Consequences, risk factors, and geography of young people not in education, employment or training (NEET)

Chapter 3 Results

This chapter will present analytical results separately for consequences, risk factors and geographies of NEET.

Consequences of NEET status

This section describes the long-term (10 and 20 year) consequences of being NEET for Cohorts 1 and 2. Since these consequences might be attributed to other factors such as deprivation that precede the period of the outcomes being investigated, such factors (limiting long-term illness, educational qualifications and living in a council area known as a NEET 'hotspot') were adjusted for. Thus the effect of NEET status on this range of outcomes from these models can be attributed to the long-term effect of having been NEET.

A number of socioeconomic and health outcomes have been examined in relation to NEET experiences. They include economic inactivity, low status occupations, limiting long-term illness, hospital admission following an A&E visit, hospital admission following an A&E visit due to self-harm, depression and anxiety prescription and drug misuse.

Key findings - Consequences

There is robust evidence that there is a scarring effect of NEET status in relation to long-term socioeconomic and health outcomes in the 20 years' follow-up.

• The NEET group remains disadvantaged in their educational attainment 10 and 20 years later. More than one in five of NEET young people in 2001 had no qualifications by 2011 compared with only one in twenty five of non-NEETs.
• There is a scarring effect in economic activity. In comparison with their non-NEET peers NEET young people in 2001 were 2.8 times as likely to be unemployed or economically inactive 10 years later.
• The scarring effect is also evident in the occupational positions that NEET young people entered. For example, NEET young people in 2001 were 2.5 times as likely as their non-NEET peers to work in a low status occupation in 2011.
• NEET experiences are associated with a higher risk of poor physical health after 10 and 20 years. The risk for the NEET group was 1.6 - 2.5 times that for the non-NEET group varying with different health outcomes.
• NEET experiences are associated with a higher risk of poor mental health after 10 and 20 years. The risk of depression and anxiety prescription for the NEET group is over 50% higher than that for the non-NEET group.
• Young people who were NEET in 1991 and remained economically inactive in 2001 consistently demonstrated significantly poorer outcomes by 2011 than those who were non-NEET in 1991 and economically active in 2001 and those who were engaged with employment or education in either 1991 or 2001. This suggests that there is a cumulative effect of being out of employment or education on later life chances and this group is the most disadvantaged that need continuing support.
• Young people who changed from NEET status in 1991 to employment or education in 2001 have lower risks of poor life outcomes compared with those who were consistently in disadvantaged positions. However, the negative effect of NEET status in 1991 was not fully discounted by the later engagement of employment or education, indicating the long-lasting detrimental effect of NEET experiences.
• Young people who changed from being non-NEET in 1991 to being economically inactive or unemployed in 2001 have higher risks of poor life outcomes compared with those who were consistently in employment or education. This suggests that economic activity in 2001 is also predictive of later labour market and health outcomes regardless of NEET status in 1991.

Profiles of samples

This section provides summary statistics on NEETs in terms of gender, age and economic activity.

From Table 1 we can see that there were 717 female and 776 male NEETs in 2001 and similarly 1014 female and 958 male NEETs in 1991. Extrapolating to the whole population, this implies that the numbers of female NEET and male NEETs were separately 14340 and 15520 in 2001, and 20280 and 19160 in 1991. Both cohorts are relatively evenly distributed with respect to age. The gender distribution for both cohorts is also fairly even.

The percentage of NEET is higher among males than among females in 2001. This was the reverse of 1991. There was a general trend, with the percentage of NEET increasing with age in both 2001 and 1991.

Table 1 Gender and age profiles of the 16-19 cohorts in 1991 and 2001

2001 (Cohort 1)	Non-NEET (%)	NEET (%)	Row Total (%)
Gender
Female	5,528 (89%)	717 (11%)	6,245 (51%)
Male	5,315 (87%)	776 (13%)	6,091 (49%)
Age in 2001
16	2,976 (93%)	227 (7%)	3,203 (26%)
17	2,662 (88%)	346 (12%)	3,008 (24%)
18	2,572 (84%)	477 (16%)	3,049 (25%)
19	2,656 (86%)	448 (14%)	3,104 (25%)
1991 (Cohort 2)	Non-NEET (%)	NEET (%)	Total (%)
Gender
Female	6,244 (86%)	1,014 (14%)	7,258 (50%)
Male	6,351 (87%)	958 (13%)	7,309 (50%)
Age in 1991
16	3,237 (93%)	262 (7%)	3,499 (24%)
17	3,204 (89%)	395 (11%)	3,599 (25%)
18	3,056 (83%)	610 (17%)	3,666 (25%)
19	3,098 (81%)	705 (19%)	3,803 (26%)

Source: SLS

The summary for economic activity for both cohorts is presented in Table 2. In 2001, 25% of females and 32% males were employed or self-employed, and 64% of females and 56% of males were studying. For females, 5% were unemployed and an equivalent percentage of them were economically inactive due to looking after home or family, or other reasons. For males, 9% reported being unemployed while only 3% reported being economically inactive due to looking after home or other reasons.

The distribution of economic activity categories for ages 16-19 in 1991 was quite different from that of 2001. The most notable change is that from being working in 1991 to being a student in 2001. The overall proportion working in 2001 was approximately half that in 1991, with this change being slightly higher for females. This trend reflects the increasing level of participation in post-compulsory education since the 1990s, and changes in labour market structure in the 1990s. However, the overall level of NEET was similar at 12-13%.

In both 1991 and 2001, males were more likely to be working or unemployed than females while in contrast, females were more likely to be in education, or looking after home or family. The level of those permanently sick was similar among males and females.

Table 2 Economic activity for Cohorts 1 and 2

Economic activity	2001 (Cohort 1)		1991 (Cohort 2)
	female (%)	male (%)	female (%)	male (%)
Working	1,553 (25)	1,931 (32)	3,717 (51)	3,911 (54)
Student	3,975 (64)	3,384 (56)	2,527 (35)	2,440 (33)
Unemployed	333 (5)	558 (9)	646 (9)	889 (12)
look after home	341 (5)	168 (3)	319 (4)	16 (0)
Sick	43 (1)	50 (1)	49 (1)	53 (1)
Total	6,245 (100)	6,091 (100)	7,258 (100)	7,309 (100)

Source: SLS

Educational attainment

This section examines whether NEETs remained disadvantaged 10 and 20 years later in terms of educational attainment.

Table 3 breaks down the level of qualification in 2011 by NEET status in 2001 and 1991. It can be seen that two thirds of those who were NEET were in the bottom two categories (Standard Grades and no qualifications) while only one third of non-NEET were in the same groups. For those non-NEET young people in 2001, 41% of them obtained degree level qualifications by 2011 compared with only 9% of NEETs in 2001. For the 2001 cohort, 22% of NEETs did not have any qualifications by 2011, over five times that of non-NEETs. Similarly 33% of 1991 NEETs did not have qualifications compared with 8% of non-NEETs by 2011.

