1 Introduction

A high labor force participation rate (LFPR) is a strong indicator of a vibrant economy. In the USA, LFPR is officially defined as the percentage of the non-institutionalized civilian population 16 years of age and older that is either employed or actively seeking employment [1]. In contrast, the more commonly used unemployment rate is based only on those who are actively seeking work and excludes those who have given up the search for unemployment or are absent from the workforce for other reasons [1]. The LFPR is an important economic indicator for several reasons. A lower LFPR means that a smaller proportion of the potential labor force is contributing to the production of goods and services, thus limiting the growth of gross domestic product (GDP). A lower LFPR also can foster higher taxes because of the smaller base that can provide government revenue. As well, lower LFPRs may generate lower unemployment rates which paradoxically can result in that economic indicator conveying a false sense of a positive economic environment.

LFPRs differ considerably across the 50 American states. For example, in April of 2023, the mean seasonally adjusted state LFPR was 62.62. The minimum was 54.5 in Mississippi and the maximum was 69.6 in North Dakota [2]. It should be noted too that the national LFPR has been in decline since 2000 [3] especially among males, and that recent data pertaining mostly to 2022 show that 34 other countries have a higher LFPR than the USA [4].

Many factors have been suggested as influences on LFPRs such as unemployment rates, business cycles, industrialization, wealth accumulation, age distributions, demand for employees, supply of labor, race, health and disability levels, education and skill levels, technological advances in the workplace, increasing globalization, declining real wages, reliance on government policies and programs, higher incarceration rates, retirement, and urbanization [e.g.,1, 3, 5]. Perhaps many of these environmental or general population factors have more influence on national LFPRs than on state differences in LFPRs. Nevertheless, it appears that no researchers have specifically focused on dispositional factors as potential predictors of LFPRs. As a step toward filling this void, the present research determines the capacity of state resident intelligence (IQ) [6] and Big Five personality [7] to predict state LFPRs without and with a selection of environmental variables statistically controlled.

The nature of the six dispositional variables have been widely documented [8]. IQ is a general ability factor underlying cognitive skills such as reasoning, problem solving, learning, memory, verbal ability, and abstract thinking. Intelligence is highly heritable with estimates ranging from 40 to 85 percent with higher estimates in adulthood than childhood [9]. The IQ scores developed by Pesta [6] for each of the 50 states were based on a combination of standardized test results assessing reading and mathematics at the fourth and eighth grade level and adult literacy and numeracy during the 2012–2019 period.

The “Big Five” is the most broadly accepted contemporary model of nonpathological personality differences [e.g., 10, 11]. It contains five important dispositional dimensions, each containing several facets [10] (see Table 1). The Big Five personality factors have strong heritability estimates of between 40 and 60 percent [e.g., 12]. The state-level Big Five scores used in the present study were developed by Rentfrow et al. [7] based on the responses of over 600,000 American residents to the Big Five Inventory [11] in an internet survey conducted from 1999 to 2005. These state scores have been demonstrated to be quite temporally stable and to have retained their predictive integrity well into the 2013–2017 period analyzed in the present study (e.g., 13).

Table 1 Big Five personality dimensions and examples of their facets
Full size table

1.1 IQ and the LFPR

Dispositional IQ probably has some impact on outcomes in most life contexts [e.g., 14]. Employment participation is not likely to be an exception. Therefore, it is reasonable to assume that higher IQ should lead to better adaptation and more successful coping strategies regarding the challenges inherent in the preparations necessary to enter the workforce and the actions required to avoid being unemployed.

Individual-level research has offered indirect evidence suggesting that IQ should be related to American state LFPRs. For example, higher IQ has been associated with a greater chance of being employed [e.g., 15], with better job performance [e.g., 16], and with greater work satisfaction [e.g., 17]. Whether state-level IQ estimates are related to state LFPRs remains open to empirical verification, but the expectation of such relations seems warranted.

