mechanism:
% interaction quality is important
% past research has either focused on conditions or cognitive-affective processes
% neither are necessary nor explain why an interaction is positive
% Several meta-analytic reviews have found that positive intergroup contact reduces prejudice and increases positive attitudes in experimental and cross-sectional studies \citep[][]{Tropp2005, Pettigrew2006}, as well as in intergroup contact interventions outside the lab \citep[][]{Beelmann2014, Lemmer2015}. 

Looking at the past literature, we can essentially separate intergroup contact theory research into a two-step problem. Firstly, we need to understand when and why contact becomes a positive contact (contact $\rightarrow$ positive contact) and, secondly, we need to understand when and why positive contacts drive better intergroup relations \citep[positive contact $\rightarrow$ better relations; e.g., see ][]{Allport1954b, Hewstone1996, Pettigrew1998}.

In recent years, research has focused on the second step of understanding the psychological processes that explain how positive contacts improve intergroup relations \citep[e.g. see,][]{Paolini2021}. Among others, researchers have explored different forms of social categorizations \citep[][]{Pettigrew1998}, the salience of social categories \citep[][]{Brown2005}, intimacy \citep[e.g.,][]{Marinucci2021} and attachment \citep[e.g.,][]{Tropp2021}, threat and intergroup anxiety \citep[e.g.,][]{Stephan2008}, as well as knowledge about the other group \citep[][]{Pettigrew2008c}. Most recently, researchers have even looked at how empowerment need fulfillment during positive intergroup contact can explain some of the beneficial intergroup effects \citep[][]{Hassler2021}. There is thus, substantial evidence on the psychological mechanisms that explain the effects of positive contact.

Research on the first step of what makes an interaction positive to begin with tends to be much older, and often more static and contextual. The most widely used approach has been the idea that equal status, common goals, collaboration, and structural support during the interaction form Allport's optimal conditions for positive contacts \citep[][]{Allport1954b, Pettigrew1969}. Following Allport's original conditions, several additional conditions of optimal contact were proposed, including, stereotype disconfirmation \citep[][]{cook1978} or common language and voluntary interaction (\citealp{wagner1986}; for a critical discussion see \citealp{Pettigrew1986}). However, despite their prominence in guiding research on this topic, meeting the contact conditions does not seem to be necessary to finding positive effects of intergroup contact \citep[][]{Pettigrew2006} and more fundamentally, the conditions often do not capture any underlying psychological mechanisms of why an interaction is perceived as positive \citep[e.g.,][]{Pettigrew1998}.

In this article, we focus on the role of motivation and need fulfillment to understand when and why exactly an interaction is perceived as positive. We propose need fulfillment in particular because needs are a fundamental aspect of the human experience that governs a significant number of emotional, cognitive, and behavioral facets \citep[][]{Kreienkamp2022d, kruglanski2002}. Importantly, need fulfillment has particularly been highlighted in explaining the success of (close) relationships, psychosocial functioning, as well as reducing conflict between groups --- all of which are essential to positive intergroup interactions.

On an individual psychological level, there is a long tradition of using need fulfillment to explain what drives human adaptation and social relations. From the early works of \citet[][]{maslow1943} and \citet[][]{lewin1926e} to more recent works by \citet[][]{Ryan2017} or \citet[][]{Steverink2006}, the fulfillment of needs have been considered a driver of psychosocial functioning. Most relevant to our proposal here, within experience sampling studies need fulfillment has been found to explain variations in well-being during daily interactions \citep[][]{Downie2008} and has been found to be important in understanding the success of close relationships \citep[e.g., see][]{knee2023}. In short, an extensive body of scholarly work underscores the significance of need satisfaction in fostering favorable social relationships and social functioning.

Beyond the individual relations literature, need fulfillment has recently also seen application as a psychological mechanism in the intergroup relations literature. Social identity theory has focused on the role of self-esteem needs in understanding how people navigate intergroup contexts \citep[e.g.,][]{abrams1988}. In the study of conflict and reconciliation, addressing differential needs of victims and perpetrators (i.e., the need for power and the need for morality respectively) increased willingness to reconcile \citep[][]{Shnabel2008}. And similarly, addressing a relevant need for identity continuity among refugees in Turkey bolstered resilience in the face of discrimination experiences \citep[][]{Celebi2017}. In short, an increasing amount of literature is emphasizing the significance of need satisfaction in understanding intergroup dynamics.

It is thus not surprising that Dovidio and colleagues propose that: "To achieve truly constructive intergroup relations, it is important that intergroup exchanges meet the psychological needs of both majority- and minority-group members." \citep[][p. 6]{Dovidio2017}. A call that has thus far remained unanswered when it comes to the basic tenet that need fulfillment underpins positive and constructive interactions.

% Difficult to test general mechanism: Either consider too many needs (situationally relevant but not feasible) or too few (feasible but not transferable). Solution: ask for main need + rating of main need  
% MAYBE MOVE THIS TO 'The Present Research Section'
One reason why motivational considerations might have remained absent from the intergroup contact literature is that there is an overwhelming number of individual motives or goals that might be relevant to a person during an intergroup interaction. Researchers considering the motivational content would, thus, either test few hyper-specific needs that might not be transferable to other intergroup contexts or they may need to assess a broad and diverse range of motives. However, while the specific need content differed within the different lines of research, what unites most motivational researchers is a focus on fulfilling the situationally relevant needs of people. This motivational experience of need fulfillment, thus, brings many of the diverse need content theories together and offers a common psychological mechanism for understanding positive intergroup contact.

Here it is important to briefly define what exactly we mean by need fulfillment and how it differs from need content theories. With motivation and need fulfillment we specifically mean the psychological experience of addressing an active and relevant need during the interaction. For our purposes, we define a need as:

\begin{definition}[Need]\label{def:need}
A tension or deficiency in the organism that elicits a (non-specific) motivational force organizing affect, cognition, and behavior to reduce this unsatisfactory situation, which is to some extent necessary for the individual’s overall well-being.
\begin{flushright}
--- based on \citet[][]{dweck2017, Hull1943, kruglanski2002, Lewin1938, McClelland1987, Ryan2017, Steverink2006}
\end{flushright}
\end{definition}

The psychological experience of need fulfillment is, thus, distinct from the content of the need (i.e., the motive or goal). The content could, for example, include physical motives (such as safety or hygiene) but also psychosocial motives \citep[such as acceptance or competence; e.g., see][]{pittman2007}. The experience of needing is a more general process that arises when any important motive is thwarted or situationally active and relevant \citep[][]{leander2020, lewin1926e, gollwitzer1985e}. It is this perceived needing and the perceived fulfillment of needs that we focus on in this article. This is not to say that considering specific motives is irrelevant to contact situations but instead, we propose that the psychological experience of perceived need fulfillment is a core mechanism in understanding interaction quality perceptions, well-being, and outgroup attitudes.