Table 3 Qualification level in 2011 by NEET status 2001 and 1991

Qualification 2011	2001 (Cohort 1)		1991 (Cohort 2)
	% non-NEET	% NEET	% non-NEET	% NEET
Degree	41	9	36	11
HNC/HND	17	11	15	12
Higher/A-Level	18	13	15	11
S-Grade/O-Level	21	45	27	33
No Qualifications	4	22	8	33
Total	7,945	996	8,980	1,265

Source: SLS

Table 4 highlights the 2011 educational qualification variable dichotomised between those with no qualifications and the rest broken down by the extended NEET classification. Those in NEET categories are more likely to have no qualifications; however, within the NEET categories those reporting permanent sickness appear far more likely than any other category to report having no qualifications. Permanent sickness here is a category from the census question asked in the context of economic activity. It relates to being out of work due to permanent sickness/disability.

In summary, there was little catch up in educational attainment over the life course for the NEET group (any gap year effect appeared small). In addition, education is a gateway to the labour market and is a protective factor for health. Lack of education qualifications 10 and 20 years later for those who were NEET implies their continued disadvantage in later life.

Table 4 Qualification level in 2011 by extended categories of NEET, 2001 and 1991

Economic activity 2001	% with qualifications, 2011	% with no qualifications, 2011	Total
Non-NEET	96	4	7,945
Unemployed	81	19	809
Permanently Sick	50	50	61
Looking after home/family	74	26	126
Economic activity 1991	% with qualifications, 2011	% with no qualifications, 2011	Total
Non-NEET	92	8	8,980
Unemployed	70	30	984
Permanently Sick	35	65	75
Looking after home/family	66	34	206

Source: SLS

Factors included in the models to control for their effects

Before we describe the analysis results of long-term effects of NEET experiences on life chances in detail we summarise the relationships between the variables which were adjusted for: gender, age, educational attainment, Carstairs deprivation, limiting long-term illness and living in a council area NEET 'hotspot' (see Statistical Methods and Appendix 1 for further details) and the outcome measures.

It should be noted that areas of residence for many young people may have changed over the follow-up period due to migration. For all outcomes there was a noticeable trend for outcomes to improve with a higher level of educational attainment. For all outcomes except drugs misuse and hospital admission following A&E visit for self-harm there was a noticeable trend for outcomes to improve with a lower level of deprivation. For all outcomes except drugs misuse there was an association between limiting long-term illness and a poorer outcome. Males were more likely to have had a hospital admission following a visit to A&E and have a record of drugs misuse whereas females were more likely to be economically inactive and have used antidepressant or antianxiety medication. Older ages were more likely to be in employment or education and more likely to have a higher status occupation. Glasgow, Dundee and North Ayrshire were associated with several poorer outcomes. This suggests that both individual factors and contextual factors at these aggregated levels were important in influencing outcomes.

Economic inactivity

In this section we examine labour market outcomes in 2011 for those from the 2001 cohort and the 1991 cohort. The labour market outcome in 2011 was derived from the 2011 census. The economic activity variable was used to classify people into those who were unemployed or economically inactive and the rest, consistent with the division between NEET and non-NEET young people. For simplicity, we use the term 'inactive' to refer to people who were either unemployed or economically inactive and the term 'active' to refer to people who were working or studying. People from both cohorts were predominantly either involved in economic activity or were inactive due to non-educational reasons when they were aged 26-29 or 36-39. Only around 2% of the cohorts were in education or training in 2011.

It can be seen for both the 1991 and 2001 cohorts that those who were NEET were more likely to be economically inactive by 2011 (Table 5). For example, those who were NEET in 2001 were more likely to report subsequent economic inactivity in 2011. About 43% of those who were NEET in 2001 did not engage in employment or study in 2011, compared with 11% of those who were non-NEET (Table 5). Similarly 28% of 1991 NEET young people were out of employment or education when they were aged 36-39, close to three times the rate for non-NEET young individuals.

Table 5 Economic activity in 2011 by 2001 and 1991 NEET status

NEET status, 2001	% Economically active	% Not active	Total
Non-NEET	89	11	7,945
NEET	57	43	996
NEET status 1991	% Economically active	% Not active	Total
non-NEET	90	10	8,980
NEET	72	28	1,265

Source: SLS

Percentages of economic activity in 2011 by the extended categories of economic activity in 2001 and 1991 are presented in Table 6. It shows that roughly 70% of those who were permanently sick in 2001 did not work or study in 2011. And over 40% of those who were unemployed in 2001 or looking after family were not in employment or study at the 2011 Census, compared with 11% of those who were non-NEET. The distribution of economic activity in 2011 for the 1991 cohort is similar to that for the 2001 cohort. One difference is that those who were inactive due to looking after home/family were more likely to participate in employment with 64% in employment or education in 2011. In comparison only half of those in the same category in the 2001 cohort were active by 2011. An explanation for this would be that mothers in the 2001 cohort would be aged 26-29 in 2011 and still likely to be looking after children whereas mothers in the 1991 cohort would be more likely to have returned to work when aged 36-39 in 2011.

Table 6 Economic activity in 2011 by 2001 and 1991 extended categories of NEET

2001 economic activity	% economically active	% economically inactive
Non-NEET	89	11
Unemployed	59	41
Permanently Sick	32	68
Looking after home/family	50	50
Total	7,601	1,340
1991 economic activity	% economically active	% economically inactive
Non-NEET	87	13
Unemployed	61	39
Permanently Sick	32	68
Looking after home/family	64	36
Total	8,530	1,715

Source: SLS

Linking 1991, 2001 and 2011 records allowed us to examine the dynamics of movement into and out of employment or education in the 20 years follow-up. Table 7 shows that those who were NEET in 1991 were more likely to report subsequent economic inactivity in 2001 and/or 2011. Over 50% of those who were NEET in 1991 were not economically active at either or both subsequent censuses, compared with 21% of those who were non-NEET in 1991 (Table 7). Nearly 30% of 1991 NEETs were economically inactive in both 2001 and 2011, compared to only 6% of their non-NEET counterparts. This suggests that this group was the most disadvantaged and would need most assistance to gain employment.

If the estimate is extended to the Scotland population, this indicates that, in total, more than 5,500 (5% SLS sample, 958*29%*20) young people who were NEET in 1991 remained out of employment or education in both 2001 and 2011. Making up only 12% of the 1991 cohort, NEET young people accounted for over 38% of those who remained out of employment or education in both years. Although some of them may be out of employment for family reasons or due to illness, the impact on the size of the workforce was substantial as this group was aged between 16 and 39 in the follow-up period when most could be expected to contribute to the economy. This implies a significant negative impact on the economy through lost output, higher welfare payments and lower tax returns.

Table 7 Economic activity in 2001 and 2011 by 1991 NEET status

Economic activity, 2001 & 2011	% non-NEET 1991	% NEET 1991
Active both 2001 and 2011	79	44
Active 2001, not active 2011	7	10
Not Active 2001, active 2011	8	17
Inactive both 2001 and 2011	6	29
Total	7,306	958

Source: SLS

Table 8 shows the results from models of being economically inactive versus being active at Census 2011. The results in Table 8 show a substantial NEET effect, independent of the other factors in the model. Young people who were NEET in 2001 were nearly three times as likely as their non-NEET counterparts to be out of employment or education in 2011.