1.2 Big Five and the LFPR

Each of the Big Five personality variables has elements that suggest a potential link to the capacity to obtain and hold a job. For example, according to a recent model of employability put forth by Bourdeaux et al. [18], to be considered employable essentially is to be perceived as a person who is Rewarding, Willing, and Able. The Bourdeaux model places the capacity to be Rewarding and the capacity to be Willing on the dispositional features of higher Conscientiousness, higher Extraversion, higher Agreeableness, and lower Neuroticism. The capacity to be Able is rooted in higher Openness to Experience and lower Neuroticism. Some aspects of IQ also are implicated in the capacity to be Rewarding and to be Able. In addition to separate Rewarding, Willing, and Able scores, the model also combines values on the three dimensions to form an overall “employability” score. Bourdeaux et al. [18] demonstrated that overall composite employability scores are associated in descending order with lower Neuroticism and higher Conscientiousness, Extraversion, Agreeableness, and Openness to Experience. They also are associated with aspects of higher IQ.

Other research results also show relations of each of the Big Five personality dimensions to criteria that are explicitly or implicitly associated with workforce participation. These studies have been conducted with individuals rather than states as the analytic units. Many of the findings are implicitly supportive of the Bourdeaux et al. [18] model of employability. However, none of these studies has specifically considered LFPRs as the outcome variable. The following is a brief overview of some of this research literature.

Being higher on Openness to Experience would appear to be related to garnering success as an employee in many settings. However, research results concerning its relation to employment remain mixed. For example, higher Openness to Experience has been found to be associated with higher ratings of perceived employability [19], but it also has been found to be negatively related to female participation in the workforce [20], and even unrelated to employment status [21].

The general tendency of persons higher on Conscientiousness to be organized and committed to hard work naturally pushes them toward higher employability. They tend to set goals and strive to succeed. To Wille et al. [19], their characteristics influence work activities and attitudes, and these in turn promote more positive perceptions of employability. Other research has found that those higher on Conscientiousness also are more likely to be employed [e.g., 20, 21], and to display higher levels of job-search networking intensity and other traditional job-search methods [22].

Extraverts tend to be more active, assertive, gregarious, warm, and positive than introverts [10]. These are all characteristics that should serve well in the workforce. Some researchers have reported that extraverts are inclined to receive higher ratings of employability [19], to manifest greater job-search networking intensity and use of other traditional job-search methods [22], and to be more likely to be workforce participants [20]. However, others have found that extraversion is unrelated to employment status [e.g., 21].

The tendency of those higher on Agreeableness to be more straightforward, trustful, cooperative, and compliant [10] also would seem to be characteristics desirable in an employee. In contrast, those lower on Agreeableness tend to be more callous, antagonistic, and cynical—all characteristics that can contribute to a toxic workplace. Surprisingly, those higher on Agreeableness have been found to receive lower ratings of employability [19] and to be less likely to be employed [21]. As well, others have found that Agreeableness is unrelated to employment status [e.g., 20].

Individuals higher on the Neuroticism dimension are prone to experience anxiety, anger, irritability, depression, and impulsiveness [10]. Such characteristics would appear to be detrimental to finding and maintaining suitable employment. Many studies support that view. For example, higher Neuroticism has been associated with lower ratings of perceived employability [19] and generally lower chances of being in the workforce [20, 21].

1.3 Environmental variables and the LFPR

As noted earlier, many environmental factors have been suggested by past researchers to have an impact on LFPR. The strategy in the present study was to focus on a manageable number of seemingly important variables with clear and readily available state-level data to serve as statistical controls. The final selection included White population percent, urban population percent, income, political preference, and age distribution. Race, urbanization, and age distribution were chosen as control variables because they were among the variables suggested as influences on LFPR in the past that also had readily available state data. Income was chosen because state data were available, and it also serves somewhat as a proxy for industrialization, wealth accumulation, wages, employment, and education and skill levels. Political preference was chosen because state data were available, and because it may also serve as a rough proxy regarding reliance on government policies and programs and higher incarceration rates.