To test such a proposal, we can rely on adaptive and responsive survey designs that allow a tailored approach based on the participants' inputs \citep[e.g.,][]{Tourangeau2017}. In particular, we propose to ask the participants to report their main goal during the interaction in a short open-ended question (i.e., name the situationally relevant need content), and with reference to their own response, the participants can then indicate how much this need was fulfilled during the interaction (i.e., need fulfillment mechanism). Such an adaptive approach allows us to take the initial step of testing whether situational need fulfillment indeed generally predicts perceived interaction quality, well-being, and positive outgroup attitudes independent of need content. 

\section{Intergroup Contact in Daily Life}
While we have argued that a need fulfillment mechanism is relevant to intergroup contact generally, its flexible and broad applicability might be ideally suited to address the pressing issue of understanding natural intergroup contacts outside the lab. Investigations of such `real-life interactions' often suffer from the difficulty that past intergroup contact research has either focused on the mechanisms of individual interactions in artificial lab studies (sometimes referred to as the intergroup interaction literature) or has focused on longer-term recall self-reports of natural interactions \citep[commonly referred to as the intergroup contact literature; also see][]{Pettigrew2006}. \citet[][]{MacInnis2015} have even pointed out that these two approaches tend to find conflicting effects --- where individual interactions (in the lab) have more negative effects and recall of real-world contact patterns have more positive effects for intergroup relations. We, thus, miss data following people in their diverse daily interactions and investigating the psychological mechanisms of contacts, especially as they compound over time. Even with extended intervention studies, the most fine-grained data available is usually limited to pre-post-control designs. 

The lack of longitudinal real-world data, however, stands in stark contrast to many of the theoretical advances that have focused on the dynamic nature of intergroup relations \citep[e.g.,][]{Pettigrew1998}, as well as the original contact hypothesis, which was focused on the daily interactions of people \citep[][]{Allport1954b}. As a result, prominent researchers in the field have long called for longitudinal \citep[][]{Pettigrew1998, Pettigrew2008, Pettigrew2011} and real-life experience-sampling data outside the lab \citep[ESM][]{MacInnis2015, McKeown2017}. Such data would be able to capture real-life interactions that include interaction-specific mechanism information close to the actual experience\footnote{Additionally, such experience-sampling data can be collected close to the intergroup interactions and would, thus, largely mitigate recall biases. Moreover, because data is nested within participants, experience-sampling data often allows capturing large amounts of high-quality data with relatively few participants \citep[][]{shiffman2008}.}.

% Used to be difficult but technological and methodological developments (e.g., FormR + HLM) → collect large body of intensive longitudinal data
In the past, such data collections were often unfeasible because they were either physically impractical or too expensive. However, recent technological developments allow us to easily collect experience sampling data on mobile devices \citep[e.g.,][]{Keil2020} or using web-based applications \citep[e.g.,][]{Arslan2020}. At the same time, analytical methods for such more complex data have become more readily available, making the analyses more approachable \citep[e.g., see][]{ODonnell2021}. Given these technological and methodological developments, we were able to collect three independent studies of extensive real-life data following the daily intergroup interactions of recently arrived migrants with the majority-group members.  

\section{The Present Research}
Using three independent sets of intensive longitudinal data (Studies 1–3), the aim of this paper is essentially threefold. We (1) seek to test the basic ideas of the contact theory within real-world experience sampling data. We (2) aim to test the core need fulfillment mechanism within the real-world data. And we (3) seek to ensure the stability, robustness, and embeddedness of our results.

Firstly, for the general contact hypothesis test, our study is among the first to test the fundamental tenets of intergroup contact and Allport's conditions in real-life intensive longitudinal data. Translating the contact hypothesis into intensive longitudinal data is not a trivial task, as past research traditions have used two fundamentally different approaches. While lab studies have tended to focus on the effect of a single positive interaction, cross-section studies have primarily investigated the frequency of positive interactions more generally. Intensive longitudinal data allows us to investigate both. We can test whether having a specific type of interaction vs. not having an interaction improves intergroup relations, but we can also use the participant's 30-day contact reports to test whether participants with more positive interactions tend to benefit more from intergroup contact. Testing both approaches to the contact hypothesis allows us to go beyond a replication of the basic theory but could disentangle individual- from aggregated contact effects and would allow for a direct comparison with both bodies of literature. 

We test the basic contact hypothesis within and across the three studies. In particular, we assess the effect of individual interactions within each study using a multilevel model, but to avoid power limitations, we test the collective effect of contact frequency and -quality after the individual studies, across all participants.
\begin{enumerate}[leftmargin=1.5\parindent]
    \item[H1:] Based on the most general understanding of the contact hypothesis, an increase in frequency and quality of contact should jointly account for more favorable outgroup attitudes within and across intensive longitudinal data.
\end{enumerate}

The test of Allport's conditions is notably restricted to measurements that report on outgroup interactions because Allport's conditions and interaction quality ratings cannot meaningfully be measured or imputed if participants did not have an interaction. Focusing on the interactions in detail, we use a multilevel regression model to test whether interactions that are higher in the fulfillment of Allport's conditions predict more favorable outgroup attitudes. We would also expect that such interactions are perceived as higher in interaction quality.
\begin{enumerate}[leftmargin=1.5\parindent]
    \item[H2:] Based on the literature about Allport’s optimal contact conditions, intergroup interactions that are higher in equal status, common goals, collaboration, and structural support should predict more favorable outgroup attitudes due to more positive interaction quality perceptions within the intensive longitudinal data.
\end{enumerate}

Once the general contact hypothesis is established within the ESM data, our second main aim is to test our main theoretical proposal that the fulfillment of situational needs is meaningfully related to more positive outgroup attitudes following intergroup interactions. As our main proposal is concerned with the mechanisms of successful intergroup contact, we again focus on outgroup interaction reports. Within a multilevel model, we expect interactions that are higher in situational core need fulfillment to be perceived as more positive, and as a result that these interactions also predict more positive outgroup attitudes. We also expect the needs mechanism to work at least as well as Allport's conditions. We particularly expect part of Allport's contact conditions to be a static set of situational needs so that the situational core need fulfillment should explain some of the same variance in outgroup attitudes.
\begin{enumerate}[leftmargin=1.5\parindent]
    \item[H3:] Based on our proposal, intergroup interactions with higher situational core need fulfillment should predict more favorable outgroup attitudes due to more positive interaction quality perceptions within the intensive longitudinal data. We also expect situational need fulfillments to work at least as well as Allport's optimal contact conditions in predicting outgroup attitudes.
\end{enumerate}

Our third main aim is to ensure that our results are robust, stable, and ecologically valid. To test the robustness of our need-fulfillment mechanism we test whether the need mechanism is indeed specific to outgroup interactions and whether the process could be explained by a smaller set of fundamental psychological needs instead. We additionally, assess the need fulfillment mechanism in predicting individual well-being benefits and check whether different types of needs or interactions change the main results. We present the full robustness analyses in \appref{app:AppendixRobustness}. To test the stability and reliability of our results, we utilize forest plots and meta-analytic estimates for our main analyses. To assess the embeddedness of our situational core needs, we use an exploratory topic model for the participants' free-text entries and compare the extracted content topics with themes commonly found within the motivational literature.