The last column in Table 8 shows results from the model for the 1991 cohort of being economically inactive versus economically active in 2011. We used a variable indicating the NEET status in 1991 and economic activity in 2001 as described in the Statistical Methods.

Table 8 Odds ratios of economic inactivity in 2011 from logistic regression

2001 cohort (Cohort 1)			1991 cohort (Cohort 2)
NEET status	Odds ratio	Significance level	NEET 1991 and economic activity 2001	Odds Ratio	Significance level
No	1
Yes	2.77	***
			Non-NEET 91 & active 2001	1
			Non-NEET 91 & inactive 2001	5.75	***
			NEET 91 & active 2001	1.91	***
			NEET 91 & inactive 2001	9.38	***
N	7,917			8,073

*P<0.10 **p<0.05 ***p<0.01, Source: SLS

The results demonstrate the longer term negative effect of NEET status. Those who were NEET in 1991 and out of work or education in 2001 were 9 times as likely not to be economically active in 2011 compared with their non-NEET and subsequently economically active peers. For those young people who were NEET in 1991, economic activity in 2001 did not cancel out the negative effect of having been NEET as they were still significantly more likely to be economically inactive in 2011 compared with the non-NEET/economically active group. This is suggestive of an ongoing 'scarring effect' due to previous NEET experiences.

Low status occupations

The occupational position of those in work is examined in this section. The occupational positions in 2011 and 2001 by 1991 NEET status are shown in Table 9. The outcome variable was based on National Statistics Socio-Economic Classification (NS-SEC), a derived variable from the census (see Appendix 3). It can be seen that 44% of people who were non-NEET in 2001 were in the Higher Professional and Lower Professional categories by 2011, compared to 20% of those who were NEET. Over half of NEET young people in 2001 were in the Semi-Routine or Routine occupational category by 2011 in contrast to only about one fifth of their non-NEET peers.

The distribution of occupation categories in 2011 by 1991 NEET status is similar to that by 2001 NEET status, which again shows the persistence of a negative effect of NEET status after a 20 year period. Ending up in low status occupations is another indicator of a scarring effect associated with NEETs.

Table 9 2011 National Statistics-Socio-Economic Classification categories by 2001 and 1991 NEET status

NS-SEC 2011	2001 cohort (Cohort 1)		1991 cohort (Cohort 2)
	non-NEET (%)	NEET (%)	non-NEET (%)	NEET (%)
Higher Professional	13	3	15	5
Lower Professional	31	17	31	19
Intermediate	20	12	16	13
Own-account workers	5	6	9	9
Lower-Tech	10	12	9	9
Semi-Routine	13	32	11	25
Routine	9	19	8	19
N	6,647	519	7,640	732

Source: SLS

Models were fitted to examine the likelihood of working in low status occupations as defined by NS-SEC categories: Semi-Routine and Routine occupations (Rose and O'Reilly, 1997). As expected, being NEET in 2001 is associated with a higher (more than twice as likely) risk of working in low status occupations by 2011 (Table 10).

The results show that the negative effect of NEET status was enduring even 20 years after the experience. Being NEET in 1991 was associated with a higher risk of working in low status occupations no matter whether the individual was or was not economically active in 2001. However, not being economically active in 2001 was also a significant predictor of low occupational status in 2011, even for those who were non-NEET in 1991.

Table 10 Odds ratio of low status occupations in 2011 from logistic regression

2001 cohort (Cohort 1)			1991 cohort (Cohort 2)
NEET status	Odds ratio	Significance level	NEET 1991 and economic activity 2001	Odds Ratio	Significance level
No	1
Yes	2.04	***
			Non-NEET 91 & active 2001	1
			Non-NEET 91 & inactive 2001	2.64	***
			NEET 91 & active 2001	1.79	***
			NEET 91 & inactive 2001	3.40	***
N	7,792			7654

*P<0.10 **p<0.05 ***p<0.01, Source: SLS

Limiting long-term illness

Whether an individual has a limiting long-term illness (LLTI) is a question that has been asked in the census since 1991. Table 11 shows LLTI status in 2011 by 2001 and 1991 NEET status. It is clear that being NEET in 2001 was associated with higher risks of reporting LLTI in 2011. About 18% of NEET people reported limiting long-term illness in 2011, while by contrast only 6% of non-NEET reported such a condition. There was an increase in the proportion of people who reported LLTI in 2011 for the 1991 cohort, which is likely associated with their older ages as this cohort was aged between 36 and 39 by 2011. Nevertheless, those who were NEET in 1991 were again more likely to report LLTI in 2011 than their non-NEET counterparts, with a rate more than double that among non-NEETs.

Proportions of people reporting LLTI in 2011 by extended NEET categories are presented in Table 12. As expected, the highest proportion reporting limiting long-term illness in 2011 were those who were permanently sick in 2001, the percentage being 67%. A similar proportion of those who were unemployed and those who were looking after family reported LLTI in 2011. Likewise, among people who reported being permanently sick in 1991, over 40% reported LLTI 20 years later.

Based on the 1991 cohort from Table 11 it can be estimated that for the Scotland population, around 6,000 (1265*24%*20) NEET young people would have limiting long-term illness by 2011. The scale of this long-term health effect of being NEET is substantial as this contributed to over 27% of those who reported having illness aged 36-39 while making up only 12% of the total in this age cohort.

Table 11 Limiting long-term illness in 2011 by 2001 and 1991 NEET status

NEET status, 2001	% no LLTI	% with LLTI	Total
Non-NEET	94	6	7,945
NEET	82	18	996
NEET status 1991	% no LLTI	% with LLTI	Total
non-NEET	91	9	8,980
NEET	76	24	1,265

Source: SLS

Table 12 Limiting long-term illness in 2011 by extended categories of NEET in 2001 and 1991

NEET status, 2001	% no LLTI	% with LLTI	Total
Non-NEET	94	6	7,945
Unemployed	85	15	597
Permanently Sick	33	67	61
Looking after home/family	85	15	338
NEET status, 1991	% no LLTI	% with LLTI	Total
Non-NEET	91	9	8,980
Unemployed	77	23	965
Permanently Sick	56	44	75
Looking after home/family	75	25	225

Source: SLS

The model results for the LLTI are presented in Table 13. These show an independent effect of being NEET in 2001 on the outcome, net of the other factors controlled in the model including having reported having LLTI in previous censuses. Young people who were NEET in 2001 were over 70% more likely than their non-NEET peers to report limiting long-term illness in 2011.

NEET status is also associated with LLTI 20 years later. Disengagement from employment and not being in education at the 1991 and 2001 censuses increases the odds of LLTI by four times in comparison with engagement in employment or education at both time points. Economic activity in 2001 did not fully nullify the effect of NEET experiences in 1991.

