Micro-inequities and immigration backgrounds in the software industry

Micro-inequities are subtle, repetitive, and often unintentional forms of negative messaging, that can account for a significant burden over time. Research shows that racial and gender minority groups are more likely to experience micro-inequities, and that micro-inequities have a significant negative effect on self-esteem, work performance and career advancement. However, research on micro-inequities among software practitioners, particularly with an immigration perspective, is non-existent. To bridge this gap, we investigate the experiences of software practitioners, regarding micro-inequities from an immigration perspective. We surveyed 135 immigrant and non-immigrant software practitioners working in technical roles about verbal, nonverbal and environmental micro-inequities. Our results show that immigrants experience nine out of 27 investigated forms of micro-inequities significantly more than non-immigrants. These include not being given credit for their work, feeling excluded from key social or networking opportunities and being assumed to be less competent, assertive or intelligent. Our study can serve as an incentive for practitioners to adopt (more) inclusive work practices and to raise awareness about micro-inequities in the community. Lay Abstract: Micro-inequities are subtle, repetitive, and often unintentional forms of negative messaging, that can account for a significant burden over time. While research shows that under-represented groups are more likely to experience micro-inequities, they have not been extensively studied in the software industry. In our work we survey 135 software practitioners regarding their experiences with micro-inequities in the software industry from an immigration perspective. Our findings show that immigrants experience some forms of micro-inequities significantly more than non-immigrants.


INTRODUCTION
Micro-inequities are small, covert, often unintentional, and hard to-prove events that occur whenever people are perceived to be different [37].These events come in form of subtle and sometimes subconscious messages which can vary in the way they are delivered, including spoken words and sentences, looks, gestures, and tones [18].The "micro" in "micro-inequities" could suggest that these inequities are trivial [38]; however, their repetitive nature and cumulative effects can account for a significant burden over time.This can ultimately result in a formidable barrier in the performance, productivity, and advancement of the affected individuals leading to damages to their self-esteem and eventual departure from the workplace [17,18].
Immigration has been a major factor in the increase of ethnic and racial diversity in the Global North during the second half of the 20th century, and will probably continue to be so in the foreseeable future [7,24].This trend is also applicable to software development teams which are increasingly becoming more ethnically diverse due to the growth in distributed software development and a globally mobile labor force.Software teams that are ethnically diverse show improved innovation and problem solving results, but only if measures are taken to enable all team members to contribute [10].Previous studies show the existence of ethnic or race related micro-inequities [9] in other fields rather than the software industry [6,12,14,19,27,37,40,42,45], and gendered micro-inequities in the software industry [16].However, there is no research on how software practitioners with an immigration background experience micro-inequities in the software industry.
We address this gap and investigate software practitioners' experiences with micro-inequities, as well as whether these experiences differ between immigrant and non-immigrant software practitioners.For this purpose, we conduct a questionnaire-based survey and report on the responses of 135 software practitioners working in technical roles.Using a typology from related literature [44,45], we make a distinction between three types of micro-inequities namely verbal, nonverbal, and environmental micro-inequities.We use statistical tests to analyze if and how immigrant practitioners experience micro-inequities differently than non-immigrants.Furthermore, we report on differences in experiences between first-and secondgeneration immigrants, and immigrants from the Global South and Global North.
This study is, to our best knowledge, the first to study immigration and micro-inequities in the software industry.Its main contributions are (i) empirical evidence on immigration and microinequities in the software industry, (ii) a survey instrument that can be used by future researchers studying micro-inequities at the workplace, and (iii) responses from 135 software practitioners to be used for further analysis in any future studies.
To encourage replication and further research we make our collected data and survey instrument available. 1