Before turning to individual studies, we would like to address a number of conceptual, practical, and methodological considerations. One key decision for our studies has been to focus on the minority experience during the contact. While the same mechanisms should hold for the experience of members in high-power groups, there is substantially more research available that focuses on the experience of the majority group, and minority perspectives are historically often understudied \citep[e.g.,][]{Dovidio2017}. At the same time, however, minority groups are often underprivileged and research is direly needed to understand the more prevalent experiences of stress and health issues among minorities \citep[e.g.,][]{alvidrez2019}. 

A second non-trivial aspect of translating the intergroup contact hypothesis into intensive real-world data was the choice of the outcome variable. For our main analyses, we chose outgroup attitudes --- the positive or negative evaluation of the other group. We chose outgroup attitudes mainly because they are the most common outcome considered within the intergroup contact literature \citep[e.g.,][]{Pettigrew2006, Paolini2021}. As the methodology is relatively new to the field, we sought to first replicate (and then extend) the most reliable effects of the contact hypothesis within the ESM data. Outgroup attitudes are, however, not without controversy, especially for minority group members. Positive outgroup attitudes can increase harmony and reduce the willingness to support social change among the disadvantaged in some cases --- even in the face of injustice \citep[e.g.,][]{dixon2012, saguy2009}. While a recent review found the effect to be less conclusive for longitudinally collected data and less consistent for positive interactions (rather than interactions generally), the backfire effect remains an important possibility for the present data \citep[see][]{reimer2023}. In order to ensure at least a direct benefit to the minority group members, we also assess the effect of the need fulfillment mechanism on well-being as the dependent variable as part of our robustness analyses below.

In terms of methodological considerations, it is important to note we tested most of our hypotheses using multilevel regression models, where measurement occasions (level 1) were nested within participants (level 2). This approach is tolerant to missing data and uneven case numbers within participants. Furthermore, we use a hierarchical modeling approach and report the final model in-text \citep[][; for the full modeling process see Supplementary Material A]{snijders2012}. Secondly, statistical power estimations for intensive longitudinal- and multilevel models are notoriously difficult due to the complex covariance structures. However, our participant- and measurement numbers are among the largest sample sizes found within the intensive longitudinal literature \citep[e.g.,][]{AanhetRot2012}. Additionally, power simulations after the first study showed that our data was sufficiently powered for even small effect sizes (see Supplemental Material B). In particular, we found that even our smallest effects of interest would be detectable with 22 participants and 24 measurements per person (assuming the effect sizes of Study 1 and focusing on a power of .8 with a .05 alpha level; see Supplemental Material B for the full analyses). We only increased the participant sample sizes in Studies 2 and 3 to allow for between-participant effects across studies and more complicated trajectory analyses, which are not necessary for the hypotheses tested here.

Finally, for our most comprehensive study (Study 3) we preregistered both the hypotheses as well as the analysis plan \citep[available at][]{KreienkampMasked2021f}. All studies received ethical approval from University Masked for Peer Review and none of the data has been published elsewhere. The detailed hypotheses and analysis plan are available in \appref{app:AppendixHypotheses}. The full surveys, code, and materials are available in our open science repository \citep[including a complete codebook;][]{KreienkampMasked2022a}. Additionally, the fully annotated analyses are available in Supplementary Materials A, B, and C.

% Methods and Results from RMarkdown render
\input{Methods-and-Results}

\section{Discussion}
% aims re-iterated
The main aim of this project was to test the basic tenets of the intergroup contact hypothesis and Allport's optimal conditions in real-life intensive longitudinal data as well as to test whether the fulfillment of situational needs meaningfully predicts positive interaction perceptions and outgroup attitudes. 

% Contact hypothesis in ESM data [inconsistent in aggregate but consistent in ML analysis]
When considering the results of the three studies jointly, we found mixed results for the basic intergroup contact hypothesis. To replicate the two common approaches to the contact hypothesis, we looked at both the within-person effects of individual outgroup interactions (mimicking the analysis of lab studies) as well as the joint effect of interaction frequency and average interaction quality between participants (mimicking the cross-section literature). For the effect of individual interactions, we find that having an outgroup interaction (vs. not having an interaction) was associated with more favorable outgroup attitudes. Similarly, we find that within outgroup interactions, the interaction quality was meaningfully associated with more favorable outgroup attitudes. Yet, considering interaction frequency and average interaction quality jointly was only possible on the aggregated between-participant level. Surprisingly, here we found independent effects of interaction frequency and average interaction quality ratings but no interaction effect. The absence of this aggregate effect could underline the fact that cross-sectional retrospective data might be misleading because (1) it presents a mixture of within and between-subjects effects \citep[][]{Hamaker2020}, or (2) suffers from recall biases (e.g., where times with no outgroup interactions are undervalued by participants during the retrospective evaluations). 

Interestingly, this effect is also inconsistent with the observations and theorizing of \citet[][]{MacInnis2015}, who observed that individual interactions showed negative effects on intergroup relations and the aggregate of past intergroup contacts showed positive effects on intergroup relations. There are, however, two important differences in our data compared to the past literature assessed by \citeauthor{MacInnis2015}. Firstly, what \citeauthor{MacInnis2015} called the intergroup interaction literature, has particularly focused on artificial lab studies where study participants meet a stranger from the outgroup. In our real-world data, such synthetic and controlled interactions are arguably less relevant. Secondly, what \citeauthor{MacInnis2015} called the intergroup contact literature, has looked at long-term recall self-reports --- where participants are asked to recall the quantity and average quality of intergroup contacts over the past month or year \citep{MacInnis2015}. The mental aggregation of such long-term recall surveys is substantially different from the aggregation we did based on the close-to-event reports \citep[][]{shiffman2008}. To truly compare our results to \citet[][]{MacInnis2015} theorizing, future studies should, thus, also collect a long-term recall report that mirrors the questions asked during intergroup contact studies.

It should also be noted that the inconsistencies with past research might in part be a data artifact (e.g., because most people reported substantially more measurements during which they did not have an outgroup interaction). There is also a possibility that statistical power was a concern given our sample was relatively small with 207 participants and the effect of interest is an amplification interaction effect. However, if we assume an effect size of \textit{r} = -.21 for the effect of positive intergroup contacts \citep[see][]{Pettigrew2006} our sample size should be at the threshold of 80\% power with a .05 alpha level (sensitivity analysis in G*Power: $f^2 = 0.04$) --- even more so if we consider the higher quality data we get from aggregating many real-world reports with less recall bias. 

% Allport's conditions in ESM data [consistent predictor and partially through quality]
Next, using the data from our third study, we find that Allport's conditions are related to higher interaction quality perceptions and more positive outgroup attitudes. When we consider interaction quality and Allport's conditions jointly, we find that interaction quality ratings assumed a larger part of the shared variance in outgroup attitudes. We thus find first evidence that Allport's conditions of optimal contact are also relevant to the daily interactions recent migrants have in their interactions with majority group members. 