Table 13 Odds ratio of having limiting long-term illness in 2011 from logistic regression

2001 cohort (Cohort 1)			1991 cohort (Cohort 2)
NEET status	Odds ratio	Significance level	NEET 1991 and economic activity 2001	Odds Ratio	Significance level
No	1
Yes	1.74	***
			Non-NEET 91 & active 2001	1
			Non-NEET 91 & inactive 2001	3.73	***
			NEET 91 & active 2001	1.47	**
			NEET 91 & inactive 2001	4.06	***
N	7,917			8,073

*P<0.10 **p<0.05 ***p<0.01, Source: SLS

Hospital Admissions

This section looks at whether health outcomes measured as hospitalisation following an accident and emergency (A&E) visit are related to NEET status. The data were provided by Information Services Division (ISD). The outcomes reported in the analysis below identify whether an individual had at least one such hospital admission between 2001 and 2010.

An admission to hospital following a visit to A&E is a negative health outcome. A relationship between NEET status and subsequent hospital admission is therefore an indicator of a health disadvantage. A relationship of this variety might also be considered indicative of an attitude to risk or a lifestyle which includes greater risk. People who have more risky lifestyles are possibly more likely to end up in a hospital A&E, and this is especially so for young people. It is also likely that those admitted to hospital following a visit to A&E were suffering from a more severe health condition than those who merely attended A&E.

The second health outcome is related to hospital admission following a visit to A & E due to deliberate self-harm. This subgroup distinguishes those who are subject to considerable psychological stress from those who have experienced an accident, for example.

Table 14 tabulates the hospital admissions following an A&E visit for the 2001 cohort and the 1991 cohort separately. As with the LLTI outcome it is evident that those who were NEET are disproportionately likely to experience hospital admission following an A&E visit, on both outcomes. For example, for the 2001 NEET group, the percentage of those admitted to hospital following an A&E visit due to self-harm was over three times that of their non-NEET peers (7% vs 2%). For the 1991 and 2001 cohorts, the percentages of those admitted to hospital following an A&E visit were very close although the cohorts are 10 years apart in terms of age.

Table 14 Admitted to hospital following an accident & emergency (A&E) visit between 2001 and 2010 by 2001 and 1991 NEET status

NEET status	A & E		A & E due to self-harm
2001 cohort (Cohort 1)	%no admission	%any admission	%no admission	%any admission	Total
non-NEET	77	23	98	2	7,445
NEET	64	36	93	7	964
1991 cohort (Cohort 2)	%no admission	%any admission	%no admission	%any admission	N
non-NEET	78	22	99	1	7,582
NEET	66	34	95	5	1,006

Source: SLS and ISD

If the statistics based on the 2001 cohort from Table 14 are applied to the Scotland population, around 6,900 NEET young people have been admitted to hospital following an A&E visit at least once over 10 years between 2001 and 2010. If policy interventions were to be successful in eliminating NEET and its damaging co-determinants among young people, the number of visits to hospital following an A&E visit might be reduced by over 2,500 (964*36%*20-964*23%*20), a 36% reduction.

Tables 15 and 16 show modelling results for the hospital admissions outcomes. Again there is a strong significant association between being NEET and each of the outcomes for the 2001 cohort. NEET young people were 75% more likely than their non-NEET peers to be admitted to hospital following a visit to A&E. Also NEET individuals were more likely to be admitted to hospital following an A&E visit due to deliberate self-harm, with the odds more than double the odds for non-NEETs.

Table 15 Odds ratio (OR) of hospital admission following a visit to accident and emergency between 2001 and 2010 from logistic regression

2001 cohort (Cohort 1)			1991 cohort (Cohort 2)
NEET status	Odds ratio	Significance level	NEET 1991 and economic activity 2001	Odds Ratio	Significance level
No	1
Yes	1.75	***
			Non-NEET 91 & active 2001	1
			Non-NEET 91 & inactive 2001	1.46	***
			NEET 91 & active 2001	1.29	**
			NEET 91 & inactive 2001	1.83	***
N	7,917			8,073

*P<0.10 **p<0.05 ***p<0.01, Source: SLS and ISD

Table 16 Odds ratio of hospital admission following a visit to accident and emergency due to self-harm between 2001 and 2010 from logistic regression

2001 cohort (Cohort 1)			1991 cohort (Cohort 2)
NEET status	Odds ratio	Significance level	NEET 1991 and economic activity 2001	Odds Ratio	Significance level
No	1
Yes	2.23	***
			Non-NEET 91 & active 2001	1
			Non-NEET 91 & inactive 2001	2.92	***
			NEET 91 & active 2001	2.63	***
			NEET 91 & inactive 2001	8.23	***
N	7,917			8050

*P<0.10 **p<0.05 ***p<0.01, Source: SLS and ISD

From the model for the 1991 cohort we can see that those who were consistently outside employment and education in 1991 and 2001 were nearly twice as likely as those who were engaged in employment or education at both time points to be admitted to hospital following a visit to A&E. The odds differential was much higher with respect to hospital admission due to deliberate self-harm, with those who were NEET in 1991 and economically inactive in 2001 being over 8 times as likely to be hospitalised compared with those who were in employment or education at both time points.

Young adults moving either from non-NEET status in 1991 into being economically inactive in 2001 or NEET status in 1991 into being economically active in 2001 were also at a higher risk of hospital admission following a visit to A&E, both in general and due to self-harm. For example, the odds among young people who moved from being non-NEET in 1991 to being economically inactive in 2001 were 3 times those of the reference group (non-NEET and economically active in 2001) to have a hospital admission following A&E due to self-harm. This is suggestive of the mitigating effect of being in employment or education at some stage but, again, the effect of having been NEET in 1991 was not fully discounted.

NEET status at least 11 years prior to hospital admission predicted a greater likelihood of having at least one hospitalisation following a visit to A&E, suggesting an ongoing or accumulating lifestyle of risk-taking and stress significantly above that of the general population.

Depression and anxiety

Prescribing data from ISD were linked to the SLS. These data provided information on the prescription of antidepressants and antianxiety medications between 2009 and 2012. If an individual was given any prescription of such medications in the period, then the individual was regarded as having suffered from depression or anxiety.

The prescription of antidepressant and antianxiety drugs between 2009 and 2012 by NEET status in 2001 and 1991 is presented in Table 17. Overall, nearly half of young people who were NEET in 2001 were treated for depression or anxiety, while slightly over a quarter of non-NEETs had the same experience.

The incidence of depression and anxiety was slightly higher for the older 1991 cohort than for the 2001 cohort. Over half of those who were NEET in 1991 were prescribed antidepressant or antianxiety medication compared with one third of non-NEETs.

The scale of the effect of being NEET on the prescription of medication for depression and anxiety can be illustrated further based on the summary statistics of the 2001 cohort in Table 17. Overall more than 10,000 prescriptions (1102*48%*20) were dispensed to NEET young people between 2009 and 2012. If these NEET young people had the same level of depression or anxiety as their non-NEET peers, in total only about 6,000 (1102*27%*20) prescriptions would have been dispensed, a reduction of 40% for the NEET group. In other words, reducing the number in the NEET group and their higher mental health risk factors, would have a substantial impact on excess mental ill health in this young group.

