Social-economics, community, campus and family: a nationwide empirical investigation on the association between adverse childhood experiences and early career choice of youths and adolescents

ABSTRACT

This study examines the influences of childhood experiences (childhood social-economic situation, childhood community and campus quality, childhood family climate) on early career choice. Using the data of a large-scale nationwide stratified random sampling survey of life history across 28 provinces and municipalities in China, empirical results show that the lower quality of childhood experiences would increase the likelihood that youths choose to work in the public sector after adulthood. Also, male youths are generally more sensitive to the childhood experiences than female ones. Youths with migration experiences are generally more sensitive to the childhood family climate (and less sensitive to the childhood community/campus quality) than those that never migrate. Moreover, such influence is distinctly reflected in the choice of the first job, and gradually recedes into insignificance for the choice of the second and third job.

KEYWORDS:

• Child development
• adverse childhood experiences
• social-economic status
• family climate
• community and campus quality
• early career choice

Introduction

Adverse childhood experiences (ACEs) have attracted increasing attention of studies regarding youths and adolescents’ behavioural and psychosocial consequences (e.g. addictive behaviour, aggression and depression, Matsuura, Hashimoto, & Toichi, 2013; Ramiro, Madrid, & Brown, 2010). The ACEs include but not limited to stressful status or events such as household dysfunction, interpersonal violence, childhood neglect or abuse. While it is widely admitted that the ACEs have strong impact on youths and adolescents’ behaviour tendency, the impact on early career choice outcome remains largely unexamined. According to the perspective of social cognitive career theory (SCCT) (Lent, Brown, & Hackett, 1994), the dynamic social-contextual factors during the growth (e.g. family, community, and campus conditions) affect the learning experiences through which career-relevant self-efficacy and outcome expectations are forged. As such, the experiences in early life (e.g. emotional neglect and abuse, family violence, parental mental health and drug/alcohol abuse) can impose implicit effect on people’s career choice after adulthood.

Figure 2. The conditional marginal effects of neighbourhood friendship on the probability of choosing public sector as the 1st job choice (according to logit model regression)

Figure 3. The conditional marginal effects of community bullying on the probability of choosing public sector as the 1st job choice (according to logit model regression)

Figure 4. The conditional marginal effects of campus safety concern on the probability of choosing public sector as the 1st job choice (according to logit model regression)

Figure 5. The conditional marginal effects of feeling lonely for no friends on the probability of choosing public sector as the 1st job choice (according to logit model regression)

Figure 6. The conditional marginal effects of relationship between parents on the probability of choosing public sector as the 1st job choice (according to logit model regression)

Figure 7. The conditional marginal effects of parental panic and feeing upset on the probability of choosing public sector as the 1st job choice (according to logit model regression)

As the most proximal social contextual factors that are ambient in one’s world during childhood come from his or her family of origin, community and campus, childhood family, community and campus experiences should be emphasized when we explore determinants driving youths and adolescents to make early career choice after adulthood. Despite having the theoretical importance, the influences of these contextual factors on youths and adolescents’ early career choice after adulthood have not been fully analysed (Hargrove, Creagh, & Burgess, 2002; Whiston & Keller, 2004). Some indirect signs show that people’s career choice can be associated with the childhood family climate and economic condition (Schröder, Schmitt-Rodermund, & Arnaud, 2011; Tang, Fouad, & Smith, 1999; Zellweger, Sieger, & Halter, 2011). People who have positive childhood experiences in such aspects can feel the confidence to cope with career challenges after their adulthood (Zellweger et al., 2011). They tend to have risk preference in their career pursuit (Whiston & Keller, 2004), and thus are less likely to put emphasis on job rewards of the public sector (e.g. job security, stable benefit, seniority-based promotion, appropriate workload and task significance; Chen & Bozeman, 2013; Chen & Hsieh, 2015; Lee & Choi, 2016). The risk preference, confidence to deal with challenges, and willingness to sacrifice job security for more career potentials would reduce the likelihood of choosing the career in the public sector. In contrast, those who are lack of the sense of security and are risk-averse are more likely to choose the public sector accordingly (Clark, 2016; Dong, 2015, 2017; Van de Walle, Steijn, & Jilke, 2015).

This study contributes to the existing literature in several aspects. First, this paper makes a preliminary exploration of the relationship between childhood experiences of youths and adolescents and their early career choice after adulthood, where empirical evidence is still absent for previous studies. Second, this paper, respectively, estimates and compares above impact by gender difference and migration experiences difference. Third, this paper evaluates the persistency of the influence of childhood experiences of youths and adolescents on career choice by estimating such impact from the 1st job to the 3rd. Fourth, the special focus on early career choice that is important for youths and adolescents’ future career trajectory is still rare in existing literature (Goldacre, Fazel, Smith, & Lambert, 2013). This study will fill this gap to some extent.

