In this chapter we use both the AIM database and our survey. We also draw upon financial information provided by the Project QUEST accounting system. These data were given to us by the QUEST staff and we made no effort to audit the data (although we have no reason to doubt these data's accuracy).

We will use the AIM database for information on the pre-QUEST employment because these data are available for everyone and because the data were collected closer to the time of the actual experience and hence are likely to be more accurate than the retrospective questions in our survey. We will use both the AIM database and our survey to describe post-QUEST outcomes. As we will see, our survey shows somewhat higher post-QUEST wages than does the AIM database but the figures are close and the correlation between post-QUEST wages in the two data sources is .72. We will also describe other outcomes which are only available via our survey

Finally, the pre-QUEST wages have been adjusted to reflect inflation and are presented in 1995 dollars [1]. Hence the pre/post comparisons reflect real wage increases, not increases due to inflation.

Program Benefits

Post-QUEST Employment

The first question to be answered is what people do after leaving QUEST. The answer to this question is drawn from our survey because it asked detailed questions about activities other than work and because the AIM database only provides postQUEST placement information on positive terminations.

Table 4-1 provides data on the current (i.e. at the time of the survey) status of all QUEST leavers as well as leavers broken out by whether their termination was positive or negative (we apply the AIM positive/negative classification to the individuals in our survey). Nearly three quarters of the group as a whole and over three quarters of the positive terminators are working and another ten percent are continuing their education or training. By contrast about 10 percent of the entire population is looking for work as is 6 percent of the positive terminators. The unemployment rate, as conventionally defined, excludes people out of the labor force (i.e. people neither working nor looking). By this definition, the unemployment rate for all QUEST leavers is 12.1 percent and for positive terminations it is 7.1 percent. The results are less strong, as would be expected, for negative terminators.

Table 4-1
Current Status of QUEST Leavers

	All	Positive Termination	Negative Termination
Working For A Firm	71.5 %	77.5%	59.1%
Self-employed	1.5	1.5	1.5
Armed Forces	0.8	1.1	0
Looking for Work	10.1	6.2	18.2
At Home	5.8	3.3	13.6
School/Training	10.6	10.6	7.6

(Source: Survey)

As a test of the accuracy of these results we can see what percentage of positive terminators in the AIM data base have a placement wage reported, i.e. what percentage were working after termination. That percentage is 83.6 percent, which should be compared to the 80.0 percent reported for positive terminators in the survey (i.e. working for firm, self-employed and Armed Forces). Clearly the results are quite similar. As noted, AIM contains no information of the activities of positive terminators who are not working nor any data on the activity of negative terminators.

What was the status of these QUEST terminators prior to entering QUEST? That is, does the 70-plus percent who are now working represent an improvement over their prior situation? Our survey asked people what they were doing just before entering Project QUEST and the results are shown in Figure 4-1. The data in this figure represent only terminators (positive and negative) and hence are comparable to the prior table in this chapter.

Figure 4-1
Activity Just Prior to QUEST for People Who Are No Longer in QUEST
(modified from a graph format to a table for on-line viewing)

Working	48.50%
Looking	29.60%
Home	9.30%
School/Training	6.30%
Home/Trn/ School	6.30%

It is apparent that there has been a sharp increase in the fraction of people who work. Just prior to QUEST less than half of the participants had worked [2] whereas at the conclusion of their QUEST experience the fraction was over 70 percent.

Wage and Hour Changes For QUEST Participants

We have seen that participation in Project QUEST is associated with a substantial increase in the rate of employment. The next question concerns whether the nature of the employment improved.

Determining the post-QUEST employment situation of people who left the program is complicated because neither data source is ideal. The AIM data contains post-QUEST placement wages for positive terminators but contains no information on people who were negative terminations. Our survey does have information on participants who were both positive and negative terminations but there is likely to be modest upward bias because the survey slightly undersampled negative terminations (see our discussion of the survey and response rates in Appendix 2. [3]

In order to gain the most accurate possible sense of the impact of Project QUEST we present below three sets of results on average post-QUEST wages and hours:

1. Estimates drawn only from the AIM database. These estimates are only for positive terminations.
   
2. Estimates taken from our survey.
   
3. Estimates which blend AIM and our survey to obtain an average post-QUEST wage: for positive terminators we use the AIM database. For negative terminators we use the AIM positive termination wage and hours but multiply this by the ratio of the negative to positive termination wage (and negative to positive termination hours) taken from our survey. [4]

Table 4-2 below shows the wages (in 1995 dollars) and average hours per week worked of QUEST participants who have left the program at the last job they had before entering QUEST. The Table also provides data on our three alternative estimates of post-QUEST wages and hours.