SCOPE AND RESEARCH QUESTIONS
We investigate the experiences of software practitioners of differing immigration backgrounds regarding micro-inequities in the software industry.Using a typology from related literature [44,45], we make a distinction between three types of micro-inequities, namely verbal, nonverbal, and environmental micro-inequities.Our research questions are: RQ1: Do software practitioners experience verbal microinequities at the workplace?How do these experiences differ between immigrant and non-immigrant software practitioners?
Verbal micro-inequities between colleagues include detrimental verbal messages about intelligence, speech, and appearance.Other examples of verbal micro-inequities are negative reactions towards accents or appearance, and raised voices during discussions.RQ2: Do software practitioners experience nonverbal microinequities at the workplace?How do these experiences differ between immigrant and non-immigrant software practitioners?Nonverbal micro-inequities are behavioral micro-inequities delivered in non-spoken form.Examples of nonverbal micro-inequities are lack of receptiveness to suggested ideas, or lack of eye contact during meetings or discussions.RQ3: Do software practitioners experience environmental micro-inequities at the workplace?How do these experiences differ between immigrant and non-immigrant software practitioners?Environmental micro-inequities are subtle micro-messaging that occurs on the systemic and environmental level.These inequities create an institutional or organizational climate that is more favorable to one person than the other [6], [44].Examples of environmental micro-inequities are being assigned unreasonable deadlines, or being excluded from key social or networking opportunities.RQ4: Do first generation and second generation immigrant practitioners experience micro-inequities to different degrees?
In our study, we use the terms immigrants or first-generation immigrants to refer to people working in another country than their home country.We consider second-generation immigrants as folk working in their home country, who have at least one parent that was born in a different country than the one they are currently working in.Note that second-generation immigrants are in fact non-immigrants, as they work in their home country and conflating their experiences together with first generation immigrants is problematic [23].We followed the guidelines proposed by Statistics Netherlands [1], a Dutch government institution, for the consideration of first-and second-generation immigrants.RQ5: Do Global South and Global North immigrants experience micro-inequities to different degrees?
We study if immigrants from different global socio-economical regions experience micro-inequities to differring degrees.We follow

METHODOLOGY
We designed a questionnaire-based survey to answer the research questions of our study.Below we describe its target population, design and distribution, as well as the data analysis process.

Population
Our target population includes practitioners working in the software industry in a technical role.Software developer, software architect, software consultant, product owner, project manager, tester, data scientist, or data engineer are examples of roles included in our target population.To ensure that we collected responses from our target population, we explicitly asked participants if they are working in the software industry in a technical role (see Table 3) at the beginning of our survey.

Survey Design
The survey consists of a series of 27 close-ended questions related to micro-inequities, all shown in Table 1.Some questions are (adaptations of) survey questions from previous work studying microinequities: [2,5,16,30,43] 2 , others were carefully adapted from surveys that studied general team dynamics in tech [47,48] and the software industry [22] through a gender perspective.While these latter studies did not explicitly name the actions as micro-inequities, they fit our definitions of verbal, nonverbal and environmental micro-inequities (see Section 2).To ensure that the immigration background of our participants was correctly considered, we explicitly asked participants to answer the micro-inequities questions based on their current working experience.In addition, the survey contains 15 demographic questions.Table 3 shows the demographic questions related to immigration factors, which are the most critical for our study.
When considering if specific questions belonged to a particular micro-inequity type, we followed our chosen typology definitions (see Section 3) as much as possible.However, the differentiation was sometimes not clear cut.We decided to opt for the verbal category when reactions are mostly spoken.For example, interruptions could be made verbally or through gestures, but they most often occur verbally, thus we classified them as a verbal micro-inequity.We chose nonverbal when the micro-inequities are behavioral and without much planning or support from organizational structure.For instance, in the software industry, menial tasks such as taking minutes or organizing lunch are likely to be given in a spontaneous fashion without having a backlog for its tracking or being formally devised for specific roles-therefore, we categorized it as nonverbal.We categorized questions as environmental when the organizational structure likely supports the concerned micro-inequity.For example, being assigned (non-menial) tasks outside of the official role or not having the authority to make necessary decisions.
To guarantee that we have enough number of responses for our analysis, all questions were mandatory.Besides the demographic questions, all other questions allow for answers that follow the  form of a Likert scale.To ensure clear formulations in all questions, we conducted a trial run with 10 participants.We adapted questions for which we received feedback accordingly.We do not report on the results from the trials in this study.

Survey distribution
We recruited participants from online networking services (LinkedIn, Reddit, and Facebook) and authors' professional networks.Furthermore, we distributed to groups/servers in Telegram and Discord.We used purposive sampling to increase the probability of receiving a balanced number of filled in surveys in terms of immigrant/non-immigrant responses.We received more responses from non-immigrants than immigrants at first.Therefore, we subsequently approached groups or subthreads that have a higher probability of attracting individuals that have an immigration background.For instance, groups that identify as being expats.We used Google forms to distribute our survey.This study reports on responses received between June 29th 2021 to June 30th 2022.