Literature review

The perspective of social cognitive career theory (SCCT)

SCCT describes the process through which people make choices in occupational pursuits (Lent et al., 1994). Anchored in general social cognitive theory developed by Bandura (1986), SCCT makes efforts to help understand how external environment (e.g. social supports and barriers, like economic conditions, parental behaviours and peer influences) shapes one’s course of career development (Lent, Brown, & Hackett, 1996, 2000; Lent et al., 1999, 1998).

According to SCCT, career choice is influenced both by distal and proximal environmental factors. Based on the relative proximity to the career choice-making process, environmental influences are divided into these two basic categories. The first, distal category of environmental influences includes background contextual factors which can shape the development of career-relevant self-efficacy and outcome expectations (e.g. ‘if I choose this career field, what will happen?’) through affecting personal learning experiences. For example, the support or discouragement one obtains from engaging in particular activities outside of his or her career. The second, proximal, category refers to those contextual factors during active phases of career decision-making, such as career network contacts or exposure to discriminatory hiring practices. This standard of division also considers the temporal dimension. Contextual factors encountered in the past and those in the present correspond to these two categories, respectively, based on their relative proximity to the career choice-making process.

Childhood experiences including childhood social-economic situation, childhood community and campus quality, and childhood family climate can be viewed as a kind of choice-distal factors that precede and help shape personal interests, self-cognitions, or other influences on career choice through affecting personal learning experiences. People can take in values, attitudes or regulatory structure from the interaction process with their family, community, and campus during childhood (learning experiences), and internalize those as their own values, attitudes or regulatory structure (Gagné & Deci, 2005; Houston, 2011; Pedersen, 2015). ACEs (poorer social-economic situation, lower community and campus quality, and worse family climate during childhood) can diminish one’s coping efficacy and enhance the degree of the need for security, stability, or control (Coleman, 2003; Davies & Woitach, 2008; Gomez, 2014; Hackett & Betz, 1981; Pan, Sun, & Chow, 2011; Sabri, Coohey, & Campbell, 2012). In other words, the low level of coping efficacy and high degree of the need for security, stability, or control can gradually forge through learning from ACEs. Such cognition and needs may further have an implicit effect on the expectations for further career, such that those with low-level of risks and challenges are more likely to be the preferred career choices. In contrast, positive childhood experiences may promote the feelings of competence and risk tolerance (Whiston & Keller, 2004; Zellweger et al., 2011), whereby people may be more willing to pursue the career with relatively higher risks and challenges.

Indirect evidence about the relation of childhood experiences and career choice

The career choice of working in the public or private sector brings people distinct life in China. Higher job security (Hui, Lee, & Rousseau, 2004; Walsh & Zhu, 2007), lower work-life conflict and stable payment (Blecher, 2002), seniority-based career promotion (Cooke, 2000; Lewis, 2003), are all among the job rewards attracting talents to the public sector (Perry & Hondeghem, 2008). In contrast, the private sector is featured with performance-orientated career promotion and wage policy (Wei & Rowley, 2009), and known to have more unstable and differentiated payment, and higher risk of layoff.

The degree to which people put emphasis on the above job rewards of working in the public sector is subject to social contextual factors in their childhood. Indirect evidence presented by recent research suggests that childhood family climate can affect individual career choice as it shapes one’s capability to cope with adverse life experiences (Davies, Forman, Rasi, & Stevens, 2002; Kouros, Merrilees, & Cummings, 2008; Schudlich & Cummings, 2007). Individuals who experience poor family climate in childhood are found to be associated with the lack of career efficacy (Coleman, 2003; Davies & Woitach, 2008; Hackett & Betz, 1981; Pan et al., 2011). They may perceive it hard to work under pressure (Mann & Gilliom, 2004), and incline to avoid risks (Dulebohn, 2002). They are more willing to choose positions with the fixed payment and stable career development path (Turban, Lau, Ngo, Chow, & Si, 2001).

Not limited within the household, the influences of childhood experiences are also extended to the community and campus. The intra-community discord and interpersonal violence reduce the quality of individual perception of safety and security (Cummings et al., 2011; Lovell & Cummings, 2001). The exposure to the community discord and violence can frustrate individual risk-taking intention and behaviour in adulthood (Margolin & Gordis, 2000). The adverse community and campus experiences would impair individual confidence in pursuit of the venturous career, and increase individual emphasis on stable career development (Sabri et al., 2012). As such, the adverse community and campus situation can make people care more about the job rewards of working in the public sector.