Table 4-2

	Before QUEST (AIM Data)	After QUEST (AIM Data)	After QUEST (Survey Data)	Blended
Wage	$5.99	$7.82	$8.41	$7.35
Hours Per Week	32.4	37.5	39.3	35.6

(Sources: AIM and Survey Data)

It is quite apparent that, regardless of which post-QUEST source is used, the gains are impressive. Using our survey the gain was $2.42 an hour and with the blended data the average wage gain was $1.36 an hour. Similarly, our survey suggests an hours gain of 6.9 hours per week and in the blended data the increase per week was 3.2 hours. Given our procedure the blended estimates seem to us to be the lower bound for wage and hour gains while our survey results unadjusted seem to us to be the upper bound. [5]

The next question to address is the annual earnings change which occurs as a result of Project QUEST. This is important if we want to compare costs and benefits However, we do not observe many program leavers for over the course of a year after the program and so we have to make additional assumptions. We will begin by assuming that a work year is 50 weeks. Therefore, the natural procedure for calculating annual earnings pre- and post-QUEST would seem to be to multiply our estimate of hourly earnings times hours worked per week times 50 weeks.

This approach, however, is unfair to QUEST because one of the impacts of QUEST, as we have seen, is to increase the probability of even holding a job. To adjust for this we calculate the 50 week annual earnings (using the hourly wages and hours from above) and then for pre-QUEST discount that figure by .485 (the fraction of participants who were working prior to QUEST) and for post-QUEST discount the figure by .736 (the fraction of leavers who were working post-QUEST according to our survey). [6]

To summarize, we present two estimates of annual earnings gains. The first uses the blended post-QUEST wages and hours while the second uses the straightforward post-QUEST wages and hours from our survey. Both use the pre-QUEST inflation-adjusted wages from AIM. Both calculate annual earnings by multiplying hourly wages times hours worked per week times 50 weeks and then adjust for the fact that QUEST participants are more likely to work after leaving QUEST than they were prior to QUEST.

According to the first estimate (using the blended wages and hours), the adjusted annual earnings pre-QUEST was $4,706 and the annual earnings post-QUEST was $9,629 for a net gain of $4,923 per year. According to the second estimate, using the survey alone for post-QUEST outcomes, the adjusted annual post-QUEST earnings $12,163 and the annual pre-QUEST earnings is as before for an annual gain of $7457. (Remember: the figures in this paragraph are not actual earnings for a person who worked. They are the earnings of people who worked adjusted downwards by the probability of an individual working. Readers of this analysis who are not comfortable with the adjustments we make here should focus instead only upon the hourly wages and the hours per week discussed above. These also show substantial gains due to QUEST).

In short, our estimate is that participation in Project QUEST increased annual earnings for its participants by between $4,923 and $7,457. [7]

As a way of calibrating these results, it is worth noting that a recent national evaluation of the Job Training Partnership Act found that for disadvantaged adults the average earnings gain as a result of participation in JTPA was $900 per year. [8] Although the results are not fully comparable (the JTPA evaluation had a control group) the stronger performance of Project QUEST is nonetheless striking.

It is also worth noting that, according to the Texas Employment Commission the average hourly earnings of manufacturing employees in San Antonio rose 14 percent between September, 1992 and December, 1995. To obtain data on all employees [9], not just in manufacturing, we need to look at Bexar County and for these between the third quarter of 1992 and the second quarter of 1995 wages rose 8 percent. Both sets of earnings gains are clearly well below the increase experienced by QUEST participants.

More detailed analysis of wage patterns

In this section we ask how the changes in wages vary on the basis of different sub-groups.