% Need fulfillment in intergroup contact [consistent predictor, through quality, slightly better than Allport and SDT]
Finally, when looking at the results regarding the importance of situational key need fulfillment, we find that in all three intensive longitudinal data sets, the fulfillment of core needs during intergroup contacts predicts higher interaction quality perceptions, more positive outgroup attitudes, as well as higher well-being. We also find that in all three studies, need fulfillment and perceived interaction quality likely shared a large part of the variance they explained in outgroup attitudes (when considering partial regression coefficients in a joint model). We would like to reiterate here that we specifically did not seek to test a mediation-style model. Even though the shared explained variance was predicted in our pre-registration based on a theoretical model, and the pattern was stable across the three studies, it is important to note that the data is none-causal, and the effect might alternatively be driven by an unobserved third variable or by multicollinearity.

We, additionally, find that need fulfillment is an important predictor even when taking basic fundamental psychological needs or Allport's conditions into account. In fact, our core situational needs measure predicted outgroup attitudes at least as good as Allport's conditions and consistently explained more variance than commonly measured psychological needs. In most cases, the core need even took over the variances previously explained by the self-determination theory needs (see \appref{app:AppendixRobustness}). We thus find strong evidence that within everyday life interactions of recent migrants with majority outgroup members, the perception that one's interaction-specific needs are fulfilled offers a meaningful and flexible predictor of interaction quality, outgroup attitudes, and well-being.

\subsection{Limitations}
% (1) Sample = minority and (voluntary) migrants, (2) Methodology = short scales (reliability) and no lagged effects, (3) Core Need Concept = non-specific
While we believe that a need fulfillment mechanism should be relevant to any inter-group contact, our samples focused on a minority- and (voluntary) migrant perspective. Without additional evidence it thus remains difficult to judge whether motivational effects will generalize to other migrant groups (e.g., forced migrants), other intergroup contexts (e.g., gender-, religious-, or sexual orientation groups), or to majority groups and their outgroup attitudes. We sought to replicate our results in three studies with different types of migrants but the fact remains that all three studies had slightly more women than men participating and were younger and more educated samples overall. While the samples were representative of the migrant group to the focus region, the generalizability of the sample is restricted by its characteristics. We know of no research suggesting that in other contexts need fulfillment would be less relevant but future research may extend our findings to build an even broader understanding of need fulfillment in intergroup contacts. Researchers may even seek to explore the role of need fulfillment in real-world interactions more broadly \citep[also see][]{Downie2008}.

A second limitation lies in our methodology. While intensive longitudinal data is close to real-life events, this method comes at the expense of longer and more robust scales. Long and repetitive scales are often not feasible in intensive longitudinal methods because of the increased burden to the participants. To circumvent this shortcoming, we have ensured that the measures we used were, whenever possible, based on past validations. However, the circumstance remains that intensive longitudinal data often does not allow the same scrutiny of measurement reliability as single–shot cross-sectional data sets. An additional methodological question lies in the unexplored potential of the longitudinal aspects of our data. For our research questions, we have focused on contemporaneous effects within the data set, yet future investigations should seek to extend the mechanism to developmental trajectories within and between participants. 

A third limitation lies in the outcome variable we chose. As we have focused on outgroup attitudes to ensure comparability to the past literature of the contact hypothesis and to replicate the most reliable patterns within ESM data \citep[][]{Pettigrew2006}. However, especially for disadvantaged minority members positive outgroup attitudes may entail negative downstream consequences such as a reduced endorsement of social change --- even in the face of injustice \citep[e.g.,][]{dixon2012}. We have additionally tested our need-fulfillment mechanism for well-being reports to test the direct benefits of the mechanism to lived realities. However, we are among the first to collect intensive longitudinal data on the experience of minority migrants and it remains an open empirical question whether the more positive outgroup attitudes following need-fulfilling interactions might ironically exacerbate inequalities for the disadvantaged group \citep[also see][]{reimer2023}. 

Finally, our conceptualization of core situational needs has been focused on the most essential test of a motivational mechanism. This comes at the expense of specificity in the situational motives (i.e., we have not explored whether different individual motives have stronger effects on interaction quality and outgroup attitudes). Such an investigation would be possible with the adaptive measurement we used (e.g., by looking at the differential effects of the clustered motives) but would not have been relevant to our theory–focused research question. Identifying cases where a specific minority faces a common need frustration could be instrumental in addressing systemic challenges. Future research may, thus, explore which situations activate or threaten specific motives \citep[e.g.,][]{gollwitzer1985e, leander2020} and which exact motives are most important in different intergroup contexts. 


\subsection{Implications}
% ESM data
Despite these limitations, we can nonetheless draw a number of implications for other researchers and practitioners --- ranging from the benefits of longitudinal data to theoretical implications. A first implication concerns the feasibility and usefulness of intensive longitudinal data for intergroup contact research and the broader field of social psychology. While setting up an intensive longitudinal study is not easy, we believe the efforts to be similar to a sizable cross-sectional data collection (i.e., for a longitudinal or a high-quality cross-sectional data set with over 3,000 intergroup interactions captured and over 10,000 data points in total). Intensive longitudinal data, thus, opens up the possibility to explore research questions that focus on real-life phenomena outside the lab or focus on phenomena that depend on changes and influences over time. In the context of intergroup contact research, we are among the first to answer calls to test intergroup contact mechanisms using extended real-life data \citep[e.g.,][]{Pettigrew2011, MacInnis2015}. In doing so, we not only collected an unprecedented amount of real-life data, but our consideration of intensive longitudinal data may present new inconsistencies in how participants perceive and cognitively aggregate their past interactions with other groups — which may suggest large-scale recall biases or conflations of within and between participant effects in conventional cross-section studies.

% Motivation and core needs 
A broader theoretical implication relates to the role of situational motivation in intergroup contacts. Our results offer a first promising test of a psychological mechanism of need fulfillment in intergroup contact. While our results are tentative given their novelty within the field, they were highly consistent across studies and may offer new theoretical avenues. Experiences of need fulfillment are a facet of the human experience that has thus far been underemphasized in the intergroup contact literature. This stands in stark contrast to the many cognitive \citep[e.g.,][]{Pettigrew1998, Brown2005} and emotional aspects investigated within the field \citep[e.g.,][]{Stephan2008}. Future research may, therefore, be able to integrate broader theoretical frameworks of intergroup contact (e.g., motivations guiding cognition and affect, which in turn drive behavior. cf., theory of reasoned goal pursuit; \citealp{Ajzen2019}. Also see \citealp{Kreienkamp2022d}). 