More recently, the effect of childhood economic situation on individual career efficacy has been examined. The economic disadvantage hinders people’s access to resources that are helpful for early career exploration and constrains their perceived confidence in career development in adulthood (Gomez, 2014). Former adverse experiences of financial stress would also curtail people’s belief in their ability to perform tasks in career pursuit (Vannatter, 2012). Family economic distress in childhood can make people more easily suffer from poor emotional health (Menasco, 2012; Toomey & Christie, 1990), and more incline to avoid risks or unstable choices (Maner et al., 2007; Raghunathan & Pham, 1999; Wray & Stone, 2005), and thus they may put more emphasis on job security. Based on the indirect evidence above, it seems to be suggested that ACEs may have a higher likelihood of driving people to choose their career in the public sector given their tendency to make a risk-averse career choice.

Method and materials

Data source

The data come from China Health and Retirement Longitudinal Study – The life history survey (CHARLS-2014), which is collected by stratified random sampling. This widely used survey collects a high quality nationally representative sample of Chinese residents over the age of 45, and aims to serve the needs of scientific research relevant to health for the mid-aged and elderly (Zhao, Hu, Smith, Strauss, & Yang, 2012). Respondents of this survey are followed up every two years and the collected data are made public one year after the data collection has been completed. Hence, this survey is available to the public since 2016. This life history survey is implemented across 28 provinces and municipalities in China. It covers 150 counties including 450 communities and villages among which about 52.6% are rural areas and 47.4% are urban areas.

Measurement

Dependent variable

Table 1 shows the details about measurements of variables. The binary dependent variable of this study is the type of job. Respondents that work in governments, public institutions, NGOs or rural collective economic organizations are public sector employees, while those that work in firms, individual or self-employed firms, or as farmers, individual households (i.e. self-employed small business) or others are private sector employees (more details about dependent variables have been shown in Table 2).

Table 1. The overview of variables (measurement)

Variables	Measure
The 1st, 2nd, 3rd job choice	=1 (risk averse choice) when respondents choose to work in governments, public institutions, NGOs or rural collective economic organizations, =0 (risk preferred choice) when respondents choose to work in firms, individual or self-employed firms, or as farmers, individual households, and others.
Childhood social-economic situation
Relative family financial status	When you were a child before age 17, compared to the average family in the same community/village at that time, how was your family’s financial situation? (1 = A lot better off than them, 5 = A lot worse off than them)
Childhood community/campus quality
Neighbourhood friendship	Were the neighbours of the place where you lived as a child very close-knit? Is it very close-knit, somewhat close-knit, not very close-knit or not close-knit at all? (1 = Very close-knit, 4 = Not close-knit at all)
Community bullying	When you were a child, how often were you picked on or bullied by kids in your neighbourhood? Is it often, sometimes, rarely or never? (1 = Often, 4 = Never)
Campus safety concern	When you were a child, how often did you feel worried about your physical safety at school? Is it often, sometimes, rarely or never? (1 = Never, 4 = Often)
Feeling lonely for no friends	When you were a child, how often did you feel lonely for not having friends? (1 = Never, 4 = Often)
Childhood family climate
Relationship between parents	How would you rate the relationship your parents had with each other when you were growing up? (1 = Excellent, 6 = Very poor)
Parental panic and feeling upset	During your childhood did your female/male guardian get upset easily or feel panicky? (1 = A little of the time, 4 = A lot of the time, evaluate female and male guardian, respectively, and take the average)
Parental depression and sadness	During the years you were growing up, had your female/male guardian showed continued signs of sadness or depression that lasted 2 weeks or more? (1 = No, 2 = Yes, evaluate female and male guardian, respectively, and take the average)
Parental anxiety and nervousness	During your childhood did your female/male guardian often feel nervous and anxious? (1 = A little of the time, 4 = A lot of the time, evaluate female and male guardian, respectively, and take the average)
Demographics
Gender	[1 = Male, 2 = Female]
Education	[……, 10 = Master program, 11 = Ph.D program]
Membership of CPC	Are you or is your spouse/partner a Chinese Communist Party (CPC) member?
Ethnic minority	Are you or is your spouse/partner a minority? (belongs to certain nationality other than the Han)
Religion	Do you or does your spouse/partner believes in religion?