In Table 4-3 (which uses blended wages and hours as post-QUEST estimates) we see that women gained substantially more than did men, largely because their pre-QUEST wage was lower. It is also striking that post-QUEST women achieve essential equality with men. However, it is important to note that men did nonetheless experience substantial gains.

Table 4-3

	Men	Women
Pre QUEST Wage	$6.43	$5.78
Pre QUEST Hours	32.4	31.7
Post QUEST Wage (Blended)	$7.48	$7.27
Post QUEST Hours (Blended)	35.3	36.0

(Source: AIM and Survey)

In Table 4-4 we break the outcomes down by whether or not the person was a positive or negative terminator. Not surprisingly, the gains are greater among the positive terminators although the negative terminators did nonetheless improve their situation.

Table 4-4

	Positive Termination	Negative Termination
Pre QUEST Wage (AIM)	$6.10	$5.69
Pre QUEST Hours (AIM)	32.7	32.0
Post-QUEST Wages (Blended)	$7.81	$6.41
Post QUEST Hours (Blended)	37.5	32.4

(Source: AIM and Survey)

It is reasonable to wonder how post-QUEST wages vary by occupational field. In Table 4-5 we show, for positive terminators, the post-QUEST wages for those occupations which had five or more terminations [10]; as well as for the remaining miscellaneous training fields. It is apparent that there is considerable variation with, in general, medical training doing considerably better than the other fields.

Table 4-5
Earnings for Positive Terminators, By Training Field

	Wage
Registered Nurse	$11.90
Plumber	8.69
Licensed Vocational Nurse	8.42
Medical Technician	8.13
Electronic Technician	7.45
Office System Technician	7.17
Facilities Administrator	6.98
Financial Services	6.89
Accounting Technician	6.64
Miscellaneous	6.62
Diesel Mechanic	6.50
Customer Service Representative	6.14
Health Unit Clerk	6.07

(Source:AIM)

These results do raise a warning flag: if we remove LVNs, RNs, and the various medical technician jobs from the sample, then the average post-QUEST wage falls from $7.82 to $6.84. If QUEST finds itself unable to rely on a large supply of well paying medical jobs, these patterns reveal that the program will face problems meeting its goals.

Finally we ran a statistical test to determine what personal characteristics were associated with higher post-QUEST wages. [11] The only considerations which proved to be important were whether the person had attended college (which increased post-QUEST hourly earnings by $.76). whether the person had an arrest record (which decreased post-QUEST hourly earnings by $1.24), and the person's age upon entering (each year of age increased hourly earnings of $.05).

Welfare Usage

There is considerable national concern about rates of welfare receipt and one goal of Project QUEST is to reduce the levels of welfare usage. In Figure 4-2 we show the fraction of Project QUEST participants who reported in the survey that they were receiving welfare or food stamps at the time they entered Project QUEST as well as the fraction who reported that they were receiving support subsequent to Project QUEST. [12] It is apparent that there has been a reduction in these rates and the savings associated with this reduction needed to be added to the earnings gains as a benefit of Project QUEST.

Figure 4-2
Welfare Recipiency
(modified from a graph format to a table for on-line viewing)

Percent Receiving Welfare or Food Stamps

When Applied	44.50%
Post Quest	33.70%

(Source: Survey)

While the reduction in welfare usage is impressive, the on-going rate of 34 percent is perhaps higher than might have been expected. The next two Figures describe the characteristics which distinguish between those who remain on welfare post-QUEST and those who do not. First, it is apparent that the rate of welfare usage is much higher, nearly twice as high, among negative terminators as among positive terminators. Second, the relationship between post-QUEST activity and welfare usage is quite striking. Among people who are working, only 18 percent are welfare or food stamp recipients whereas the rates are far higher among those looking for work, keeping house, or in school.

Figure 4-3
Receiving Welfare or Food Stamps Post QUEST
(modified from a graph format to a table for on-line viewing)

Positive Termination	25%
Negative Termination	43.5%

(Source: Survey)

Figure 4-4
Receiving Welfare or Food Stamps Post QUEST, by Activity
(modified from a graph format to a table for on-line viewing)

Working	18%
Looking	68%
At Home	80%
School/Training	64%

(Source: Survey)

How Participants View Their New Jobs

We asked several questions to probe the attitudes of program leavers toward their current jobs. The results of these questions are presented in Figures 4-5 to Figure 4-7.