Retirement was not chosen because it is somewhat reflected in the age distribution variable. Health and disability levels were not chosen largely because that would necessitate the inclusion of several additional variables to cover the spectrum of what “health and disability” entails, and because health and disability levels are somewhat indirectly reflected in the age distribution and income variables. Other factors such as business cycles, demand for employees, supply of labor, technological advances in the workplace, and increasing globalization pertain more to the nation than to individual states, and the ready availability of appropriate state-level data is much less likely. Therefore, the selection of control variables generally was based on suitability and data availability, while keeping the number at a minimum to preserve degrees of freedom and thereby increase the chances of finding significant relations among the variables that are most important to the thrust of the present study. The following is a very brief account of research concerning these five chosen state environmental factors that suggests they might be predictive of state LFPRs.

There is evidence that race is related to employment factors. Participation in the U.S. workforce varies by race [e.g., 3]. Such rates are lower for Blacks than Whites and they have declined most acutely for Black men in recent years [e.g., 1, 3]. Using U.S. data from 2011 to 2015, Brucker et al. [23] also have shown that scores on measures of “striving to work” differ according to race. Lower LFPRs for Blacks may occur because they tend to live in more economically depressed areas where educational attainment levels are lower and fewer employment opportunities exist [24]. High Black incarceration rates may also have direct and indirect impacts on their LFPR [25].

The sparse research results pertaining to relations between urbanization and LFPRs tends to indicate that it should be positively related to state LFPRs. Greater urbanization has been shown to relate to higher LFPRs in advanced national economies [5]. Greater urbanization also has been associated with higher rates of employment [26]. Higher LFPRs may occur in urban areas because urban populations tend to be somewhat younger than rural populations, and older persons are less likely to participate in the labor force [27]. As well, urban populations have higher educational attainment levels, and more educated persons also are more likely to be labor force participants [28].

Both higher income and higher educational attainment also have been found to be associated with greater LFPRs in advanced national economies [5]. Over the past several decades, there has been a sharper decline in the job possibilities for those with lower educational attainment [e.g., 1, 3]. However, of these two important aspects of socioeconomic status (SES), educational attainment—which also is predictive of income—has been excluded in the current research because it can be regarded as a proxy for IQ [e.g., 14], which is one of the six dispositional variables viewed as deserving primary potential predictor status in relation to state LFPRs in the present study.

No research could be located that has directly examined the relation of political preference in the USA to LFPRs. However, lower unemployment rates have been associated with greater Republican control of the White House and Congress [29] and with state populations that tend to identify as Republican [30]. Whether such relations transfer to state LFPRs is an open question.

Shifts in age distributions also can have an impact on LFPRs. For example, a table by Aaronson et al. [31] based on U.S. data from 2005 indicates that the LFPR for males was 30.6 for ages 16–17, 57.9 for ages 18–19, 79.1 for age 20–24, 90.7 for ages 25 to 29, and then it steadily declined to 85.8 for ages 50–54 and down to 52.5 for ages 62–64. It also seems reasonable to assume that the size of a particular age group should make it easier or more difficult to secure jobs that are available and appropriate for that age group.

1.4 The present study

The current research was conducted to determine the relations of state LFPRs to state-level IQ, state Big Five personality variables, and the five selected state environmental variables. Total, male, and female state LFPRs based on those 20–64 years of age served as the three outcome variables. Statistical strategies analyzed the predictive capacities of the 11 potential predictors using Pearson correlations and sequential and simultaneous multiple regression strategies. The prime focus was on the independent predictive capacities of IQ and the Big Five when the environmental variables were statistically controlled.

Variables in the present study only retained data for the 48 contiguous American states. Alaska and Hawaii were excluded for several reasons. The distant and remote nature of Alaska and Hawaii somewhat impedes and limits the flow of diverse commercial transactions, industrial developments, and employment opportunities that are common between the 48 contiguous states. As well, Alaska has the harshest northern climate and Hawaii has by far the lowest White population percent. These are all factors which conceivably could influence LFPRs and potentially distort the results of the present inquiry. However, another benefit of such exclusion concerns the desirability of using only states that have borders with other states in the spatial autocorrelation procedures used here to examine the potential impacts of spatial autocorrelation on the interpretation of the ordinary least squares (OLS) multiple regression equation results.