Table 2. The overview of dependent variables (frequency and percentage)

Variables	Sector	Details	Frequency	Percentage
The 1st job choice	Public sector		12,101	81.29%
		Government	351	2.36%
		Rural collective economic organization	11,020	74.02%
		Public institution	715	4.80%
		NGO	15	0.10%
	Private sector		2789	18.71%
		Firms	2006	13.47%
		Individual firms	482	3.24%
		Farmers	98	0.66%
		Individual household (i.e. self-employed small business)	33	0.22%
		Others	167	1.12%
The 2nd job choice	Public sector		2742	42.80%
		Government	466	7.27%
		Rural collective economic organization	1716	26.79%
		Public institution	538	8.40%
		NGO	22	0.34%
	Private sector		3664	57.20%
		Firms	2329	36.36%
		Individual firms	977	15.25%
		Farmers	102	1.59%
		Individual household (i.e. self-employed small business)	62	0.97%
		Others	194	3.03%
The 3rd job choice	Public sector		1289	28.26%
		Government	358	7.85%
		Rural collective economic organization	558	12.23%
		Public institution	356	7.81%
		NGO	17	0.37%
	Private sector		3272	71.74%
		Firms	1733	38.00%
		Individual firms	1141	25.02%
		Farmers	137	3.00%
		Individual household (i.e. self-employed small business)	73	1.60%
		Others	188	4.12%

About the 1st job choice (Job 1), there are 14,887 answers in total, 12,101 (81.29%) are ‘public sector’, 2786 (18.71%) are ‘private sector’. There are 6406 answers in total reporting the 2^nd job choice (Job 2), 2742 (42.80%) are ‘public sector’, 3664 (57.20%) are ‘private sector’. Besides, there are 4561 answers in total reporting the 3^rd job choice (Job 3), 1289 (28.26%) are ‘public sector’, 3272 (71.74%) are ‘private sector’.

Independent variable

Independent variables in this study involve three categories of childhood experiences, (1) childhood social-economic situation, (2) childhood community/campus quality, (3) childhood family climate. More details have been shown in Table 3.

Table 3. The overview of independent and control variables (frequency and percentage)

Variables	Item	Frequency	Percentage
Childhood social-economic situation
Relative family financial status	A lot better off than them	250	1.23%
	Somewhat better off than them	1748	8.62%
	Same as them	10,317	50.89%
	Somewhat worse off than them	3221	15.89%
	A lot worse off than them	4736	23.36%
Childhood community/campus quality
Neighbourhood friendship	Very close-knit	8860	44.57%
	Somewhat close-knit	10,111	50.87%
	Not very close-knit	694	3.49%
	Not close-nit at all	212	1.07%
Community bullying	Often	765	3.86%
	Sometimes	1886	9.52%
	Not very often	2733	13.80%
	Never	14,426	72.82%
Campus safety concern	Often	442	2.81%
	Sometimes	482	3.06%
	Not very often	755	4.80%
	Never	14,051	89.33%
Feeling lonely for no friends	Often	1193	6.03%
	Sometimes	1273	12.47%
	Not very often	1651	8.35%
	Never	15,663	79.19%
Childhood family climate
Relationship between parents	Excellent	4267	23.10%
	Very good	5990	32.50%
	Good	3370	18.29%
	Fair	4226	22.93%
	Poor	382	2.07%
	Very poor	204	1.11%
Parental panic and feeling upset	1	12,187	70.72%
(1 = a little of the time, 4 = a lot of the time, evaluate male and female guardian, respectively, and take the average)	1.5	1049	6.09%
	2	1984	11.51%
	2.5	510	2.96%
	3	843	4.89%
	3.5	95	0.55%
	4	564	3.27%
Parental depression and sadness	1	13,601	79.99%
(1 = No, 2 = Yes, evaluate female and male guardian, respectively, and take the average)	1.5	1811	10.65%
	2	1591	9.36%
Parental anxiety and nervousness	1	11,059	63.26%
(1 = a little of the time, 4 = a lot of the time, evaluate male and female guardian, respectively, and take the average)	1.5	1321	7.56%
	2	2298	13.14%
	2.5	652	3.73%
	3	1150	6.58%
	3.5	176	1.01%
	4	826	4.72%
Demographic information
Gender	Male	9706	47.46%
	Female	10,743	52.54%
Education	Ph.D.	5	0.02%
	Master	15	0.07%
	Four-year college	188	0.92%
	Three-year college	387	1.89%
	Vocational school	598	2.93%
	High school	1639	8.02%
	Middle school	5271	25.79%
	Primary school and below	12,337	60.36%
Membership of CPC	Yes	3732	18.27%
	No	16,690	81.73%
Ethnic minority	Yes	2013	9.86%
	No	18,399	90.14%
Religion	Yes	2691	13.18%
	No	17,727	86.82%

The independent variable ‘relative family financial status’ belongs to ‘childhood social-economic situation’. The financial status of majority respondents (50.89%) is near the average level of the community or village. About 9.85% and 39.25% are, respectively, above and below the average level.