Table 17 Prescription of antidepressant and antianxiety drugs between 2009 and 2012 by 2001 and 1991 NEET status

NEET status, 2001	% no	% yes	Total
Non-NEET	72	27	7,468
NEET	52	48	1,102
NEET status, 1991	% no	% yes	Total
Non-NEET	67	33	7,553
NEET	48	52	1,120

Source: SLS and ISD

Logistic regression models were fitted to investigate the relationships between NEET status and the risk of depression and anxiety.

Table 18 shows that being NEET in 2001 was associated with a higher risk of depression and anxiety around a decade later, indicating that this group was over 50% more likely to be treated for depression or anxiety than their counterparts who were non-NEET in 2001.

Having a limiting long-term illness in 2001 was also associated with a higher risk of depression and anxiety, and the size of the effect is similar to that of being NEET (see Appendix 2). This is not unexpected as it reflects the long-term nature of some mental health problems, as well as the association between chronic physical conditions and mental health.

From the model for the 1991 cohort, we can see that young adults who were disadvantaged in both 1991 and 2001 (NEET and economically inactive respectively) were 2.8 times as likely as their counterparts who were advantaged at both time points to be treated for depression or anxiety. Those who were non-NEET in 1991 but became economically inactive in 2001 also showed a higher risk of depression or anxiety compared with the reference group, with nearly double the odds of those who were non-NEET in 1991 and economically active in 2001. Young people who moved from NEET status in 1991 to become economically active in 2001 also had higher risks of depression or anxiety, again suggesting the long lasting negative effect of the NEET experience.

Overall, the results show that NEET experiences are associated with increased antidepressants and antianxiety treatment 10 years and 20 years later and that this effect is independent of a number of socio-economic factors at both individual and area levels.

Table 18 Odds ratio of being prescribed with antidepressant or antianxiety drugs between 2009 and 2012 from logistic regression

2001 cohort (Cohort 1)			1991 cohort (Cohort 2)
NEET status	Odds ratio	Significance level	NEET 1991 and economic activity 2001	Odds Ratio	Significance level
No	1
Yes	1.56	***
			Non-NEET 91 & active 2001	1
			Non-NEET 91 & inactive 2001	1.92	***
			NEET 91 & active 2001	1.56	***
			NEET 91 & inactive 2001	2.76	***
N	7,917			8073

*P<0.10 **p<0.05 ***p<0.01, Source: SLS and ISD

Drug misuse

Drug misuse usually refers to the illicit use of any opiate or benzodiazepine. Drug misuse data were collected from the Scottish Drug Misuse Database held by ISD. This database contains anonymised data on individuals at the point of first contact with a range of drug services, including non-statutory agencies and general practitioners. The data covers the period between 2006 and 2012. Drug misuse by 2001 and 1991 NEET status is presented in Table 19. Nearly one out of every twenty five of those who were NEET in 2001 were recorded as users of illicit substances. In contrast, only one out of one hundred non-NEET young people were recorded as having misused drugs in the same period. For the 1991 cohort, the results were similar: about 4% of NEETs had a record of drug use compared with 1% of non-NEETs.

Table 19 Drug misuse between 2006 and 2012 by NEET status in 2001 and 1991

NEET status, 2001	% no	% yes	Total
Non-NEET	99	1	7,945
NEET	96	4	996
NEET status, 1991	% no	% yes	Total
Non-NEET	99	1	8,980
NEET	96	4	1,265

Source: SLS and ISD

We used logistic regression to examine the relationship between NEET status and drug misuse (Table 20). Being NEET in 2001 is found to be associated with a higher risk of drug misuse between 2006 and 2012, with this NEET group being more than 2 times likely to use these drugs than their non-NEET counterparts.

For the 1991 cohort, analytical results showed that young adults who were excluded from employment and education in both 1991 and 2001 were more than 9 times as likely as their counterparts who were advantaged at both time points to be involved in drug misuse. For those who were non-NEET in 1991 but moved to being economically inactive in 2001 the risk of drug misuse was four times that of the reference group who were non-NEET in 1991 and economically active in 2001. Young people who had moved from NEET status in 1991 to being economically active in 2001 did not show a higher risk of drug misuse. This suggests that while being NEET has long-term negative effects, moving into employment substantially mitigates the risk of drug misuse.

Table 20 Odds ratio of drug misuse between 2006 and 2012 from logistic regression

2001 cohort (Cohort 1)			1991 cohort (Cohort 2)
NEET status	Odds ratio	Significance level	NEET 1991 and economic activity 2001	Odds Ratio	Significance level
No	1
Yes	2.47	***
			Non-NEET 91 & active 2001	1
			Non-NEET 91 & inactive 2001	3.91	***
			NEET 91 & active 2001	0.35
			NEET 91 & inactive 2001	9.18	***
N	7,917			8,073

*P<0.10 **p<0.05 ***p<0.01, Source: SLS and ISD

Risk factors of becoming NEET

This section describes the characteristics of Cohorts 3 and 4 and the results of modelling the risk of being NEET. Lists of the independent variables considered for Cohorts 3 and 4 are given in Appendices 5 and 6.

Key Findings - Risk Factors

There is strong evidence that being NEET is associated with the following demographic and socioeconomic factors. These risk factors seemed to be similar for young people growing-up in the 1990's compared to the 2000's.

• Risk factors are consistent across two cohorts and between males and females.
• Educational qualification is the most important factor. No qualifications increased the risk of being NEET by 6 times for males and 8 times for females in Cohort 3. No qualifications at SCQF level 5 or higher obtained by school stage S4 increase the risk of being NEET by 10 times for males and 7 times for females in Cohort 4.
• Other school factors are important including the proportion of time absent from school and the number of exclusions.
• Two factors are especially important for females: being an unpaid carer for more than 20 hours per week and teenage pregnancy.
• Household factors are also important. Living in a social renting household, living in a family that is not headed by a married couple, living in a household with no employed adults, having a large number of siblings all increased the risk of becoming NEET.
• Local NEET rate is an important factor for both cohorts and genders, with the risk of NEET increasing with local NEET rate.
• A risk score derived from the statistical modelling has potential to identify young people who are at risk of becoming NEET.

The 2001 NEETs

This sample consists of all SLS members aged 16-19 in 2001 who were also present in the 1991 census (Cohort 3).

There are slightly more females than males in the sample. The NEET rate is slightly higher for males than for females. The NEET rate is lowest for 16 year olds, increasing for ages 17 and 18, with age 19 being approximately the same as age 18. Younger ages are more likely to still be in education or training, the minimum age for leaving education is usually 16 in Scotland, and there will be further numbers leaving school education as age increases, with very few remaining beyond age 18. The overall NEET proportion is 12.5% which is consistent with previously published data.

Table 21 NEET status 2001 by gender and age for Cohort 3

	Non-NEET (%)	NEET (%)	Total (%)
Gender
Male	4,362 (86.9%)	659 (13.1%)	5,021 (49.3%)
Female	4,561 (88.2%)	613 (11.9%)	5,174 (50.8%)
Age in 2001
16	2,552 (93.0%)	191 (7.0%)	2,743 (26.9%)
17	2,294 (88.7%)	293 (11.3%)	2,587 (25.4%)
18	2,074 (83.9%)	399 (16.1%)	2,473 (24.3%)
19	2,003 (83.7%)	389 (16.3%)	2,392 (23.5%)

Source: SLS

The 1991 Census variables, teenage pregnancy before 2002, local NEET rate in 2001, highest educational qualification and unpaid carer in 2001 were considered as potential risk factors (see Appendix 4 for details). Modelling results are shown in tables 22 and 23.