This research was conducted from the perspective of geographical psychology [e.g., 7, 32], a prime objective of which is to understand how spatial dispersions of standings on dispositional factors relate to other variables at the macro level. For theoretical background and the development of hypotheses, users of the approach often rely on psychological theory and empirical evidence generated with individuals as the analytic units [32]. A foundational assumption is that the aggregate position on a dispositional variable in a geographical unit is reflective of the central tendency of the residents of that unit and is associated with the pervasiveness in that unit of the psychological and behavioral proclivities connected to that dispositional variable. Proponents of this developing perspective are thoroughly cognizant of inherent fallibilities in assumptions about cross-level generalizations such as those considered by other researchers such as Robinson [33] regarding the ecological fallacy and Pettigrew [34] regarding the compositional fallacy. Nonetheless, it also is thought that judicious extrapolative reasoning supported by empirical evidence can be used to speculatively deduce potential implications for aggregate-level relations from individual-level processes and for individual-level processes from aggregate-level relations.

2 Method

2.1 Measures

2.1.1 Labor force participation (LFPR)

The study used three LFPR variables: total, male, and female percentages of persons 20–64 years of age in the labor force at some point over a previous 12 month span. The data were based on American Community Survey 2013–2017 5-year estimates [35]. Higher scores indicate greater participation in the state workforce.

2.1.2 General population IQ

State IQ values recently developed by Pesta [6] served in the present study. They were based on state-level results from the fourth and eighth grade math and reading scores of the National Assessment of Educational Progress (NAEP) and the adult numeracy and literacy scores from the Program for the International Assessment of Adult Competencies (PIAAC). The NAEP scores were from the years 2015, 2017, and 2019. The PIAAC scores were from the years 2012 to 2017. Test–retest reliabilities over the years used were 0.84 for mathematics, 0.86 for reading, 0.99 for numeracy, and 0.99 for literacy. The final state IQ estimate was the mean of the chosen NAEP and PIAAC scores.

2.1.3 Big Five personality

Based on responses to the 44-item Big Five Inventory [11] by 619,397 state residents in an internet survey carried out from December of 1999 to January of 2005, Rentfrow et al. [7] computed state z scores for Openness to Experience, Extraversion, Agreeableness, Conscientiousness, and Neuroticism. They concluded that the sample generally was representative of the national population and that the survey drew participants from each state in direct proportion to the 2000 census figures regarding population and racial composition. However, the sample was somewhat less representative regarding social class and was younger than the general population. They also concluded that “the state-level factor structure was virtually identical to the factor structure commonly found at the individual level” (p. 349) and that the Big Five assessments were reliable with mean Cronbach alphas of 0.89 at the state level and 0.81 at the individual level. Validity was demonstrated by relations between the state Big Five z scores and various indicators of state crime, health, employment, religiosity, social capital, and political values [7]. The state scores also show temporal stability and have been successfully used in various other contexts [e.g., 13, 36].

2.1.4 White population percent

White population percent for each state was obtained for 2010 from the Kaiser Family Foundation (KFF) [37], and for 2020 from the U.S. Census Bureau [38]. The Pearson correlation was 0.98 between the White population percents for 2010 and 2020. Prorated estimates were created for 2013, 2014, 2015, 2016, and 2017. For example, the 2010 value was subtracted from the 2020 value, the difference was divided by nine, and six times this result was added to the value for 2010 to produce an estimated value for 2016 for each of the 48 contiguous states. The mean of the years 2013–2017 served as the White population percent variable.

2.1.5 Urbanization

Urban population percent for each state was obtained for 2010 from Iowa state University (ISU) [39]. An alternative urbanization measure developed by Rakich [40] was used for 2020. The Pearson correlation was 0.90 between the two measures. Prorated urban population percent estimates then were created using the procedure for White population percent. However, the prorated estimates were based on standardized urbanization values for 2010 and 2020 because the 2020 variable yielded values not directly comparable to those for 2010. The results then were standardized to produce the final urbanization variable. Higher scores indicate a larger urban population percent.

2.1.6 Income

Per capita personal income for each state in 2013, 2014, 2015, 2016, and 2017 was obtained from the Bureau of Economic Analysis (BEA) [41]. The mean for the 5 years functioned as the measure of income in Study 1. The measure has an extremely high Cronbach alpha of 0.995.