The independent variables ‘neighbourhood friendship’, ‘community bullying’, ‘campus safety concern’ and ‘feeling lonely for no friends’ belong to ‘childhood community/campus quality’. For ‘neighbourhood friendship’, majority answers (50.87%) are somewhat close-knit. For ‘community bullying’, about 72.82% respondents never suffer from community bullying. For ‘campus safety concern’, the majority of respondents (89.33%) never have such concern. For ‘feeling lonely for no friends’, about 79.19% respondents never have such feeling.

The independent variables ‘relationship between parents’, ‘parental panic and feeling upset’, ‘parental depression and sadness’ and ‘parental anxiety and nervousness’ belong to ‘childhood family climate’. For ‘relationship between parents’, about 73.89% answers are good and above, and the remaining 26.11% are fair and below. For ‘parental panic and feeling upset’, 70.72% (3.27%) respondents score the scale at the lowest (highest) level. For ‘parental depression and sadness’, 79.99% (9.36%) respondents score the scale at the lowest (highest) level. Finally, for ‘parental anxiety and nervousness’, 63.26% (4.72%) respondents score the scale at the lowest (highest) level.

Control variable

Following previous literature, we also control the effects of a series of control variables. More details have been shown in Table 3.

Gender difference is controlled in this study. Previous studies show that males display higher career aspiration, more entrepreneurship orientation and risk preference compared with females, and such gender differences affect career choice (Wigfield, Battle, Keller, & Eccles, 2002; Wilson, Kickul, & Marlino, 2007). In this study, 9706 (47.46%) respondents are males and 10,743 (52.54%) respondents are females.

Education is controlled as previous studies show that lower education is associated with a greater preference for public sector employment which is viewed as a safe career option (Block, Fisch, Lau, Obschonka, & Presse, 2016; Van de Walle et al., 2015). In this study, 2.90% respondents are college educated and above, 2.93% respondents are vocational school level educated, 33.81% respondents are middle school level educated, 60.36% respondents are primary school level educated and below.

Membership of CPC is controlled as previous studies show that the membership of Chinese Communist Party (CPC) is viewed as a kind of political social capital bringing wage and other rewards and thus serves as an incentive for CPC members to enter the public sector (Appleton, Knight, Song, & Xia, 2009). In this study, 18.27% respondents have CPC membership and 81.73% respondents do not.

Ethnic minority is controlled as previous studies show that ethnic minority group are more prejudiced and restricted in the private sector employment, and they are found more attracted by the public sector work (Battu & Zenou, 2010; Doverspike, Qin, Magee, Snell, & Vaiana, 2011). In this study, 9.86% respondents are ethnic minority and 90.14% respondents are not.

Religious belief is controlled as previous studies show that people with religious belief are more likely to choose public sector career and more dedicated to the public interest for their conception to take the public work as a kind of divine ‘calling’ (Hirsbrunner, Loeffler, & Rompf, 2012; Teresa Flanigan, 2010). In this study, 13.18% respondents have religious belief and 86.82% respondents do not.

Grouping difference

We match the data and construct the dataset that contains variables above without missing observations. The estimated sample thus includes 9293 non-missing observations. The overview of variables predicting the 1st job choice is displayed in Table 4.

Table 4. The overview of variables (grouping by the 1st job choice)

	Public sector	Private sector	Combined		The diff.
Variables	Mean	Mean	Mean	S.D.	t-value
The 1^st job choice	1.0000	0.0000	0.8129	0.3900
Childhood social-economic situation
Relative family financial status	3.5601	3.2716	3.5152	0.9816	−13.972**
Childhood community/campus quality
Neighbourhood friendship	1.6240	1.5370	1.6105	0.6096	−6.7140**
Community bullying	3.5813	3.5445	3.5558	0.8179	−2.1456*
Campus safety concern	1.2016	1.1280	1.1936	0.6213	−5.3549**
Feeling lonely for no friends	1.4092	1.2760	1.3931	0.8539	−7.3616**
Childhood family climate
Relationship between parents	2.5555	2.4165	2.5170	1.1948	−5.2791**
Parental panic and feeling upset	1.4337	1.2659	1.3998	0.7496	−9.8287**
Parental depression and sadness	1.1616	1.0971	0.1468	0.3141	−8.9655**
Parental anxiety and nervousness	1.5503	1.4107	1.5237	0.8434	−7.2759**
Demographic information
Gender	1.5190	1.4761	1.5254	0.4994	−4.0800**
Education	3.5946	4.9433	3.7850	2.1766	29.9094**
Membership of CPC	0.1866	0.2266	0.1827	0.3865	4.8057**
Ethnic minority	0.0909	0.0902	0.9862	0.2982	−0.1103
Religion	0.1279	0.1388	0.1318	0.3383	1.5373