Data Analysis
For our analysis, we split our data into two independent groups: (1) Immigrants, i.e., software practitioners working outside of their home country3 and (2) Non-immigrants i.e., software practitioners working in their home country.Figure 1 shows the strategy we followed to categorize the immigration status of participants according to their responses.
For the data analysis we follow the recommendations by McCrum-Gardner [29].The difference in responses to survey questions is compared over immigrant and non-immigrant software practitioners; we later compare between immigrants (first generation immigrations) and second-generation, which are part of our non-immigrant group.Finally, we compare between immigrants from the Global South and Global North.
Because the survey answers are in the form of Likert scale data, we analyze the data quantitatively through a Wilcoxon rank sum test.The same test is used when testing for differences between first-and second-generation immigrants, and immigrants from the Global South and Global North.
Before applying our tests we inverted the answer scales to [Never,...,Always] and [Strongly Disagree,...,Strongly Agree] and converted them to numerical values in the 1 to 5 range.
We tested one variable, immigration, for each question in our survey two times.First we tested differences between immigrants and non-immigrants.Later, between immigrants (first generation immigrants) and second-generation, which are part of our nonimmigrant group.Finally, between immigrants originating from different global and socio-economic regions.Due to these three comparisons we applied a Benjamini-Hochberg correction on our test results.We consider the results to be statistically significant if the p-value is less than 0.05.
In total, we received 170 responses.A small number of participants (17) reported that they do not work in the software industry in a technical role.Few responses were only partially filled in (16) or did not make sense (2).For example, a respondent stated that they work in their home country, but their answers concerning

RESULTS
We mark all results with statistical significant difference between both groups with an * and only report on the test results of these.Figure 2 shows the statistically significant differences in microinequity experiences of immigrant and non-immigrant software professionals.
Most of our respondents identified themselves as men (82.2%), fewer as women (17%) and one as non-binary (0.8%).The majority of respondents belonged to the age-group of 26-35 (45.2%).In total, participants reported living in 21 different countries, with the largest proportions living in the Netherlands (33.1%) and United States (19.1%).Table 4 shows the geographic distribution of home and work countries of immigrant respondents along different continents.Table 5 shows this grouping along two socio-economic and political global regions: the Global South and Global North [21].Most responses are from software practitioners with a bachelors (55.9%) or master's degree (34.6%).The majority of participants (87.5%) reported working full time (35+ hours per week), mostly in the role of software developers (47.8%).Participants were evenly distributed among companies of different sizes, with two exceptions: (1) a very low number of participants reported working for extremely large (100.000+employees) or (2) very small organizations (less than 10 employees).