2.1.7 Political preference

The gauge of political preference was a composite based on liberal-conservative ideological orientation and Democratic-Republican political party support. Citizen ideology values for each state in 2012 and 2016 were provided by Fording [42]. Higher scores indicate greater liberalism. The percent of the popular vote won in each state by the national Democratic and Republican presidential candidates in 2012 and 2016 was taken from Leip [43]. To compute a Democratic party preference variable for 2012 and for 2016, the Republican percentages were subtracted from the Democratic percentages. To produce a final political preference variable, the four ideology and party preference variables were transformed to z scores, summed, and the resulting composite was standardized. Higher scores indicate a greater tendency to favor a liberal ideology and a Democratic presidential candidate. The final political preference variable has a Cronbach alpha of 0.96.

2.1.8 Age distribution

Suitable age data for the percent of the population 20–64 years of age in each state for the 2013–2017 period could not be located. Therefore, computations were based on data obtained from the KFF [44] for a close alternative age category of 19–64 for the same period. Displayed percents for those up to 18 years of age and for those 65 and over in 2013, 2014, 2015, 2016, and 2017 were summed, divided by five, and subtracted from 100 to create the age distribution variable used here. It has a Cronbach alpha of 0.99.

2.2 Analytic strategy

Planned Pearson correlations and multiple regressions determined the relations between total, male, and female LFPR and the six dispositional and five environmental variables. For each LFPR criterion, two 2-step multiple regression equations were computed. In the first equation, the six dispositional variables were forced to enter as a block on the first step, and the five environmental variables were forced to enter as a block on the second step. The second equation was similar, but the block order was reversed. Each of the six regression equations also was tested for spatial autocorrelation. Two-tailed significance tests and an alpha level of 0.05 were used throughout.

3 Results

Table 2 displays the means, standard deviations, and Pearson correlations for each of the 14 variables. Among the most notable correlations, total LFPR correlated significantly with IQ (0.82), Neuroticism (− 0.34), White population percent (0.31), income (0.69), political preference (0.29), and age distribution (0.35). Male LFPR correlated significantly with IQ (0.78), Neuroticism (− 0.49), and income (0.65). Female LFPR correlated significantly with IQ (0.78), Conscientiousness (− 0.29), White population percent (0.35), income (0.66), political preference (0.40), and age distribution (0.41). Correlations between the three LFPR variables were high: Total LFPR correlated with male LFPR (0.95) and female LFPR (0.96), and male LFPR correlated with female LFPR (0.83).

Table 2 Means, standard deviations, and Pearson correlations
Full size table

The results of multiple regression equations demonstrating the relations of the total LFPR criterion to IQ, Big Five personality, White population percent, urbanization, income, political preference, and age distribution are presented in Table 3. In Step 1 of Equation 1 with total LFPR as the criterion, IQ and the Big Five as a block accounted for 74.2% of the LFPR variance, F(6, 41) = 19.62, p < 0.001. In Step 2 of Equation 1, White population percent, urbanization, income, political preference, and age distribution entered as a block accounted for another 11.0%, F(5, 36) = 5.32, p < 0.001. The significant βs for Equation 1 were 0.67 for IQ and − 0.20 for Neuroticism. Equation 2 simply reversed the order of Step 1 and Step 2. Now, the five environmental variables entered together on Step 1 accounted for 62.1% of the variance, F(5, 42) = 13.79, p < 0.001, and the six dispositional variables entered together on Step 2 accounted for a further 23.0%, F(6, 36) = 9.28, p < 0.001. Of course, the significant βs from Equation 1 also apply to Equation 2.

Table 3 Multiple regression equations demonstrating the relations of total LFPR to IQ, Big Five personality, white population percent, urbanization, income, political preference, and age distribution
Full size table

The multiple regression results for male and female LFPR are displayed in Supplementary Table S3A. The results for males paralleled those for total LFPR, and the β of − 0.29 for Neuroticism was somewhat larger than that found for total LFPR. With female LFPR as the criterion, the results were somewhat different. IQ had a significant β of 0.59 but Neuroticism did not emerge as an independent predictor. Instead, Openness to Experience had a significant β of − 0.29 and political preference had one of 0.44.