The independent t-test of the mean is implemented. ^†p < .10, *p < .05, **p < .01

The results of Independent Sample T-test indicate that the mean values of independent variables significantly differ between respondents choosing the public sector and the private sector (Relative family financial status, t = −13.972; Neighbourhood friendship, t = −6.7140; Campus safety concern, t = −5.3549; Feeling lonely for no friends, t = −7.3616; Relationship between parents, t = −5.2791; Parental panic and feeling upset, t = −9.8287; Parental depression and sadness, t = −8.9655; Parental anxiety and nervousness, t = −7.2759; all above with p-value<0.01; Community bullying, t = −6.7140; with p-value<0.05).

Statistical analysis

Binary choice models

The statistical software Stata 13.1 is used in this study. Three types of binary choice models (i.e. the probit model, logit model and linear probability model) are used in the present study. The regression equations are displayed as follows:

Probability (Career choice = 1 | x₁, x₂, …, x_n) = G (β₀+ x₁β₁+…+x_nβ_n)

Where, x_i are independent variables, and β_i are estimated parameters. The function G(·) is the cumulative distribution function of the normal distribution/logistic distribution for the probit/logit model, while G(·) is linear when the linear probability model is used as the regression model.

The estimation of conditional marginal effect

Although the above estimation can provide the rough capturing of the effect of independent variables on dependent variable, the results of estimated nonlinear models (probit/logit model) reporting the mean marginal effect can cover the slight discrepancy of the marginal effect at different sample points. Therefore, for providing more detailed results, it becomes worthwhile to estimate the conditional marginal effect (CME) of independent variable when covariates are at different percentile.

More specifically, the marginal effect of independent variable x_i to the response probability P (y = 1 | x₁, x₂, …, x_n) can be derived as

Δ P (y = 1 | x₁, x₂, …, x_n)/Δ x_i ≈ g (β₀+ x₁β₁+…+x_nβ_n)β_i

where g(·) is the probability density function of the normal/logistic distribution for the probit/logit model. Since g(·) is a nonlinear function rather than a constant, the above CME Δ P (y = 1 | x₁, x₂, …, x_n)/Δ x_i varies for different value combination of (x₁, x₂, …, x_n). In practice, the common regression reports the mean marginal effect. If the independent variable x_i is continuous, the consistent estimator can be expressed as E[g(x_1,…x_n; β_1,…β_n)]β_i = [n⁻¹Σ_1≤i≤n g (x_1,…x_n; β_1,…β_n)]β_i. If the independent variable x_i is discrete, the consistent estimator can be expressed as E[ΔG(x_1,…x_n; β_1,…β_n)] = n⁻¹[Σ_1≤i≤n G (β₀+ x₁β₁+…+x_i-1β_i-1+ β_i.)- G(β₀+ x₁β₁+…+x_i-1β_i-1)].

In this study, instead of just reporting the mean marginal effect, we estimate and illustrate the CMEs of independent variables when covariates are at the 5, 30, 70, 95 percentile, respectively (logit model), which can provide more details.

Empirical results

The effect of childhood experiences on the 1st job choice

Table 5 reports the main results of this study. We estimate the impacts of childhood social-economic situation/childhood community and campus situation/childhood family climate on the 1st job choice. Results confirm that the 1st job choice is significantly associated with above aspects of childhood experiences. In detail, the deterioration of relative family financial status, poor neighbourhood friendship, community bullying, campus safety concern, feeling lonely for no friends, poor relationship between parents, parental panic and feeling upset, parental depression and sadness are shown to significantly increase the likelihood of choosing the public sector in the 1st job choice.