% practical use of needs
Additionally, situational motivations in intergroup contact also offer promising avenues for practitioners and policy-makers. Intergroup contact theory is among the most implemented psychological theories \citep[e.g.,][]{Pettigrew2006, AlRamiah2012a, Reimer2021}. Given our findings that need fulfillment in everyday intergroup contacts was at least as powerful as Allport's conditions in predicting outgroup attitudes, considerations of people's needs offer a substantially more immediate mechanism to address. In cases where some or all optimal contact conditions are not possible to be fulfilled, needs offer an even more compelling alternative (e.g., where equal status is contextually not possible or in cases where people help despite a lack of institutional support). Additionally, our conceptualization of situational needs offers a clear opportunity for practitioners and interventions. Instead of addressing needs as a one-size fits all solution (e.g., simply focusing on competence needs), one may at times ask outgroup interaction partners what they need during an interaction. This is not to say that we should not explore which motives tend to be relevant to specific groups in specific intergroup contact contexts. Rather, during interventions for which data on important need contents are not available or infeasible to collect, a flexible and reactive approach of inquiring momentary intergroup contact needs might be more fruitful.


\subsection{Conclusion}
% Conclusion paragraph
In sum, we used intensive longitudinal methodologies to capture real-life interactions of recent migrants with the majority outgroup. Our three studies showcase the feasibility and utility of such data to test intergroup contact theory. We provide evidence that the fulfillment of situational needs during real-life intergroup contacts meaningfully predicts perceived interaction quality and positive outgroup attitudes. Our results point to motivational needs as an understudied aspect of intergroup contact that is important in understanding when and why an interaction is perceived as positive and will lead to more positive outgroup attitudes.


% Tables
\input{Tables/descrFullWide}
\input{Tables/descrOutWide}

\input{Tables/mdlContactGeneralLong}
\input{Tables/mdlTheorylong}

% Figures
\begin{figure}
  \caption{Partial Regression Diagrams of Situational Needs Model across Studies}
  \label{fig:MainPaths}
  \begin{center}
    \input{PathModels.tex}
  \end{center}
  \caption*{Note: Coefficients are standardized (partial) regression coefficients. Statistical significance markers are based on the unstandardized regression results (as presented in \tblref{tab:intergroupNeedsTblLong}). Note that we do not test a mediation model. The diagram only illustrates the included concepts and partial regression parameters.\\
  **** p < .0001, *** p < .001, ** p < .01, * p < .05}
\end{figure}

\begin{figure}
  \caption{Contact Hypothesis}
  \label{fig:ContactHypothesis}
  \centering\includegraphics[width=\textwidth]{Figures/forestParametricREMLGeneralLmer.png}
  %\begin{subfigure}{\textwidth}
  %  \caption{}
  %  \centering\includegraphics[width=0.75\textwidth]{Figures/forestParametricGeneralLm.png}
  %\end{subfigure}
  %\begin{subfigure}{\textwidth}
  %  \caption{}
  %  \centering\includegraphics[width=0.75\textwidth]{Figures/forestParametricFEGeneralLmer.png}
  %\end{subfigure}
  \caption*{Note: \\
  Summary of mixed models results of the contemporaneous contact effects.\\
  General: Random effects meta-analytic results are presented for completeness.}
\end{figure}

\begin{figure}
  \caption{Core Need Fulfillment}
  \label{fig:AllportNeedFulfillment}
  \centering\includegraphics[width=\textwidth]{Figures/forestParametricTheoryComb.png}
  %\begin{subfigure}{\textwidth}
  %  \caption{}
  %  \centering\includegraphics[width=0.6\textwidth]{Figures/forestParametricFETheoryQualityCore.png}
  %\end{subfigure}
  %\begin{subfigure}{\textwidth}
  %  \caption{}
  %  \centering\includegraphics[width=0.6\textwidth]{Figures/forestParametricFETheoryAttitudeCore.png}
  %\end{subfigure}
  %\begin{subfigure}{\textwidth}
  %  \caption{}
  %  \centering\includegraphics[width=0.6\textwidth]{Figures/forestParametricFETheoryAttitudeCoreQuality.png}
  %\end{subfigure}
  \caption*{Note: \\
  (a) Core Need Fulfillment predicting Interaction Quality.\\
  (b) Core Need Fulfillment predicting Outgroup Attitudes.\\
  (c) Core Need Fulfillment and Interaction Quality predicting Outgroup Attitudes.\\
  General: Random effects meta-analytic results are presented for completeness.}
\end{figure}


\printbibliography

\appendix

\section{Hypotheses and Analysis Plan}
\label{app:AppendixHypotheses}

\newlength{\mdfmar}
\setlength{\mdfmar}{1.5em}
\mdfdefinestyle{mdfhypothesis}{
    innerleftmargin = +1.5\mdfmar, %
    innerrightmargin = +1.5\mdfmar, 
    innertopmargin = +\mdfmar, 
    innerbottommargin = +\mdfmar, 
    skipabove = 12pt
}
\newlength{\subhypskip}
\setlength{\subhypskip}{1.5em}
\newlength{\eqskip}
\setlength{\eqskip}{3.5em}

In this appendix, we present the expanded hypotheses and their associated analysis plan. Given the nested structure of much of our data, we test many of our hypotheses using a multilevel approach, where $y_{ti}$ denotes the response at measurement occasion $t$ ($t = 1, ..., T_i$; level 1) for individual $i$ ($i = 1, ..., n$; level 2). All multilevel assumptions are tested as follows (e.g., for random slopes model with $j$ within-person predictors):

\begin{flalign}
  &\textrm{Level 1 Variance:}\ e_{ti} \sim \mathcal{N}(0,\sigma^2) \\
  &\textrm{Level 2 Variance:}\ \begin{bmatrix} u_{0i}\\ \vdots\\ u_{ji}\end{bmatrix} 
  \sim \mathcal{N}
  \begin{pmatrix}
    \begin{bmatrix} 
      0 \\ 
      \vdots \\
      0
    \end{bmatrix}, 
    \begin{bmatrix} 
      \tau_{00}^2 &  & \\ 
      \vdots & \ddots & \\
      \tau_{j0} & \ldots & \tau_{jj}^2
    \end{bmatrix}
  \end{pmatrix}
\end{flalign}

For our main aims we sequentially focus on four sets of models to test and validate our hypotheses:

\subsection{1. Contact hypothesis in intensive longitudinal data.} 
Within the individual studies, we begin by testing the most basic assumption of the intergroup contact hypothesis that outgroup attitudes should be more positive after outgroup interactions but not after non-outgroup interactions. For this we use multilevel regression analyses, predicting outgroup attitudes from outgroup interaction and non-outgroup interaction dummy variables. We also include the participant means as level two predictors to fully disentangle within- (level 1) and between-participant (level 2) effects of intergroup contact \citep[e.g.,][Section 4.6]{snijders2012}.