3.1 Spatial autocorrelation analysis

The full regression equations in Table 3 and Supplementary Table 3A were tested with the global Moran’s I test for detecting the presence of residual spatial autocorrelation [45]. To facilitate the tests using the R spdep, spatialreg, and haven packages [46], all variables were transformed to z scores. A Queen’s binary neighborhood matrix was also created that treated each state that touched another state at any point as a neighbor. According to the R program requirements, each of the full original regression equations was tested by entering all the variables from the original equation simultaneously. For Equation 1 and Equation 2, Moran’s I statistic standard deviate was 0.18 (p = 0.429) for total LFPR, 0.12 (p = 0.453) for male LFPR, and 0.14 (p = 0.443) for female LFPR. These results indicate that spatial autocorrelation was not an issue in the present study.

3.2 Additional explorations

Multiple regression equations were computed to determine the variance accounted for in total, male, and female LFPR by their respective block of significant independent contributors. IQ and Neuroticism as a block accounted for 69.3% of the total LFPR variance, F(2, 45) = 50.83, p < 0.001, and 70.7% of the male LFPR variance, F(2, 45) = 54.24, p < 0.001. IQ, Openness to Experience, and political preference jointly accounted for 77.4% of the female LFPR variance, F(3, 44) = 50.37, p < 0.001.

Exploratory analyses also indicated that Pearson correlations and multiple regression equations showed little substantive difference when Alaska and Hawaii were included. For example, respectively for 50 and 48 states, total LFPR correlated significantly 0.81 and 0.82 with IQ, − 0.36 and -0.34 with Neuroticism, 0.69 and 0.69 with income, 0.35 and 0.29 with political preference, and 0.30 and 0.35 with age distribution. The respective correlations for White population percent were a nonsignificant 0.23 and a significant 0.31. For the Table 3 multiple regression analyses, Equation 1 showed that IQ and the Big Five accounted for 73.8% of the LFPR variance and Equation 2 showed that they accounted for 26.5% in the 50-state version, compared to 74.2 and 23.0%, respectively, in the 48-state version. The significant βs were 0.61 for IQ and − 0.20 for Neuroticism for 50 states, compared to respective βs of 0.67 and − 0.20 for 48 states. Income also attained significance with a β of 0.32 for 50 states but had a nonsignificant β of 0.28 (p = 0.061) for 48 states.

4 Discussion

State resident IQ was the predominant predictor of total, male, and female state LFPRs. IQ correlated 0.82 with total LFPR, 0.78 with male LFPR, and 0.78 with female LFPR. The IQ βs for Equation 1 and Equation 2 with each of the three outcome variables were 0.67, 0.71, and 0.59, respectively. The only other significant independent predictors were Neuroticism with βs of -0.20 for total LFPR and − 0.29 for male LFPR, Openness to Experience with a β of − 0.29 for female LFPR, and political preference with a β of 0.44 for female LFPR. Unreported regression equations showed that the two significant predictors could account for 69.3% of the variance in total LFPR and 70.7% in male LFPR, and the three significant predictors could account for 77.4% in female LFPR.

These state-level results pertaining to IQ and Neuroticism as predictors of LFPRs are generally in accord with earlier individual-level research indirectly suggesting the possibility of such relations [e.g., 15, 17, 19]. However, the present results are somewhat discrepant from previous findings regarding Openness to Experience and political preference. Positive associations between Openness to Experience and work-related variables have been reported [e.g., 19], but higher Openness to Experience also has been associated with a lower probability of female workforce participation [20]. As well, previous research based on unemployment rates suggested that higher LFPRs would be associated with Republican rather than Democratic preference [e.g., 29, 30].

This is the first study to explicitly examine the relations of state resident dispositional factors to state LFPRs, and therefore the first to demonstrate that state resident IQ, Neuroticism, and Openness to Experience are important predictors of state LFPRs. Furthermore, the relation for IQ surfaced for total, male, and female LFPR even with White population percent, urban population percent, income, political preference, and age distribution statistically controlled. The relation of Neuroticism to total and male LFPR, and of Openness to Experience for female LFPR, also emerged with these statistical controls.