Table 5. The effects of childhood experiences on the 1st job choice

	Dependent variable: Job 1
Variables	Linear model	Probit model	Logit model
Childhood economic situation
Relative family financial status	0.0302** [0.0048]	0.1065** [0.0174]	0.1929** [0.0303]
Childhood community/campus quality
Neighbourhood friendship	0.0237** [0.0071]	0.0860** [0.0272]	0.1498** [0.0474]
Community bullying	0.0215** [0.0057]	0.0752** [0.0205]	0.1361** [0.0352]
Campus safety concern	0.0192** [0.0061]	0.0778** [0.0277]	0.1408** [0.0500]
Feeling lonely for no friends	0.0091^† [0.0055]	0.0414^† [0.0229]	0.0766^† [0.0414]
Childhood family climate
Relationship between parents	0.0083* [0.0038]	0.0284* [0.0142]	0.0524* [0.0247]
Parental panic and feeling upset	0.0161* [0.0074]	0.0718* [0.0315]	0.1289* [0.0567]
Parental depression and sadness	0.0376* [0.0736]	0.1606* [0.0630]	0.2833* [0.1135]
Parental anxiety and nervousness	0.0068 [0.0063]	0.0223 [0.0248]	0.0446 [0.0437]
Demographic variables
Gender	−0.0440** [0.0093]	−0.1467** [0.0323]	−0.2561** [0.0558]
Education	−0.0757** [0.0053]	−0.2385** [0.0161]	−0.3925** [0.0276]
Membership of CPC	0.0491** [0.0128]	0.1655** [0.0452]	0.3024** [0.0804]
Ethnic minority	0.0198 [0.0200]	0.0762 [0.0715]	0.1286 [0.1248]
Religion	−0.0291^† [0.0160]	−0.1099* [0.0543]	−0.1834^† [0.0939]
Intercept term	1.1078** [0.0634]	1.9414** [0.2493]	3.1793** [0.4443]
Number of observations	9293	9293	9293
Log likelihood	–	−4630.68	−4634.71
Ward χ^2 statistics [P-value]	–	408.32 [0.0000]	381.50 [0.0000]
F-statistics [P-value]	26.70 [0.0000]	–	–

Job 1 = The types of the first job. Estimated standard errors have been clustered on the community level and reported in brackets. ^†p < .10, *p < .05, **p < .01.

The conditional marginal effect (CME) of childhood experiences on the 1^st job choice

To provide more detailed estimated results of nonlinear models, we also compute the CMEs of above independent variables on the 1st job choice at 5, 30, 70, 95 percentile of covariates, respectively (although we choose the logit model here, probit model results do not show differences in significance level). Results of Figures 1–8 show that there have the nonlinear CMEs of above independent variables on the probability to choose the public sector as the 1st job choice. More specifically, for a given level of relative family financial status, the CME is ranked from high to low when covariates are at 70, 5, 30, 95 percentile, and for other independent variables, the CME is ranked from high to low when covariates are at 5, 70, 30, 95 percentile. Besides, for covariates are at given percentile, the CMEs are higher when the independent variables are lower, and thus all the Figures 1–8 show a decreasing tendency (a down-stairs shape).

Figure 1. The conditional marginal effects of relative family financial status on the probability of choosing public sector as the 1st job choice (according to logit model regression)

Figure 8. The conditional marginal effects of parental depression and sadness on the probability of choosing public sector as the 1st job choice (according to logit model regression)

The effect of childhood experiences on the 1st job choice (by gender and migration grouping)

Table 6 reports the respective regression results by gender (male and female) and migration experiences (migrants and people that never migrate). Results show males are generally more sensitive to the childhood experiences for the 1st job choice compared with females, and this is more specifically reflected in some aspects of childhood experiences (community bullying, feeling lonely for no friends, parental panic and feeling upset, parental depression and sadness). Besides, migrants are generally more sensitive to the childhood family climate compared with people that never migrate for the 1st job choice. People that never migrate are generally more sensitive to the childhood community/campus quality compared with migrants for the 1st job choice.

Table 6. The effects of childhood experiences on the 1st job choice (sub-groups)

Estimated standard errors have been clustered on the community level and reported in brackets. (light blue)^† p < .10, (moderate blue)* p < .05, (deep blue)** p < .01.

The effect of childhood experiences on the 2nd and 3rd job choice

Another important issue raising our concern is whether childhood experiences would have long and lasting effect on the career choice. In order to respond to this concern, we examine the effects of childhood experiences on the 2nd and 3rd job choice. Among observations, we have 4365 non-missing ones reporting the 2nd job and 3294 non-missing ones reporting the 3rd job. Table 7 reports the results of the effects of childhood experiences on the career choice for the 2nd and 3rd job. The 2nd job choice is found to still be significantly affected by community bullying, and parental panic and feeling upset, but to no longer be significantly affected by other determinants mentioned above. While for the 3rd job choice, the influences of all the determinants above recede into insignificance.