Figure 4-5
Does Current Job Utilize Skills Learned in QUEST
(modified from a pie chart format to a table for on-line viewing)

To A Great Extent	49%
To Some Extent	23%
Very Little	8%
Not At All	20%

(Source: Survey)

Figure 4-6
Does the Current Job Have a Good Future in Terms of Pay and Promotion
(modified from a pie chart format to a table for on-line viewing)

Yes, Definitely	59%
Yes, Somewhat	22%
No	17%
Not Sure	2%

(Source: Survey)

Figure 4-7
Do You Plan to Stay in the Current Job?
(modified from a pie chart format to a table for on-line viewing)

Yes, Indefinitely	35%
Yes, For Next Several Years	39%
Will Leave Within a Year	18%
Not Sure	8%

(Source: Survey)

First, we wanted to know if the jobs in fact utilized the skills the participants learned as a result of Project QUEST and it is clear that for most program leavers the answer is yes.[13] There is also a strong relationship (not shown in the Figures) between whether the skills are utilized and wages: among those who said the skills were utilized to a great extent the average wage was over $2 higher per hour than those who reported that the skills were not at all utilized.

Figure 4-6 asks people's assessments of whether the job has good future prospects in terms of pay increases and promotions and the next Figure asks if people intend to stay at the job. The strong majority believe that their job has a future and most people plan to stay, at least for several years. There is, unsurprisingly, a strong relationship between wages and the answers to these questions.

Other Outcomes

Beyond purely economic outcomes for participants, Project QUEST hopes to improve the prospects of their families. This is a difficult outcome to measure but we did ask in the survey whether the family or children of the participants have shown more interest in education as a result of the experience of the participants in Project QUEST. The reply to this question is shown in Figure 4-8 and it is clear that, according to this measure, Project QUEST has had a substantial impact upon the families of participants. It is also worth noting that, in response to another question 62.6 percent of the participants said that their involvement in Project QUEST had made it easier for them to think about going back to school themselves and 74.3 percent said that they had definite plans to do so.

Figure 4-8
Have Family or Children Shown More Interest in Education Due to Project QUEST?
(modified from a pie chart format to a table for on-line viewing)

Yes, Very Much	57%
Yes, Somewhat	16%
No	27%

(Source: Survey)

How Participants View Project QUEST

We consider one further gauge of the impacts of Project QUEST: the satisfaction of QUESTers with their program. We asked in our survey about participants' views of a number of aspects of QUEST, ranging from occupational choice to their relationships with counselors.

Eighty-two percent of QUEST participants were satisfied or very satisfied with the first occupation they were matched with for training. There are three primary reasons those dissatisfied give for their reaction: that the pay in their training occupation is too low, that it offers limited promotional potential, or that there are too few job openings available after graduation. Only 38 percent of those who were dissatisfied ultimately changed career tracks. (Overall 15 percent of QUESTers shifted from one trailing stream to another.) The level of satisfaction declined in the second career tracks; only 71 percent were satisfied or very satisfied with their new occupational track. The root of most of the dissatisfaction of this group is in inadequate pay opportunities in the career they were being trained for. In the end, 81 percent of QUESTers are satisfied with their final training occupation.

We've documented in earlier chapters the important role counselors play in Project QUEST. Participants' perceptions reinforce this. Only 11 percent of QUESTers rate their counselor as not important to their staying in the program, as Figure 4-9 shows below. (Chapter 5 below contains a fuller discussion of program retention.)

Figure 4-9
Importance of Counselor to Participants' Completion of Program
(modified from a pie chart format to a table for on-line viewing)

Very Important	60%
Important	17%
Fairly Important	11%
Not Important	12%

(Source: Survey)

QUEST participants appreciate the information on program details and requirements that counselors provide. They also value the aid their counselors give in gaining financial assistance and their general emotional support. Almost 90 percent of QUESTers rate their counselor's overall supportiveness as good. In general, 96 percent of QUESTers state that the financial assistance they received in the program was an important determinant of their ability to complete the training.