The likelihood of being male NEET is increased if the childhood home was rented, all economically active adults in the childhood home were unemployed, the childhood household type was not 'married couple' or the young person had a higher number of siblings, a lower level of educational qualification, was a teenage parent or lived in an area with a high local NEET rate.

The likelihood of being female NEET is increased by the same factors although there is an extra factor (being an unpaid carer of at least 20 hours per week) and the household type does not quite reach significance. Not surprisingly, being a teenage parent is a more important factor for females than for males.

Teenage pregnancy is the most significant factor for females, however having a baby when a teenager is a relatively uncommon factor - only 6.7% of females had a teenage birth.

This contrasts with 15% of the sample that recorded no qualifications (or were missing these data) which increases the risk of NEET by approximately 6 and 8 times for males and females respectively compared with those having a degree, HNC or Higher qualification. For around 20% of SLS members who lived in an area where the local NEET rate is over 19.5%, the risk increases by approximately 2.5 times for males and 2 times for females. There are trends with both these variables with the risk of being NEET decreasing with higher levels of education and lower local NEET rates.

Table 22 Odds ratio of being NEET in 2001 from logistic regression for males

Variable	Odds ratio	Significance level	N
Tenure, 1991
Owner occupied	1.00		2,951
Social Renting	2.03	****	1,957
Private Renting	1.91	**	113
Employment of household, 1991
2 persons, both employed	1.00		2,288
1 person, employed	1.04		1,532
Some economically active persons are unemployed	1.18		367
All economically active persons are unemployed	1.89	****	296
No economically active & employed persons	1.26		538
Household type, 1991
Married couple	1.00		3,997
Other households	3.91	**	21
Lone parent	1.46	***	832
Cohabiting couples	1.66	**	171
Number of siblings, 1991
0	1.00		621
1	1.24		2,528
2	1.46	**	1,365
3 or more	1.90	****	507
Qualification, 2001
Degree\HNC\Higher	1.0		1,705
Standard grade	2.64	****	2,455
None\missing	6.06	****	861
Proportion NEET, 2001
Lowest proportion, <=6.5	1.0		1,163
>6.5 &= <13	1.58	***	1,687
>13 & =<19.5	1.85	****	1,229
Highest proportion, >19.5	2.52	****	942
Teenage father
No	1.0		4,938
Yes	1.75	**	83

N=5021, **P<0.05, ***P<0.01, ****P<0.001, Source: SLS

Table 23 Odds ratio of being NEET in 2001 from logistic regression for females

Variable	Odds ratio	Significance level	Number
Tenure, 1991
Owner occupied	1.00		3,064
Social Renting	1.60	****	2,015
Private Renting	1.54		95
Employment of household, 1991
2 persons, both employed	1.00		2,316
1 person, employed	1.45	***	1,613
Some economically active persons are unemployed	1.58	**	383
All economically active persons are unemployed	2.14	****	324
No economically active & employed persons	2.00	****	538
Number of siblings, 1991
0	1.00		575
1	1.36		2,552
2	1.56	**	1,459
3 or more	2.42	****	588
Being an unpaid carer, 20+ hours per week, 2001
No	1.00		5,120
Yes	2.42	**	54
Qualification, 2001
Degree\HNC\Higher	1.00		2,151
Standard grade	3.36	****	2,325
None	7.74	****	698
Proportion NEET, 2001
Lowest proportion, <=6.5	1.00		1,162
>6.5 &= <13	1.41		1,712
>13 & =<19.5	1.55	**	1,255
Highest proportion, >19.5	2.05	****	1,045
Teenage pregnancy
No	1.0		4,826
Yes	12.52	****	348

N=5174, **P<0.05, ***P<0.01, ****P<0.001, Source: SLS

This contrasts with 15% of the sample that recorded no qualifications (or were missing these data) which increases the risk of NEET by approximately 6 and 8 times for males and females respectively compared with those having a degree, HNC or Higher qualification. For around 20% of SLS members who lived in an area where the local NEET rate is over 19.5%, the risk increases by approximately 2.5 times for males and 2 times for females. There are trends with both these variables with the risk of being NEET decreasing with higher levels of education and lower local NEET rates.

A higher local NEET rate might reflect fewer opportunities for an individual actively engaged in trying to find employment. There may also be other factors at play which relate to geographies of NEET: there may be local cultures where it is more acceptable to be unemployed (because a relatively high proportion of the working age population are unemployed) leading to a lack of engagement, a peer effect (the number of close friends and siblings who are NEET), and areas where services are not sufficiently resourced to deal with the problem.

The parents' education levels measured in the 1991 Census recorded only whether or not a post 18 years old qualification was gained and so it is not possible to differentiate between those with no qualifications at all and those with a school level qualification. Most of these parents would have reached the age of 16 in the 1970s and early 1980s when post-18 qualifications were less common. In 1981, only 18% entered a higher education course by the age of 21 (Paterson, 1997). The parental qualification variables are therefore unlikely to show a strong relationship with NEET status.

Tenure and the number of employed adults in the childhood home have been included in these models. These variables are more important in predicting the risk of being NEET than other variables related to deprivation such as Carstairs quintile and overcrowding (see Statistical Methods for further details on the model selection procedure).

The 2011 NEETs

School Census Multivariate models

This sample consists of all SLS members aged 16-19 in 2011 with economic activity recorded in the 2011 Census (Cohort 4) and school census data available at school stage S4. The restriction of using only those with school stage S4 data is to avoid the problem described in Appendix 3, namely that in this sample we do not have complete school census data at all ages.

There are slightly more male than female SLS members in the cohort. The proportion of male NEETs is higher than the proportion of female NEETs. There are very few sample members aged 19 as most pupils of this age would have been beyond school stage S4 when the linked school census data begins in 2007. The proportion of NEETs increases with age, with younger ages more likely to still be in education or training.

Table 24 Proportion NEET for school census analysis by gender and age for Cohort 4

Variable	Non-NEET (%)	NEET (%)	Total (%)
Gender
Male	3,624 (86.8%)	549 (13.2%)	4,173 (51.6%)
Female	3,543 (90.7%)	613 (9.3%)	3,908 (48.4%)
Age in 2001
16	2,434 (93.7%)	163 (6.3%)	2,597 (32.1%)
17	2,283 (88.9%)	285 (11.1%)	2,568 (31.8%)
18	2,166 (84.6%)	394 (15.4%)	2,560 (31.7%)
19	284 (79.8%)	72 (20.2%)	356 (4.4%)

Source: SLS

The overall NEET rate is 11.3%, which is below the usually reported rate of about 13%. This is most likely due to the fact that we are only including those with S4 school census data who are in the younger age groups and attending state schools (school census data does not include independent schools).