\begin{mdframed}[style=mdfhypothesis]
    \begin{hyp}[H\ref{hyp:contactHyp}] \label{hyp:contactHyp}
        Based on the most general understanding of the contact hypothesis, positive intergroup contacts should be associated with more favorable outgroup attitudes across intensive longitudinal data.
    \end{hyp}
    
    \begin{subhyp}[H\ref{hyp:contactDummyML}] \label{hyp:contactDummyML}
    \addtolength{\leftskip}{\subhypskip}
    Outgroup attitudes should be more positive after an intergroup interaction compared to a non-outgroup interaction.
    \end{subhyp}

    \begin{fleqn}[\eqskip] 
    \begin{equation} \label{eq:ContactDummy}
      \begin{split}
        \textrm{Level 1:} &
          \begin{aligned}[t]
            \ Attitude_{ti} =  &\ \beta_{0i} + \beta_{1i}OutgroupInteraction_{ti} + \\
                               &\ \beta_{2i}NonOutgroupInteraction_{ti} + e_{ti}
          \end{aligned} \\
        \textrm{Level 2:} &
            \begin{aligned}[t]
                \ \beta_{0i} = &\ \gamma_{00} + \gamma_{01}MeanOutgroupInteraction_{i} + \\
                               &\ \gamma_{02}MeanNonOutgroupInteraction_{i} + u_{0i} \\
                  \beta_{1i} = &\ \gamma_{10} + u_{1i} \\
                  \beta_{2i} = &\ \gamma_{20} + u_{2i}
            \end{aligned}
      \end{split} 
    \end{equation}
    \end{fleqn}
\end{mdframed}

% We then seek to investigate intergroup contacts and perceived interaction quality jointly. Notably, however, participants are only able to report their interaction quality perceptions if they had an interaction. This is considered to be structurally missing data and cannot meaningfully be imputed for modeling. To deal with this issue, we mimic the procedure commonly used within the cross-section literature. In particular, we conduct a person-level linear regression by aggregating the number of outgroup interactions participants had, their average interaction quality perceptions, as well as their average outgroup attitudes \citep[for possible alternative approaches see,][]{Enders2011}. To avoid an underpowered analysis, we use the aggregated data from all three studies. 
    
We then seek to test the full contact hypothesis by investigating intergroup contacts and perceived interaction quality jointly. To do so, we conduct a linear regression using person-level aggregated data from all three studies. In particular, we aggregate the number of outgroup interactions participants had, their average interaction quality perceptions, as well as their average outgroup attitudes. This approach has three main benefits: (1) Interaction quality ratings are only available if participants had an interaction, and the aggregation deals with this structural missingness. (2) Using the participant-level data from all three studies, we avoid potential power issues. (3) This analysis mimics the analyses conducted within the cross-section literature, where participants are asked to recall how many interactions they had over a one-month period, how positive these interactions were, and what their general attitudes towards the outgroup are.

\begin{mdframed}[style=mdfhypothesis]
    \begin{subhyp}[H\ref{hyp:contactCor}] \label{hyp:contactCor}
    \addtolength{\leftskip}{\subhypskip}
    Participants with more intergroup interactions should have more favorable outgroup attitudes.
    \end{subhyp}

      \begin{fleqn}[\eqskip] 
        \begin{equation} \label{eq:ContactCor}
            r_{\left(ContactFrequency_{i},\ AverageQuality_{i}\right)} > 0
        \end{equation}
      \end{fleqn}

    \begin{subhyp}[H\ref{hyp:contactQualLM}] \label{hyp:contactQualLM}
    \addtolength{\leftskip}{\subhypskip}
    Participants with more intergroup interactions should have more favorable outgroup attitudes depending on the average interaction quality.
    \end{subhyp}

      \begin{fleqn}[\eqskip] 
        \begin{equation} \label{eq:contactQualLM}
            \begin{split}
              AverageAttitude_{i} = &\ \beta_{0} + \beta_{1}ContactFrequency_{i} + \\
                                    &\ \beta_{2}AverageQuality_{i} +\\
                                    &\ \beta_{3}ContactFrequency_{i} * AverageQuality_{i}
            \end{split}
        \end{equation}
      \end{fleqn}  
      
      We additionally control for the participant's study membership.
\end{mdframed}
Because this analysis uses the data from all three studies, the results of this analysis are presented in the `Robustness, Stability, and Embeddedness across Studies' section.

\subsection{2. Core need fulfillment during intergroup contact.}
The main proposal of this manuscript has been the assertion that the fulfillment of situation core needs during an interaction will be associated with more positive interaction quality perceptions and, ultimately, more positive outgroup attitudes. Thus, for the main set of analyses, we focus on the reported outgroup interactions only. For each study, we will use multilevel regressions to test the main three assertions of our proposal (mirroring the basic steps of a traditional mediation analysis).

\begin{mdframed}[style=mdfhypothesis]
    \begin{hyp}[H\ref{hyp:keyNeed}] \label{hyp:keyNeed}
    Based on our proposal, intergroup interactions with higher situational core need fulfillment should predict more favorable outgroup attitudes due to more positive interaction quality perceptions.
    \end{hyp}
    
    \setcounter{subhyp}{0}
    \begin{subhyp}[H\ref{hyp:keyNeedPred}] \label{hyp:keyNeedPred}
    \addtolength{\leftskip}{\subhypskip}
    Core need fulfillment during outgroup interactions should predict more positive outgroup attitudes.
    \end{subhyp}
    
    \begin{fleqn}[\eqskip]
      \begin{equation} \label{eq:SlopesAttCore}
        \begin{split}
            \textrm{Level 1:} &\ Attitude_{ti} = \beta_{0i} + \beta_{1i}KeyNeedFulfill_{ti} + e_{ti}\\
            \textrm{Level 2:} &\ \beta_{0i} = \gamma_{00} + u_{0i} \\
                              &\ \beta_{1i} = \gamma_{10} + u_{1i}
        \end{split}
      \end{equation}
    \end{fleqn}
    
    \begin{subhyp}[H\ref{hyp:keyNeedQual}] \label{hyp:keyNeedQual}
    \addtolength{\leftskip}{\subhypskip}
    Core need fulfillment during outgroup interactions should also predict higher perceived interaction quality.
    \end{subhyp}
    
    \begin{fleqn}[\eqskip]
      \begin{equation} \label{eq:SlopesQltCore}
        \begin{split}
            \textrm{Level 1:} &\ InteractionQuality_{ti} = \beta_{0i} + \beta_{1i}KeyNeedFulfill_{ti} + e_{ti}\\
            \textrm{Level 2:} &\ \beta_{0i} = \gamma_{00} + u_{0i} \\
                              &\ \beta_{1i} = \gamma_{10} + u_{1i}
        \end{split}
      \end{equation}
    \end{fleqn}
    
    \begin{subhyp}[H\ref{hyp:keyNeedMediation}] \label{hyp:keyNeedMediation}
    \addtolength{\leftskip}{\subhypskip}
    The effect of core need fulfillment on outgroup attitudes should be reduced when considered together with perceived interaction quality.
    \end{subhyp}
    
    \begin{fleqn}[\eqskip]
      \begin{equation} \label{eq:SlopesAttCoreQual}
        \begin{split}
          \textrm{Level 1:} &
            \begin{aligned}[t]
              \ Attitude_{ti} =  &\ \beta_{0i} + \beta_{1i}KeyNeedFulfill_{ti} + \\
                                 &\ \beta_{2i}InteractionQuality_{ti} + e_{ti}
            \end{aligned} \\
          \textrm{Level 2:} &\ \beta_{0i} = \gamma_{00} + u_{0i} \\
                            &\ \beta_{1i} = \gamma_{10} + u_{1i} \\
                            &\ \beta_{2i} = \gamma_{20} + u_{2i}
        \end{split} 
      \end{equation}
    \end{fleqn}
\end{mdframed}