Verbal micro-inequities (RQ1)
More than half (63%) of the participants reported never being told by others that they are not smart, competent or articulate.The answers are similarly distributed in the two groups.Around 17% of immigrant participants have very often experienced being told that they are smart, competent, or articulate, after being initially assumed not to be*, compared to 5% of non-immigrants.The difference is significant (Wilcoxon rank-sum test: W = 1668, p-value = 0.038).
While no participants reported always or very often getting a negative reaction about the way they speak* from colleagues, over 40% of the immigrants and 20% of the non-immigrants reported that this sometimes or rarely happens to them.The difference between  both groups is significant (Wilcoxon rank-sum test: W = 1433, pvalue = 0.0018).
Only non-immigrants declared always or sometimes being told that they are too aggressive or bossy*, albeit these were few.The majority of all the respondents (64%) and a large population (> 70%) of non-immigrants reported never being told that they are too aggressive or bossy*.A significantly larger proportion of the immigrant group (54%) reported that this sometimes or rarely happens to them (Wilcoxon rank-sum test: W = 1556.5,p-value = 0.012).
Participants from both groups reported never or rarely receiving negative reactions on their appearance in similar proportions (immigrants 91%, non-immigrants 94%).
More immigrants than non-immigrants reported rarely (32% and 20%, respectively) or sometimes (15% and 6%, respectively) having colleagues raising their voice when talking to them*, even though they had no difficulties understanding them.In a similar line, more nonimmigrants (64%) claimed never experiencing such a behavior from their coworkers than immigrants (47%).The difference between both groups is significant (Wilcoxon rank-sum test: W = 1625, pvalue = 0.0216).More immigrants (43%) than non-immigrants (18%) rarely experienced colleagues talking more slowly to them*, even though they had no difficulties understanding them.This difference is significant (Wilcoxon rank-sum test: W = 1457, p-value = 0.0023).
Immigrants and non-immigrants experienced verbal harrassment in similar proportions.Most immigrants (89%) and non-immigrants (90%) reported never or rarely experiencing verbal harassment when working in the software industry.In contrast, considerable proportion of immigrants (15%) disclosed sometimes having witnessed verbal harassment* when working in the software industry, while relatively few non-immigrants (4%) did.The difference between both groups is significant (Wilcoxon rank-sum test: W=1702, p=0.039).Around half of the participants from both groups (55% non-immigrants, 43% immigrants) declared sometimes being interrupted during discussions or meetings.Other answers are fairly evenly distributed among them.
Most participants declared that they never (47% non-immigrants, 43% immigrants) or rarely (39% for both groups) not get the chance to state their questions, while other colleagues did.More immigrants (30%) answered that sometimes colleagues were not receptive to their ideas, compared to non-immigrants (18%).Similarly, more immigrants (32%) experienced sometimes having their ideas not taken seriously compared to non-immigrants (19%).
Overall 10% of both immigrants and non-immigrants reported colleagues sometimes presenting negative body language towards them.More immigrants (41%) said this issue happened rarely than non-immigrants (31%).More immigrants experienced colleagues taking credit for their work without properly attributing them* both very often (9%) and sometimes (24%), compared to non-immigrants (2% and 16% respectively).The difference between both is significant (Wilcoxon rank-sum test: W = 1676, p-value = 0.038).
Answers were more or less evenly distributed between immigrants and non immigrants when reporting if they have been asked to do menial tasks that other colleagues were not asked to do.More immigrants (31%) experienced coworkers sometimes asking colleagues a question, that due to their experience level, should have been asked to them, compared to non-immigrants (18%).The difference, however, is not significant.

Environmental micro-inequities (RQ3)
A considerable number of respondents reported being assigned tasks with unreasonable deadlines very often (24% immigrants, 14% nonimmigrants) or sometimes (49% immigrants, 26% non-immigrants).Most respondents (immigrants 74%, non-immigrants 79%) are rarely or sometimes assigned tasks that require other types of skills than those entailed in their professional role.
The largest proportion of both immigrants and non-immigrants disagreed that there exists an organizational climate in their work environment that is not inviting to all races, ethnicities, skin colors and (migration) backgrounds*.However, substantially less immigrants (34%) strongly disagreed with the statement compared to non-immigrants (58%).Similarly, a slightly larger proportion of immigrants agreed (8%) with the statement as opposed to nonimmigrants (7%).The difference is significant (Wilcoxon rank-sum test: W = 1589.5,p-value = 0.0216).
Both groups showed even proportions of participants (immigrant 32%, non-immigrant 34%) being responsible for managing or leading others in their work environment.Similarly, both groups reported having the support and authority necessary to make decisions in similar proportions.A larger proportion of immigrants (63%) agreed with the statement compared to non-immigrants (55%).Finally, most participants from both immigrant (76%) and non-immigrant (85%) groups felt that their team values and acknowledges their contributions.

Micro-inequities among first-and second-generation immigrants (RQ4)
Among non-immigrant respondents, 20 (15%) are secondgeneration immigrants and all of the immigrant participants, 46 (34%), are first-generation immigrants, (see Table 3 and Figure 1).We detail the micro-inequities where there was a one-sided significant difference between first-and second-generation immigrant software practitioners (see Figure 3) as follows.
First-generation immigrants (50%) reported getting a negative reaction to the way they speak* from colleagues more often than second-generation immigrants (15%).The difference is significant (Wilcoxon rank-sum test: W = 444.5,p-value = 0.016).Firstgeneration immigrants also experienced being told that they are aggressive or bossy* to a significant higher degree (56.5%) than second-generation immigrants (20%), (Wilcoxon rank-sum test: W = 445, p-value = 0.016).
Significantly more first-generation immigrants (74%) experienced colleagues taking credit for their work without properly attributing them* than second-generation immigrants (45%), (Wilcoxon rank-sum test: W = 438.5,p-value = 0.017).Similarly, more first-generation immigrants (32.6%) reported a colleague being asked a question by a coworker, that due to their experience, should have been asked to them*, than second generation immigrants (10%).The difference is significant (Wilcoxon rank-sum test: W = 443, p-value = 0.017).
In contrast with the previous cases, significantly more secondgeneration respondents (65%) reported being assigned tasks that are not entailed in their professional role* than first-generation immigrants (39%), (Wilcoxon rank-sum test: W = 238, p-value = 0.036).