The school census variables, teenage birth before 2010 and local NEET rate were considered as potential risk factors (see Appendix 5 for details).

The variables selected were the same for the male and female models, with the exception that the male model did not include teenage birth before 2010 or urban-rural area in 2011. The number of births in this particular sample is small - 19 to males (see Appendix 3 for further explanation) and is therefore unlikely to be significant. Although teenage birth is a significant risk factor for females, there were only 62 and so this only explains a small proportion of the female NEET group.

The number of passes at SCQF level 5 or higher obtained by school stage S4 is an important predictor for both males and females showing a clear trend. Males and females with no passes were 10 and 7 times respectively more likely to be NEET than those with at least 6 passes. The numbers with no passes at S4 are large: 32% of males and 26% of females. There appears to be benefit to having at least 6 passes over having 3-5 passes, suggesting that relatively small differences in early attainment might change the likelihood of being NEET.

The proportion of time absent from school also shows a clear trend. Males and females who were absent for at least 20% of the time are 4 and 7 times respectively more likely to be NEET than those absent for less than 5% of the time. The numbers absent for at least 20% of the time are sizeable: 12% of males and 13% of females. The group who were absent for at least 20% of the time are about twice as likely to have had a limiting long-term illness or have been an unpaid carer for a least 19 hours per week (data not shown). However these groups only account for 16% and 11% of the male and female high absentee group.

The proportion that were excluded from school is lower (14% and 5% for males and females respectively) than the proportion with no qualification or >20% time absent from school. This variable may capture those with more disruptive behaviour that also affects others (compared to school qualification and absences which only affect the individual themselves). Males and females who had at least 4 exclusions are about 3 times more likely to be NEET than those with no exclusions. It is intuitive that this group would be less likely to find employment or wish to remain in education.

Being registered for free school meals is predictive of becoming NEET with those registered being over 30% more likely to become NEET for males and females. The local NEET rate may be important, because in some areas, the labour market could be so competitive that pupils exhibiting none of the other risk factors cannot find employment.

Table 25 Odds ratio of being NEET in 2011 from logistic regression for males

School Census factors	Odds ratio	Significance level	N
Registered for free school meals
No	1.00		3,556
Yes	1.32	**	617
Proportion of time absent from school
<5%	1.00		1,424
>=5% & <10%	1.32		1,351
>=10% & <20%	2.34	****	898
>=20%	3.92	****	500
Number of exclusions
0	1.00		3,583
1	1.85	****	298
2-3	1.97	***	142
4 or more	3.30	****	150
Number of passes at SCQF level 5 or higher obtained by school stage S4
>=6	1.0		799
3 -5	2.54	***	745
1-2	5.19	****	1,308
0	10.36	****	1,321
Local NEET rate
<7.5%	1.00		863
>=7.5% & <13%	1.53	**	1,177
>=13% & <19%	1.77	***	1,068
>=19%	1.76	***	1,065

N=4, 173, **P<0.05, ***P<0.01, ****P<0.001, Source: SLS

Table 26 Odds ratio of being NEET in 2011 from logistic regression for females

School Census factors	Odds ratio	Significance level	N
Registered for free school meals
No	1.00		3,358
Yes	1.45	**	550
Proportion of time absent from school
<5%	1.00	--	1,141
>=5% & <10%	1.73	**	1,263
>=10% & <20%	3.46	****	983
>=20%	7.12	****	521
Number of exclusions
0	1.00	--	3,707
1	1.30		111
2-3	2.15	***	49
4 or more	2.66	***	41
Number of passes at SCQF level 5 or higher obtained by school stage S4
>=6	1.00	--	996
3 -5	2.61	***	846
1-2	3.12	***	1,057
0	7.27	****	1,009
Teenage birth before 2010
No	1.00	--	3,846
Yes	11.06	****	62
Local NEET rate
<7.5%	1.00	--	834
>=7.5% & <13%	1.40		1,080
>=13% & <19%	1.87	***	903
>=19%	2.02	***	1,091
Urban Rural
Rural	1.00	--	605
Urban	1.66	**	3303

N=3, 908, **P<0.05, ***P<0.01, ****P<0.001, Source: SLS

Risk scores for becoming NEET

It is an inefficient use of resources to apply an intervention to the whole population as this will include individuals who will not benefit from the intervention. Identification of a subset of the population which is most at risk and which can be targeted for intervention will be more cost-effective. For a model that predicts the probability of being NEET, we can consider the subgroup to be most at risk as those that have a predicted probability in excess of a certain cut-off value, for example we could look at the subgroup that has a predicted probability which is at least 90%.

The ability to correctly identify such a subgroup will depend on data quality and the number of important risk factors which can be measured and are available to users of the risk score. Since there is always measurement error, there will inevitably be some individuals predicted to be at low risk who would benefit from an intervention and some predicted to be high risk who would not benefit from the intervention (false negatives and false positives). As the cut-off value of the predicted probability increases, there is a trade-off between reducing the size of the intervention group and increasing the number of young people at risk of being NEET that do not receive the intervention (the false negatives).

The models shown in Tables 25 and 26 use school census risk factors, published local NEET rate and teenage birth for females. As these data are available to careers guidance officers, we use these models as our risk score models. The predicted probability takes a value between 0 and 1. This can also be thought of as a risk score ranging from 0 - 100 if we multiple the probability by 100. These probabilities have been grouped into risk groups (as shown in Table 27), and the observed number of NEETs and non-NEETs in our sample are reported for each category. Five risk groups have been chosen so that the middle category is centred on 13% - the Scottish NEET rate.

Table 27 Distribution of NEET and non-NEET by risk score group

Gender	Risk score group	NEET	% NEET	Non-NEET	% Non-NEET	Total
Male	0 - < 3	11	0.9	1,169	99.1	1,180
	3 - < 7.5	46	4.9	886	95.1	932
	7.5 - < 20	146	13.0	976	87.0	1,122
	20 < 40	187	29.9	439	70.1	626
	>=40	159	50.8	154	49.2	313
Female	0 - < 3	17	1.0	1,732	99.0	1,760
	3 - < 7.5	52	5.9	837	94.2	852
	7.5 - < 20	95	12.6	659	87.4	782
	20 < 40	85	28.2	217	71.9	320
	>=40	116	54.2	98	45.8	194

Source: SLS

For the data in Table 27, the overall male and female NEET proportions are 13% and 9% respectively. Table 27 shows that there is an increasing trend in the proportion who are NEET as the risk score increases with the highest risk group (scores of at least 40), having a NEET rate of approximately 51% and 54% for the males and females respectively.

If this group with scores of at least 40 were targeted for an intervention, only 29% and 32% (159/549 and 116/365) of male and female NEETs would receive an intervention. This is known as the sensitivity of a score. The majority of those at risk of becoming NEET would not be targeted.

To increase the proportion of NEETs that are targeted, the threshold would have to be reduced. If a threshold score of at least 20 is used (the two highest risk groups in Table 27) then the sensitivity is increased to 63% for males and 55% for females (346/549 and 201/365). The increase in sensitivity comes at the cost of about a threefold increase in the size of the intervention: for males, this increase is from 7.5% (313/4173) to 22.5% (939/4173) and for females, this increase is from 5.0% to 13.2%.