\subsection{3. Allport's conditions in intensive longitudinal data.}
Within the third study, we formally measure all of Allport's optimal contact conditions. We use multilevel regression models to test whether the fulfillment of Allport's conditions in real-life data predicts more positive outgroup attitudes and higher perceived interaction quality, using the same approach as for the core need fulfillment above.

\begin{mdframed}[style=mdfhypothesis]
    \begin{hyp}[H\ref{hyp:Allport}] \label{hyp:Allport}
    Based on Allport's optimal contact conditions, intergroup interactions with equal status, common goals, collaboration, and structural support should predict more favorable outgroup attitudes due to more positive interaction quality perceptions.
    \end{hyp}
    
    \setcounter{subhyp}{0}
    \begin{subhyp}[H\ref{hyp:AttAllport}] \label{hyp:AttAllport}
    \addtolength{\leftskip}{\subhypskip}
    Higher fulfillment of Allport's conditions during outgroup interactions should predict more positive outgroup attitudes.
    \end{subhyp}
    
    \begin{fleqn}[\eqskip]
      \begin{equation} \label{eq:SlopesAttAllport}
        \begin{split}
            \textrm{Level 1:} &\ Attitude_{ti} = \beta_{0i} + \beta_{1i}Allport_{ti} + e_{ti}\\
            \textrm{Level 2:} &\ \beta_{0i} = \gamma_{00} + u_{0i} \\
                              &\ \beta_{1i} = \gamma_{10} + u_{1i}
        \end{split}
      \end{equation}
    \end{fleqn}
    
    \begin{subhyp}[H\ref{hyp:QltAllport}] \label{hyp:QltAllport}
    \addtolength{\leftskip}{\subhypskip}
    Higher fulfillment of Allport's conditions during outgroup interactions should also predict higher perceived interaction quality.
    \end{subhyp}
    
    \begin{fleqn}[\eqskip]
      \begin{equation} \label{eq:SlopesQltAllport}
        \begin{split}
            \textrm{Level 1:} &\ InteractionQuality_{ti} = \beta_{0i} + \beta_{1i}Allport_{ti} + e_{ti}\\
            \textrm{Level 2:} &\ \beta_{0i} = \gamma_{00} + u_{0i} \\
                              &\ \beta_{1i} = \gamma_{10} + u_{1i}
        \end{split}
      \end{equation}
    \end{fleqn}
    
    \begin{subhyp}[H\ref{hyp:AttAllportQual}] \label{hyp:AttAllportQual}
    \addtolength{\leftskip}{\subhypskip}
    The effect of higher fulfillment of Allport's conditions on outgroup attitudes should be reduced when considered together with perceived interaction quality.
    \end{subhyp}
    
    \begin{fleqn}[\eqskip]
      \begin{equation} \label{eq:SlopesAttAllportQual}
        \begin{split}
          \textrm{Level 1:} &
            \begin{aligned}[t]
              \ Attitude_{ti} =  &\ \beta_{0i} + \beta_{1i}Allport_{ti} + \\
                                 &\ \beta_{2i}InteractionQuality_{ti} + e_{ti}
            \end{aligned} \\
          \textrm{Level 2:} &\ \beta_{0i} = \gamma_{00} + u_{0i} \\
                            &\ \beta_{1i} = \gamma_{10} + u_{1i} \\
                            &\ \beta_{2i} = \gamma_{20} + u_{2i}
        \end{split} 
      \end{equation}
    \end{fleqn}
\end{mdframed}

We then compare the effect of Allport's contact conditions with our core need fulfillment by comparing the model fit statistics of the two individual models and by adding both concepts to a joint multilevel regression model, to see whether the two approaches explain the same variance in outgroup attitudes.

\begin{mdframed}[style=mdfhypothesis]
    \begin{hyp}[H\ref{hyp:NeedAllport}] \label{hyp:NeedAllport}
    Based on our proposal, the fulfillment of the core situational need should predict outgroup attitudes at least as well as Allport's conditions.
    \end{hyp}

    \setcounter{subhyp}{0}
    \begin{subhyp}[H\ref{hyp:compModel}] \label{hyp:compModel}
    \addtolength{\leftskip}{\subhypskip}
    The need model (H\ref{hyp:keyNeedPred}) should predict more variance in outgroup attitudes than the model based on Allport's conditions (H\ref{hyp:AttAllport}).
    \end{subhyp}

    \begin{fleqn}[\eqskip] 
        \begin{equation}
            \begin{split}
                AIC_{KeyNeedModel} & < AIC_{AllportModel} \\
                BIC_{KeyNeedModel} & < BIC_{AllportModel}
            \end{split}
        \end{equation}
    \end{fleqn}

    \begin{subhyp}[H\ref{hyp:compTogether}] \label{hyp:compTogether}
    \addtolength{\leftskip}{\subhypskip}
    The  effect of key need fulfillment on outgroup attitudes should  persist even when taking other Allport's conditions into account. Thus, the effect of key need fulfillment on outgroup attitudes should remain strong even after controlling for Allport's conditions.  
    \end{subhyp} 
    
    \begin{fleqn}[\eqskip]
      \begin{equation} \label{eq:SlopesAttCoreAllport}
        \begin{split}
          \textrm{Level 1:} &
            \begin{aligned}[t]
              \ Attitude_{ti} =  &\ \beta_{0i} + \beta_{1i}KeyNeedFulfill_{ti} + \\
                                 &\ \beta_{2i}Allport_{ti} + e_{ti}
            \end{aligned} \\
          \textrm{Level 2:} &\ \beta_{0i} = \gamma_{00} + u_{0i} \\
                            &\ \beta_{1i} = \gamma_{10} + u_{1i} \\
                            &\ \beta_{2i} = \gamma_{20} + u_{2i}
        \end{split} 
      \end{equation}
    \end{fleqn}
\end{mdframed}

\subsection{4. Robustness, stability, and embeddedness across studies}
Within the final set of analyses, we look at the broader picture of our results and leverage the data from all participants to contextualize our results.  

\paragraph{Robustness within studies}
To build further confidence in the effect of core need fulfillment during outgroup interactions, we conduct two additional robustness analyses for each study.