Micro-inequities between immigrants from different global regions (RQ5)
Among the 46 immigrant participants, nearly 46% were immigrants from the Global North (GN) working in the Global South (GS), the same proportion were immigrants from the Global North working in the same global region; 8% were immigrants from the Global South working in the Global South (see Table 5).Due to the sparsity of the latter, we only compared the first two groups.With the exception of three micro-inequities we found that survey participants experienced micro-inequities in similar manner.We detail those where there was a significant difference as follows.More GS->GN immigrants (14%) experienced significantly more very often being excluded from key social/networking opportunities at work* than GN->GN immigrants (0%) (Wilcoxon rank-sum test: W = 123.5,p-value = 0.007).A larger proportion of GN->GN immigrants (90%)(strongly) disagreed that their work environment is less inviting to coworkers with diverse backgrounds than GS->GN immigrants (67%).The difference is significant (Wilcoxon rank-sum test: W = 130, p-value = 0.007).Furthermore, significantly more GN->GN immigrants (100%) (strongly) agreed that their team acknowledges or values their contributions than GS->GN immigrants (80%) (Wilcoxon rank-sum test: W = 307, p-value = 0.007).

DISCUSSION
Our results show that experiences of micro-inequities between immigrant and non-immigrant software practitioners differ.Of the three types of micro-inequities we investigated, all include microinequities which immigrants experienced significantly more than non-immigrants.In contrast, with one exception, we found no micro-inequities which non-immigrants experienced significantly more than immigrants.
Verbal micro-inequities are the most prevalent (RQ1).Immigrants experienced six of them significantly more than nonimmigrants: being assumed to be less intelligent, competent or articulate, getting negative reactions to their speech, being told they are too aggressive or bossy, having colleagues raise their voice at them and talking more slowly to them even though they had no difficulties understanding their colleagues, and witnessing verbal harassment.These findings are generally inline with those of existing studies, though in different work fields such as libraries and health care, e.g., [2,13,20,46].A previous study [16] analyzing micro-inequities in the software industry through a gendered perspective found that women experienced significantly more often being told that they are bossy or aggressive, and significantly more harassment than men.They did not examine the other verbal micro-inequities we considered in our study.
Only one out of the ten investigated nonverbal micro-inequities (RQ2)-colleagues taking credit for the respondents' work without proper attribution-had a statistical significant difference between immigrants and non-immigrants.Not being given credit can have a negative emotional impact on the affected individual and ethical consequences for the organization [15,28,35].Future research should analyze the context in which this micro-inequity occurs and further investigate how it is experienced by both marginalized and privileged groups in the software industry.
Our results regarding nonverbal micro-inequities oppose those of previous work [2] which show that racial minority groups experience numerous nonverbal microaggressions to a significantly higher degree than non-minority groups.These nonverbal microaggressions include opinions being overlooked in discussions and colleagues having a negative body language towards them, both because of race.A possible explanation for this disparity could be the differing contexts in which both studies were performed (libraries vs software industry).Further, our study was conducted during the corona pandemic, when many software practitioners were working online.The increased usage of digital work environments probably made nonverbal micro-inequities less obvious altogether in the software industry (and other industries that require less face to face contact) during this time.Additionally, while the previous work analyzed race as a differing factor, we analyzed immigration background.While sometimes related, these variables are in fact different.
With respect to environmental micro-inequities (RQ3), our results show that immigrant software practitioners feel significantly more excluded from key social or networking opportunities than non-immigrants and that significantly more immigrants agree that their workplace is not inviting to all races, ethnicities, and immigration backgrounds.These findings are inline with previous work analyzing ethnic-and gender-based bias in tech [41], where respondents reported experiencing ethnic-based typecasting, hitting glass ceilings, or being excluded from key networks due to ethnicity.Although our study addresses only a part of ethnic-based biasnamely (subtle) bias on the basis of being an immigrant or not-our results also show significant differences in the feeling of exclusion from social events and the work environment.This finding is supported by previous studies which found that underrepresented groups, specifically immigrants working in tech, lack the feeling of inclusion that non-minority groups (such as non-immigrants), tend to have at the workplace [6,14,33,41].When analyzing differences in micro-inequities experienced by first-and second-generation immigrants (RQ4), we found that firstgeneration immigrants reported getting a negative reaction to the way they speak from colleagues, being told they are aggressive or bossy, having colleagues take credit for their work without proper attribution, and experiencing being skipped on questions that due to their experience, should have been asked to them significantly more than second-generation immigrants.Three of the four abovementioned micro-inequities prevail from the comparison of immigrants and non-immigrants.As micro-inequities, by their very nature, occur when people are perceived to be different [18], a possible explanation for these differences is that second-generation immigrants are perceived as less different by their coworkers than first-generation immigrants.We stress that while we used the term second-generation immigrant in this study, this group is in fact non-immigrant.