In line with the tenets of geographical psychology, it is speculated that the present findings regarding the state-level relations of IQ and Neuroticism to LFPR are anchored in and reflect the accumulation of similar individual-level relations. Although relations specific to LFPR have not been researched at the individual level, there is evidence that higher IQ is associated with higher employability scores [18], a higher chance of being employed [e.g., 15], better job performance [e.g., 16], and greater work satisfaction [e.g., 17]. There also is evidence that higher Neuroticism is associated with lower employability scores [18], lower perceived employability [19], and lower chances of being employed [20, 21].

Speculation that the present state-level relations of IQ and Neuroticism to LFPR emanate from individual-level relations also is strengthened by the theoretical plausibility of explanations based on the nature of IQ and Neuroticism. IQ represents a highly heritable and stable [e.g., 9] general cognitive capacity likely to have an impact on most life challenges [e.g., 14]. Higher IQ persons are more likely to successfully adapt to the challenges of searching for a job, securing employment through application and interview, and retaining the job over a long period through strong work performance and social relations. Neuroticism also is highly heritable and quite stable in adulthood [e.g., 12], with features that make it more likely that those on the higher end are less able to land and keep a job. For example, those higher on Neuroticism are hindered by being easily upset, anxious, poor stress managers, impulsive, less likely to be relaxed, irritable, discontented, and less likely to remain calm in strained situations [e.g., 10, 11].

Geographical psychology advocates [e.g., 7, 32] emphasize the importance of place of residence as an abiding factor in the generation of a host of beliefs, attitudes, and behaviors—some of which may pertain to LFPR. It is argued that state gene pools are somewhat different from each other largely because they have been subjected to selective migration partially motivated by the nature of the adaptation characteristics of being nearer one end or the other on dimensions such as IQ and Neuroticism that have strong genetic determination. Therefore, it is not especially surprising that state LFPRs should be related to modal state resident IQ and Neuroticism. One can also logically speculate that those relations also are likely to exist with individuals as the analytic units.

4.1 Limitations and issues

The current study has a few apparent limitations and issues that most readers might not perceive as especially problematic after further scrutiny. First, by conventional standards, the sample size of 48 is far less than optimal for multiple regression analysis. A low ratio of cases to predictors generally is associated with instability in the independent predictors that surface and the magnitude of their regression coefficients. But the sample and population here are the same 48 states. So, the parameters and the relations that are found for the sample also pertain directly to the population. Similar multiple regression strategies often have been successfully employed with such smaller sample sizes [e.g., 47, 48].

Second, readers should be aware that Pearson correlations and multiple regression coefficients often are somewhat larger when based on aggregates rather than individuals [e.g., 49]. The aggregation procedure tends to cancel out measurement errors. Therefore, the possible inflation of correlations and regression coefficients should be considered when making interpretations concerning the magnitude of the current relations. However, heightened relations using aggregates might also be seen as a strength because of the capacity to enhance reliability through measurement error reduction.

Third, the common caution that causation cannot be inferred from correlation must be followed. This is a cross-sectional study with analyses solely based on underlying correlations. Nothing in the study permits confident causal deduction. Causal interpretations must remain in the realm of pure speculation.

Fourth, some readers might have concerns that the state Big Five scores were developed using 1999–2005 data, but the other variables were based on 2010–2020 data with a concentration on 2013–2017. However, state modal personality scores are unlikely to have changed appreciably over this time span. For example, Elleman et al. [13] have shown that there was little fluctuation in such estimates for 1999–2015, the years covered by their research. Relations of state scores to other sociodemographic variables also were quite temporally stable. The state scores also have been employed fruitfully in various state-level research contexts, often pertaining to more recent time periods [e.g., 36].

Fifth, given the correlations among the 11 predictors in the multiple regression equations, multicollinearity might be considered a possible problem by some readers. However, according to Regorz [50], this was not so. No predictors had a VIF over the threshold of 10. In fact, the only VIFs over 5 were IQ at 7.43, White population percent at 7.00, and personal income per capita at 5.04. All others were between 1.82 and 3.41.