Firstly, to check for the role of alternate psychological needs, we add the fulfillment of self-determination theory needs (i.e., competence, autonomy, and relatedness) to the multilevel regression. We then also compare the model with models that predicts outgroup attitudes from self-determination theory need fulfillments or core need fulfillments only. 

\begin{mdframed}[style=mdfhypothesis]
    \begin{hyp}[H\ref{hyp:keyNeedSDT}] \label{hyp:keyNeedSDT}
    The effect of key need fulfillment on outgroup attitudes should persist even when taking other fundamental psychological needs into account. Thus, the effect of key need fulfillment on outgroup attitudes should remain strong even after controlling for autonomy, competence, and relatedness fulfillment during the interaction (cf., self-determination theory). 
    \end{hyp}
    
    \begin{fleqn}[\eqskip-\subhypskip]
      \begin{equation} \label{eq:SlopesAttCoreSdt}
        \begin{split}
          \textrm{Level 1:} &
            \begin{aligned}[t]
              \ Attitude_{ti} =  &\ \beta_{0i} + \beta_{1i}KeyNeedFulfill_{ti} + \beta_{2i}Autonomy_{ti} + \\
                                 &\ \beta_{3i}Competence_{ti} + \beta_{4i}Relatedness_{ti} + e_{ti}
            \end{aligned} \\
          \textrm{Level 2:} &\ \beta_{0i} = \gamma_{00} + u_{0i} \\
                            &\ \beta_{1i} = \gamma_{10} + u_{1i} \\
                            &\ \beta_{2i} = \gamma_{20} + u_{2i} \\
                            &\ \beta_{3i} = \gamma_{30} + u_{3i} \\
                            &\ \beta_{4i} = \gamma_{40} + u_{4i}
        \end{split} 
      \end{equation}
    \end{fleqn}
\end{mdframed}

To ensure that the core need fulfillment is outgroup contact specific, we return to the full sample of intensive longitudinal measurements within each study and test whether there is an interaction effect of outgroup contact (vs. no outgroup contact) and core need fulfillment. We expect that the effect of core need fulfillment is specific to outgroup interactions and not merely due to a more need-fulfilled life in general.

\begin{mdframed}[style=mdfhypothesis]
    \begin{hyp}[H\ref{hyp:keyNeedContactType}] \label{hyp:keyNeedContactType}
    \addtolength{\leftskip}{1em}
    The effect of key need fulfillment on outgroup attitudes should be specific to intergroup interactions and not be due to need fulfillment in general. Thus, the effect of key need fulfillment on outgroup attitudes should be stronger for intergroup interactions than for ingroup interactions. 
    \end{hyp}
    
    \begin{fleqn}[\eqskip-\subhypskip]
      \begin{equation} \label{eq:SlopesAttCoreXContact}
        \begin{split}
          \textrm{Level 1:} &
            \begin{aligned}[t]
              \ Attitude_{ti} =  &\ \beta_{0i} + \beta_{1i}KeyNeedFulfill_{ti} + \\
                                 &\ \beta_{2i}OutgroupInteraction_{ti} + \\
                                 &\ \beta_{3i}KeyNeedFulfill*OutgroupInteraction_{ti} + e_{ti}
            \end{aligned} \\
          \textrm{Level 2:} &\ \beta_{0i} = \gamma_{00} + u_{0i} \\
                            &\ \beta_{1i} = \gamma_{10} + u_{1i} \\
                            &\ \beta_{2i} = \gamma_{20} + u_{2i} \\
                            &\ \beta_{3i} = \gamma_{30} + u_{3i}
        \end{split} 
      \end{equation}
    \end{fleqn}
\end{mdframed}
The results of the robustness analyses are presented in \appref{app:AppendixRobustness} to allow for a more concise presentation of our main hypotheses in the main text.

\paragraph{Stability across studies}
We assess the stability of our main analyses. We use forest plots (including meta-analytical estimates) to visualize the direction and effect sizes of our three studies.

\begin{mdframed}[style=mdfhypothesis]
    \begin{hyp}[H\ref{hyp:Stability}] \label{hyp:Stability}
    \addtolength{\leftskip}{1em}
    The effects of our main hypotheses and robustness analyses should be consistent across studies.
    \end{hyp}
\end{mdframed}

\paragraph{Embeddedness of code needs}
We, finally, use the qualitative data from the participants' self-identified core needs to contextualize our results. We leverage machine learning to extract a topic model of the free-text entries across the three studies. We describe the extracted topics and themes and compare them to the need contents usually found and measured within the psychological literature. Full methodological details and visualizations are available in Supplemental Material C.

\begin{mdframed}[style=mdfhypothesis]
    \addtolength{\leftskip}{1em}
    \textit{This analysis is data-driven and exploratory. As such, the analysis has no associated hypothesis.}
\end{mdframed}



\section{Robustness Analyses}
\label{app:AppendixRobustness}

In this appendix, we present the empirical details of our additional robustness analyses. These analyses are specifically designed to check for alternative models and contextualize our results. We (1) check whether core need fulfillment is indeed outgroup contact specific. For this analysis, we return to the full sample of intensive longitudinal measurements and test whether there is an interaction effect of outgroup contact (vs. no outgroup contact) and core need fulfillment. We expect that the effect of core need fulfillment is specific to outgroup interactions and not merely due to a more need-fulfilled life in general. We then check (2) whether the need mechanism is relevant to both planned and accidental outgroup interactions, and (3) extends to the more individual-focused experience of well-being. In a final set of analyses, we (4) check whether the need fulfillment mechanism is relevant to different types of need content (i.e., motives) and (5) remains relevant even when accounting for the fulfillment of fundamental psychological needs (i.e., self-determination theory needs: competence, autonomy, and relatedness).

\input{Robustness-Methods-and-Results.tex}

% Tables
\input{Tables/descrFullWideAppB.tex}
\input{Tables/descrOutWideAppB.tex}

\subsection{Conclusion}
Across the wide variety of robustness analyses, we thus find that the experience of need fulfillment is a robust and flexible predictor of positive outgroup attitudes even when accounting for a range of other and even alternate predictors. However, we also find that the core need fulfillment mechanism does not account for all need-related variance in outgroup attitudes. Notably, the fulfillment of relatedness (and to some extent competence) needs explained additional variance in outgroup attitudes. Nonetheless, the situational core need fulfillment remained extremely reliably a core predictor of outgroup attitudes.

\input{Tables/mdlRobustlong}

% \begin{figure}
%   \caption{Robustness Analyses}
%   \label{fig:Robustness}
%   \centering\includegraphics[width=\textwidth]{Figures/forestParametricRobustnessComb.png}
%   %\begin{subfigure}{\textwidth}
%   %  \caption{}
%   %  \centering\includegraphics[width=0.65\textwidth]{Figures/forestParametricFERobustContact.png}
%   %\end{subfigure}
%   %\begin{subfigure}{\textwidth}
%   %  \caption{}
%   %  \centering\includegraphics[width=0.65\textwidth]{Figures/forestParametricFERobustSDT.png}
%   %\end{subfigure}
%   \caption*{Note: \\
%   (a) Need Fulfillment and Intergroup Contact predicting Outgroup Attitudes (full sample).\\
%   (b) Core Need Fulfillment predicting Outgroup Attitudes while controlling for self-determination theory needs (intergroup contact sample).\\
%   General: Random effects meta-analytic results are presented for completeness.}
% \end{figure}


\end{document}