More second-generation respondents experienced being assigned tasks that are not entailed in their professional role.This might be due to the fact that expats, who are first-generation immigrants, usually get hired to do specific high-skilled tasks in the software industry.There is, to our best knowledge, no research on micro-inequities experienced by first-and second-generation immigrants specifically.Future studies should be conducted on the topic, as it could offer insights on whether the experience of microinequities by people with immigration background improves over generations.
Respondents belonging to countries from the Global South (RQ5) experienced significantly more environmental micro-inequities, such as exclusion from key social and networking opportunities, not finding the work environment inclusive for diverse demographics and lack of recognition for their contributions from colleagues.The perceived competence and likability of immigrant groups are significantly influenced by their national origin [25].Previous research [8,34] shows that immigrants originating from nations in the Global South are more susceptible to encountering discriminatory behavior in both formal settings (contractual situations) and informal interactions (social relations) within the workplace.Potential factors for the absence of statistically significant findings related to verbal and nonverbal micro-inequities in the particular context of the Global South and Global North, but also in the rest of our analysis, can be attributed to the limited size of our dataset comprising only 21 immigrants from both the Global North and Global South, and the survey being conducted during COVID times when a substantial portion of communication occurred in an online format, rendering such micro-inequities less noticeable.
The aforementioned impact of immigrants' origin also explains our finding that although a majority of our immigrant respondents (37%) have relocated within European countries; they still reported experiencing various micro-inequities.A plausible reason for this may lie in the notable disparities encompassing political, cultural and socio-economic variations that persist across European countries, despite their shared geographical region.
Diversity and inclusion are interrelated concepts, yet it is important to note that while diversity can be legislated and measured, inclusion cannot.More extensive awareness of the environmental manifestations of micro-inequities among practitioners in the field are needed as micro-inequities at the workplace can cumulatively convey messages about who does and does not belong at the workplace [6].This in turn can affect employees' intellectual performance, motivation level, and self-concept [26]; eventually leading to leakage pipes for marginalized groups and a lack of diversity in leadership positions.
Call for action: Our results provide evidence that software practitioners are aware of, and when given a chance open up about the micro-inequities that they experience or witness in their professional life.Micro-inequities can cause feelings of discouragement and impair performance in the workplace [49] and are usually caused by unconscious biases [4].It is therefore crucial to create more awareness in the community about the commonality of their occurrence.Software companies should provide safe, accessible and credible mechanisms for underrepresented groups to share their experiences regarding micro-inequities.Awareness about subtle discrimination can be raised through company newsletters, staff meetings and training programs [39].While these actions are not enough to solve the problem they can be catalysts for inner-questioning and conversations among colleagues.The current situation on the work floor regarding micro-inequities could be assessed through occasional questionnaire-based surveys.For this purpose, our survey instrument can be helpful as it has been designed taking the software industry into consideration.
Researchers should further investigate the organizational factors that contribute to micro-inequities and the intersectional impact of immigration status with other variables such as race, gender and age.Comparative studies regarding the perception and experiences of management and other employees regarding micro-inequities will help determine the relevant impact of role and seniority.For immigrant software practitioners' who experience micro-inequities, though difficult, direct and open communication with the instigator can be the first step to address micro-inequities due to the possibility of these actions being done unintentionally.Finally, all software practitioners should consciously perform small, symbolic actions-micro-inclusions-to include colleagues who are excluded or othered.