Sixth, there is the question of generalization. The present study focused on LFPRs among the 48 contiguous American states during the period 2013–2017. It can only be speculated whether the results pattern would generalize to American cities and counties as analytic units, to other nations and their corresponding administrative units, to studies using individuals as analytic units, or to other time periods.

4.2 Potential implications for application

Here, state resident IQ and Neuroticism were the only significant independent predictors of total and male state LFPRs among the 11 potential predictors tested in a multiple regression framework. As well, state resident IQ, Openness to Experience, and political preference were the only significant independent predictors of female state LFPRs. It is speculated that these predictive relations may have profound implications for policies aimed at increasing state and national LFPRs, and eventually for promoting greater income equity.

Greater opportunities for additional education and skill training are often proposed as remedies for low LFPRs [e.g., 1]. However, lower IQ seriously jeopardizes the chances of success for this approach, and this jeopardy is not limited just to those who are somewhat below average on IQ. Over the past few decades educational and skill acquisition content have become increasingly sophisticated and complicated, requiring a somewhat higher cognitive capacity than was necessary to cope with such demands in earlier times. If more persons without somewhat above average intelligence are going to participate in the workforce in numbers that they once did, then accommodations to enable that outcome are necessary. The first requirement is for policy strategists to recognize that intelligence, which is not effectively amenable to planned alteration, is positively related, most likely in a speculated causal manner, to LFPRs. Given this restriction, accommodative approaches could include greater attention to realistic IQ requirements for participation in conventional education and skill training programs, the development of alternative programs aimed at simpler jobs and those with less cognitive ability, and the offering of greater subsidies for employers to create less complex but still productive jobs for those somewhat farther up the scale of IQ than the current cutoff points now allow.

Neuroticism presents a different challenge. Persons toward the higher end on Neuroticism are probably not as highly represented in the workforce mainly because of their heightened vulnerability to stress. Some of the negative characteristics of Neuroticism might be marginally improved through focused mental health interventions, but the core of the heritable dispositional factor is not likely to change in the face of planned intervention. Therefore, as with IQ, an accommodative process seems like the most effective tack. Again, the first requirement is that policy planners and employers come to accept Neuroticism as a factor regarding LFPRs. In addition, increased attention could be directed toward the assessment of Neuroticism in the hiring process, honest and informed appraisals could be made of the particular stress triggers inherent in the job, efforts could be made to reduce or eliminate such work stressors where possible, job categories could be created and promoted that are less likely to provoke stress reactions among the most vulnerable, and human resources personnel could be more proactive in helping more Neurotic employees constructively deal with workplace stress.

4.3 Future research

There is a paucity of research on the relation of IQ and Neuroticism to actual participation in the workforce with individuals as the analytic units. The results of further empirical inquiry in this context could provide direct cross-level confirmation of the present state-level findings. Such parallel results would bolster the speculation that the present state-level results are rooted in the state accumulation of similar relations occurring at the individual level.

Additional state-level investigation also could be beneficial. For example, with American states as the units of analysis, it would be useful to determine whether any other environmental variables could supplant the independent capacity of IQ or Neuroticism, or both, to predict state LFPRs. There might be additional candidate predictors with little or no relation to state resident IQ and Neuroticism levels that could add to the predictive model. Other state-level variables related to IQ and Neuroticism and LFPRs might even be found that substantially reduce or eliminate the currently demonstrated dispositional relations to LFPRs.

4.4 Conclusion

This is the first study to show that state resident IQ and Neuroticism are significant independent predictors of state LFPRs. The main findings are that higher IQ is associated with higher total, male, and female LFPRs, and that lower Neuroticism is associated with higher total and male LFPRs. Furthermore, these relations are evident with state White population percent, urban population percent, per capita personal income, political preference, and age distribution statistically controlled. With the same statistical controls, higher female LFPRs were found to be associated with liberal political preference and lower Openness to Experience. Awareness of these novel predictors of state LFPRs and their potential applied implications is important given the salience of LFPRs in the functioning of the economy [1], the declines in American LFPRs since 2000 [1], and the 35th-place standing of the USA among other developed nations by 2022 [4]. Low LFPRs do not bode well for a sound economic future. Reductions in the number of American idle adults should be a pressing socioeconomic goal.