Table 7. The effects of childhood experiences on the 2nd and 3rd job choice

Job 2, Job3 = The types of the 2^nd and 3^rd job. Estimated standard errors have been clustered on the community level and reported in brackets. (light blue)^† p < .10, (moderate blue)* p < .05, (deep blue)** p < .01.

Discussion

The special attention of this study to the early career choice can be particularly worthwhile. Previous studies show that the prestige and quality of the 1st job can lay the foundation for the future career development, and the early career can significantly affect people’s entire career path (Judge, Kammeyer-Mueller, & Bretz, 2004). Also, the significant associations between the 1st job and subsequent jobs in many aspects, such as the job type, earnings loss, temporality, have been identified by previous studies (Gebel, 2010; Hamaaki, Hori, Maeda, & Murata, 2013; Wright & Christensen, 2010). As such, the early career choice has great implications for people life.

Also, the exploration of childhood experiences as determinants of the early career choice has special importance. Different from the present study that focuses on the experiences during a comparatively long period of early life, the existing literature that addresses the early career decision-making mainly investigates the traits of college students (e.g. career aspiration, entrepreneurship intention; Farmer, Wardrop, Anderson, & Risinger, 1995; Pihie & Akmaliah, 2009; Rothstein & Rouse, 2011). This treatment considers the traits of college students to be given and stable, and thus can inevitably omit that the traits are gradually formed in the long period of childhood. In this case, the investigation of childhood experiences in the present study can be more robust in tracing the origins.

Besides, the empirical finding suggests that male adolescences’ career decision-making is more susceptible to ACEs than females’. This finding is in line with previous empirical evidence that the career decision-making of males and females are dichotomous (Larson, Butler, Wilson, Medora, & Allgood, 1994). Males are socialized primarily to be the family financial providers, a role expressed through career endeavours (Larson et al., 1994). Consequently, males take their career decision-making extremely seriously and become more susceptible to their past experiences. In addition, empirical finding suggests that migrants are more sensitive to childhood family climate (less sensitive to childhood community/campus quality) compared with people that never migrate. This finding is consistent with the perception that people that never migrate can have more frequent interaction within the community and thus become more susceptible to the community influence than migrants. Also, migrants who experience residence place change can more heavily rely on their within-family support compared with people that never migrate, and thus more susceptible to the childhood family climate.

However, the determinants of career choice are complex, and it is very likely that the childhood experiences are not the only perspective of exploring the antecedents of young adolescents’ early career choice. In contrast with the focus on the childhood experiences in our study, many studies concentrate on the linkages between the career choice and internal traits, such as public service motivation (Carpenter, Doverspike, & Miguel, 2012; Clerkin & Coggburn, 2012; Gabris & Simo, 1995; Pedersen, 2013), altruism (Jirwe & Rudman, 2012), and personality (Antony, 1998; Garcia-Sedeñto, Navarro, & Menacho, 2009). Although far beyond the scope of this study, these perspectives have shown that the motivations of early career decision-making are dynamic and the precise prediction of young adolescents’ early career choice can be difficult. As such, the findings of this study may roughly shed some light on how childhood experiences can affect young adolescents’ early career choice, rather than provides a solution to accurately predict their career decisions merely according to their childhood experiences.

Conclusion

This study empirically investigates the associate between childhood experiences and the early career choice of young adolescents. Empirical results show that the ACEs can significantly increase the likelihood of choosing a career in the public sector for the 1st job, but such impact turns less significant for the 2nd job and insignificant for the 3rd job. Also, males are generally more sensitive to the childhood experiences for the 1st job choice compared with females. Migrants are generally more sensitive to the childhood family climate (less sensitive to the childhood community/campus quality) for the 1st job choice compared with people that never migrate. These findings are consistent with the cognition that individual career development is dynamic with the growth of experiences, expansion of vision, and accumulation of social connections and resources. Thus childhood experiences can be of most importance for the early career choice, and such impact is just distinctly reflected in the choice of the 1st job.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This research was funded by Philosophy and Social Science “13-5” Planning Project of Guangzhou, (“Health and Medical Big Data Development in Guangzhou”, grant number 2019GZQN04).

Notes on contributors

Bocong Yuan

Bocong Yuan is a research associate professor of School of Tourism Management, Sun Yat-sen University. His current research focuses on development economics and public policy (industrial, environment and public health).

Jiannan Li

Jiannan Li is a research associate professor of International School of Business & Finance, Sun Yat-sen University. Her main research focuses on organizational behavior and occupational health.