LIMITATIONS
Immigrant participants were particularly difficult to reach.Thus, we settled for a 1:2 ratio (34% immigrants, 66% non-immigrants) due to time constraints.A more even distribution of respondents would have been preferable.A similar limitation lies in the analysis of the differences between first-(34%) and second-generation (15%) immigrants.While the amount of responses from secondgeneration immigrants is still above the recommended minimum sample size (within groups) for performing the Wilcoxon rank sum test (n>4) [11], an even greater number of responses could increase the power of our conclusions concerning the differences in experiences between first-and second-generation immigrants.The latter also applies for our Global South and Global North analysis.
Personality differences also contribute to the likelihood of perceiving/provoking discriminatory behavior [31].We did not analyze this variable in our work.In the same line, besides immigration background, there might be other variables that influence the analyzed experiences and perceptions of our participants, including age, gender, country of residence, race, ethnicity, job role, and seniority.Along this line, we stress that while we treated both immigrant and non-immigrant groups as homogeneous entities in our study both groups entail diverse sets of people which are shaped by different aspects forming their identity.
The generalizability of our findings is restricted by the predominant representation of respondents from specific geographical locales, such as the Netherlands and the United States.Also, we did not study the impact of immigrants' home country, work location and the number of years spent in their work country.A potential future study could investigate whether these factors affect immigrants experience of micro-inequities in the software industry; as well as the affect of immigration time.
For our consideration of immigrants and non-immigrants, we allowed respondents to self-define their home country (i.e., the country participants felt the strongest bond to 4 ).This was a determining factor for characterizing respondents as immigrants and non-immigrants.In this characterization, respondents implicitly had the freedom to choose a country different than the one they were born in.This decision was motivated by the authors' own observations as first-generation immigrants 5 and their experiences with bonds to the country in which they were born, the one(s) they grew up in, the country they are currently working in, and the ones they are officially citizens of.
Our study centers on 27 instances of micro-inequities, a scope that does not encompass the completeness of possible manifestations within the workplace.Furthermore, the categorization of most micro-inequities as verbal, non-verbal or environmental was straightforward for most of the questions in our survey.However, there were few cases that were not clear cut.For instance, negative reactions to personal appearance or interruptions during meetings could be done both verbally and non-verbally.We classified these by taking into account the manner in which we thought those messages were most often delivered (see Section 2).We emphasize that regardless of which specific category the micro-inequity belongs to, it is important to note that these exist and, as our and other research suggests, affects underrepresented groups more.
Ethnic bias and micro-inequities can be controversial topics, and as a consequence self-selection bias is a validity threat to our study.This could result in people with strong opinions on the subject responding the survey more often, potentially skewing the results.When distributing, we mitigated this risk by using neutral titles and description texts, containing neutral words like "teamwork" and "challenges" instead of "micro-inequities".
Our sample consists of 135 answers from respondents.This is a drawback in regards to the accuracy of the statistical analysis.We tried to mitigate this by only considering one demographic variable instead of all collected.We advertised the survey through social media and personal and professional networks, therefore, we could not compute the response rate.We also cannot claim that our participants are representative of the whole software engineering industry.Another limitation is the distribution of the survey during the pandemic when countries were in (partial) lockdown.People working remotely are less likely to experience subtle inequities, such as negative reactions to appearance, being asked to do menial tasks (e.g., order food), or colleagues showing negative gestures, as there is no shared physical workspace with colleagues.This offers an explanation for why we found only one out of ten forms of nonverbal micro-inequities differing with statistical significance.
With our strategy for classifying immigrants and nonimmigrants, participants working remotely from their home country for a company located abroad were incorrectly typed as immigrants.Working remotely from abroad could be a moderating variable between experiencing micro-inequities and immigration background.However, the threat works both ways: both non-immigrants and immigrants could be working remotely from abroad.We anticipated this threat, and therefore explicitly asked in our questionnaire if participants work remotely from abroad.A relatively small number of participants reported they were (20 of 135).A future study could be done to investigate if working remotely from abroad and experiencing micro-inequities are significantly related.