Our figures of 63% and 55% are considerably better than if interventions were given at random (i.e. 13% and 9%) but to target a higher proportion of those likely to be NEET would require an even larger intervention.

The better results for females may be because the model for the females includes more factors. It cannot be attributed only to the fact that having a birth before 2010 is a strong predictor for being female NEET because this is a relatively uncommon event in this cohort.

Britton et. al. (2011) described a score to measure at ages 13-14 years whether a young person would become NEET. There is some overlap between the factors in their model and those included in our model. For their selected threshold of low KS2 score plus at least 5 characteristics, the expected probability of being NEET is 20%. This is equivalent to our score being at least 20. A lower proportion of the population would be targeted - 8%. However no data is given about the number of NEETs and non-NEETs in this group. Pseudo R² statistics provided for their model suggest that their model has a poorer fit than ours (0.15 compared to 0.22 and 0.27 for our male and female models respectively). This may be because we have included variables measured later than age 13-14 such as school qualifications and teenage pregnancy.

It may be that the risk score given here can either be the starting point for developing a final risk score or used as a screening tool to clarify cases where there is doubt. We do not have access to other data that careers guidance officers have that would alter their estimation of risk - such as an individual being involved in crime, being in care, having health issues or having to cope with exceptional circumstances. The personality of an individual including their motivation and resilience could also be considered (Bynner and Parsons, 2002; Yates, et al 2010). The risk score from the model could therefore be modified by the user by including their knowledge of these other factors.

It should also be remembered that the model identifies only the most significant factors at a population level. It may be possible to build on this risk score using feedback from users to incorporate factors that are either relatively uncommon or not available to us by giving these factors a score, as agreed by the users.

Extension of model to include 2011 Census factors

The number of risk factors considered was extended to include 2001 Census factors, birth weight, prescription data, being an unpaid carer in 2011 Census and urban/rural status of local area in 2011 (see Appendix 6 for final models).

The sample size reduces slightly as not all individual records include 2001 Census data. The model for females includes the same factors shown in Table 26 with two new additions replacing being registered for free school meals: the school census variable of 'ever attended a special school' and the 2001 Census variable of the number of employed adults in the household. These two factors are of only marginal significance, the number of pupils ever attending a special school was small (32) and therefore the significance of this factor varies more with changes to the model and sample used.

The model for males includes the same factors shown in Table 25 with two new additions replacing being registered for free school meals: tenure and the number of siblings. The likelihood of being NEET increases if the childhood home was rented from a public landlord and with increasing numbers of siblings. The 2001 Census variables that are significant were also found to be significant in the analysis of the earlier cohort who were NEET in 2001.

Comparison of two cohorts

The two cohorts show consistent results, both showing that the level of school qualification, local NEET rate and teenage pregnancy are strong predictors for an individual being NEET. These factors are recorded at the time or at most a few years before NEET status being measured. The model for the later cohort considered a number of additional factors, and found that school census variables such as the number of exclusions and proportion of time absent were highly predictive. Some variables measured 10 years previously were also important: namely tenure, number of employed adults in the household, household type and number of siblings for the earlier cohort, and a subset of these was significant for the later cohort.

Geographies of NEET

We used Censuses from 1991, 2001 and 2011 to describe how the proportion of NEETs in Scotland varied by urban rural category, deprivation category and local authority. In 1991, students in full time education were enumerated at their home address whereas in 2001 and 2011 they were enumerated at their term time address. This may affect the comparisons shown here.

Local authority

There is considerable variation between the local authorities with the highest rates being more than two times the lowest rates, in 2011 the rates varied between 18.9% in West Dunbartonshire and 8.0% in Aberdeen City, in 2001 the rates varied between 19.0% in Glasgow and 6.5% in East Renfrewshire and in 1991 the rates varied between 25.2% in Glasgow and 7.0% in Aberdeenshire (Figure 1). In general, the males and females show similar distributions, charts showing the distribution by gender can be seen in Appendix 7. In 2011 and to a lesser extent 2001, the male NEET rate for Falkirk was higher than the female rate. This is most likely due to Polmont, the young offender institution for males. Since 2003 all male young offenders (approximately 400-500) have been placed here (previously there were another two institutions in Dumfries and Clackmannanshire). The male NEET rate for Falkirk is below the Scotland male rate if the figures are adjusted to remove the effect of Polmont.

Scottish Government proposed seven NEET hotspot areas in 2006: Glasgow, West Dunbartonshire, North Ayrshire, East Ayrshire, Clackmannanshire, Inverclyde and Dundee (Scottish Executive, 2006). These hotspots were selected based on several factors: percentage NEET (census data), benefit claimant rates, school leavers' destinations, attendance rates and exclusion rates. The local authorities with a NEET rate consistently more than 1% higher than the national average in 1991, 2001 and 2011 were West Dunbartonshire, North Lanarkshire, North Ayrshire, Inverclyde and Glasgow. These are all the NEET hotspot areas except for North Lanarkshire.

Carstairs deprivation

There is a striking trend of a decreasing proportion of 16-19 year olds that were NEET as deprivation decreases, seen for both males and females (Figure 2). The proportion that were NEET in the most deprived quintile is approximately four times that seen in the least deprived quintile. This pattern can be seen for Census 1991, 2001 and 2011 data however the trend is stronger in 1991 and 2011 when Scotland was experiencing a recession. More information on this measure of deprivation is available in Appendix 1.

Urban Rural Category

The six-fold categories are those defined by the Scottish government. They are:

1. Large urban areas
2. Other urban areas
3. Accessible small towns
4. Remote small towns
5. Accessible rural
6. Remote rural

Urban Rural categories were first produced after the 2001 Census. The 1991 graph has been produced using the earliest (2003) output area urban rural classifications to approximate the urban rural distribution in 1991. This graph is therefore to be used only as a general guide and is less accurate than the graphs for 2001 and 2011. It will be more usual for areas to become more urban over time due to new building and improved roads although this is not always the case. In addition, students were enumerated at their home address in 1991, which should result in the not NEET category being enumerated in the less urban areas and the most urban areas having an inflated rate compared to 2001 and 2011.

There is a trend for the proportion of young people that were NEET to decrease as the categories become more rural for both males and females (Figure 3). This trend appears to be less marked over time. One exception to this trend is the 'remote small towns' group which has the highest female rate in 1991 and second highest female rate after the most urban category in 2001. These rates exceed the corresponding male rates. This high rate is not so evident in the 2011 graph. Category 4 is the smallest category, accounting for less than 3% of the population however the small numbers do not explain this feature.

Figure 1 Proportion NEET by local authority, 1991, 2001 and 2011 Census

Source: National Records of Scotland, Scotland Census 1991, 2001, 2011

Figure 2 Proportion of NEET by gender and Carstairs deprivation quintile

Source: National Records of Scotland, Scottish Census 1991, 2001, 2011

Figure 3. Proportion NEET by gender and 6-fold urban rural classification

Source: National Records of Scotland, Scottish Census 1991, 2001, 2011