RELATED WORK
Existing work, discussed below, shows that people working in tech with an immigration background are more likely to experience inequities which could be closely related to micro-inequities.
Mor-Barak and Levin [32] surveyed 3,400 employees with diverse backgrounds in the high-tech industry in Southern California and found that ethnic minorities, among other underrepresented groups such as gender minorities, are more likely to feel excluded from organizational information networks and important decisionmaking processes.Amrute [3] explored the relationship between Seattle's tech industry, migration, and hate incidents.Despite Seattle's high ethnic diversity ranking, they found that immigrants from India and China working in Seattle's software industry experience racially motivated harassment in local communities.
Pugalia and Kozanioglu [36] identified several barriers that serve as a rationale for the low participation of immigrant women entrepreneurs in the high tech sector.These barriers include racial discrimination, different behavioral rules, and a limitation of independence and mobility due to cultural factors.
Research surveying the perceptions of 2,686 employees working in tech, revealed that racial and ethnic minority men and women saw a higher value in diversity and felt more comfortable with diversity than their Caucasian male co-workers [32].Two-thirds of respondents from 54 semi-structured interviews of high-skilled, white and Asian men and women working in Silicon Valley, reported that they have experienced some form of ethnic or gender bias.Experiences include ethnic typecasting, job segregation and glass ceilings, as well as being excluded from key "old boys" networks [41].
While research shows the existence of ethnic or race related microaggressions and gendered micro-inequities in different work fields, e.g., [6,12,14,19,27,37,40,42,45], research on the topic in the software industry is sparse.To our best knowledge, there is only one study [16] that has analyzed micro-inequities in the software industry.This work analyzed seven micro-inequities, as well as other prominent barriers and sources of unfair treatment in the software industry through a gender-perspective using a different dataset.Besides the distinct perspectives, our study distinguishes itself from this work by focusing exclusively on micro-inequities, further distinguishing between verbal, nonverbal and environmental.While the aforementioned study analyzed seven micro-inequities, we analyzed 27.We believe that our study could also be extended so that a gender-perspective is accounted for.To our best knowledge, there have been no studies done on the occurrence of micro-inequities among software practitioners with an immigration background focus specifically.Our work provides initial evidence on this topic by surveying software practitioners and investigating micro-inequities and their relation to immigrants in the software industry.

CONCLUSION
We report on the results of a survey of 135 immigrant and nonimmigrant software practitioners working in technical roles.We analyzed their experiences in the software industry with respect to verbal, nonverbal, and environmental manifestations of microinequities.Our results show that the experiences of immigrants and non-immigrants differ with statistical significance in nine of the 27 investigated forms of micro-inequities.The largest number of differences were found in the experience of verbal micro-inequities.Immigrants significantly reported experiencing them more often than non-immigrants in six out of nine investigated verbal forms of micro-inequities and three forms of verbal and environmental micro-inequities.With the exception of three micro-inequities, we found that immigrants from the Global South and Global North experienced micro-inequities in a similar manner.We also found that first-generation immigrants experience more micro-inequities, particularly verbal and nonverbal, compared to second-generation immigrants.
Our results offer practitioners in the field an added incentive, alongside related literature, to promote work practices that encourage inclusion in the workplace.Moreover, our study can be used by other researchers to take stock of the current situation with regards to equity and the everyday workplace experiences of immigrant and non-immigrant software practitioners.

Figure 1 :
Figure 1: Strategy for categorizing immigration status

Figure 3 :
Figure 3: Significant differences regarding micro-inequities experienced by first-and second-generation immigrants working as software practitioners.
Q25 I have felt "out of place" or excluded at my workplace.[Always;Very often; Sometimes; Rarely; Never] [43] Q26 I have felt excluded from key social/networking opportunities at work.[Always; Very often; Sometimes; Rarely; Never] [47] Q27 In my work environment, I believe there exists an organizational climate that is not inviting to coworkers of all races, ethnicities, skin colors, and (migration) backgrounds.[S.agree; Agree; Neutral; Disagree; S. disagree] Own work Q28 I am responsible for managing or leading others in my work environment.[S.agree; Agree; Neutral; Disagree; S. disagree] [22] Q29 I have the support and authority necessary to make decisions.[S.agree; Agree; Neutral; Disagree; S. disagree] [22] Q30 I feel that my team values and acknowledges my contributions.[S.agree; Agree; Neutral; Disagree; S. disagree] [22]

Table 2 :
Demographic questions related to technical role and immigration background.
#Question Response OptionsQ1Are you currently working in the software industry in a technical role (e.g.software developer, software architect, software consultant, product owner, project manager, tester, data scientist, or data engineer)?

Table 4 :
Overview of immigrants home and work continents

Table 5 :
Immigrants' home and work socio-economic and political regions