Our survey also provides several measures of QUEST participants' perspectives on the training itself. Virtually all, ninety-one percent, view their training program as high in quality. Approximately one quarter of participants in QUEST did raise one concern: they feel that the pace of training is too fast. This concern was echoed by QUESTers in our interviews and focus groups.

Substitution

One concern about Project QUEST, indeed about all employment and training programs, is whether any success they have is due to simply re-shuffling people. In the extreme, one might wonder whether the successful placement of a Project QUEST graduate simply comes at the expense of someone else who would have otherwise gotten the job. If this were true then while the program would have positive benefits for the people involved, from a societal viewpoint there would be no gains.

Virtually no evaluation of any employment and training program has addressed this issue because it is impossible to know what would have happened. Even a control group methodology cannot address this because most employment and training program are too small relative to the overall labor market to have any impact upon the outcome for the control group. This study of Project QUEST is no different than any other evaluation in that it does not purport to resolve this issue. We were, however, sensitive to it and did ask some questions in our interviews aimed at getting at the question. We have several points to make.

• An important part of the Project QUEST story is that hospitals had been hiring nurses from outside the San Antonio area, indeed from as far away as Canada and the Philippines. To the extent that Project QUEST substituted local people for these "outsiders" there is a clear net gain from the viewpoint of San Antonio and even Texas as a whole.
  
• In some instances QUEST students were given priority over others for slots in community college courses and in these cases there may be have been some substitution. However, it is also true that several community colleges added new classes to accommodate Project QUEST students and on these occasions there was not a simple substitution of Project QUEST students for other potential community college students.
  
• Some substitution is inevitable. However, even this may be acceptable if it leads to better equity outcomes. There is evidence in some of our interviews that Project QUEST students in RN programs come from more economically disadvantaged backgrounds than do the typical RN students. This is less true for LVN, diesel and OST students, and we have no comparative information for the other occupations Project QUEST trained for. We do, however, have strong evidence that Project QUEST students faced serious barriers to success in the labor market (this evidence was presented above).

Summary of Benefits

Project QUEST has very substantially increased the hourly wages, the hours worked and hence the annual earnings of its participants. The impact on wages is particularly notable since other job training programs often achieve their gains by increasing hours worked but not wages.

Project QUEST has also had other benefits. It has reduced the usage of welfare and food stamps and has increased chances that participants and their family members will pursue additional education. Project QUEST participants are pleased with these outcomes, as judged by their view that their jobs have a future and their intention to remain on these jobs. Most participants are also very satisfied with their training, counseling and financial support.

Program Costs

In this section we describe the costs of Project QUEST using data provided to us from the program's accounting system.

The natural approach to examining costs is to look for a per participant figure. However, arriving at this figure is complicated because of the varying lengths of time individuals stay in the program. Hence in the tables below we provide two alternative cost bases:

1. The costs per individual.
   
2. The costs per month. This enables us to control for the fact that different people spend different lengths of time in the program and also permits us to examine trends in costs over time.

We also look at several groups of people:

1. All participants. This provides the broadest view of the program.
   
2. Cost data broken down by outcome (positive and negative termination). This enables us to control for the fact that some people left the program early.

The cost data broken down into these three groups is presented in Table 4-6.

Table 4-6
Total Cost Data Per Participant

	All Participants	Positive Terminations	Negative Terminations
Tuition and Fees	1646	1854	1198
Support Payments	4012	4575	2885
Counselors and Occupational Analysts Salaries	2328	2454	2031
Administrative Costs	2091	2183	1877
Total Costs	$10,077	$11,066	$7,991

Several conclusions stand out:

1. The average cost for all the participants is $10,077. For positive terminations (who spent an average of 17.7 months in the program) the figure is $11,066 and for negative terminations (who spent an average of 14.6 months) the figure is $7,991.

2. The breakdown by cost categories is
   • Direct payment for participants (tuition, child care, food, books, transportation) = 56.1% of total costs

   • Payment to staff who work directly with participants (counselors and occupational analysts)=23.1% of total costs

   • Indirect costs for administrative staff and expenses = 20.7% of total costs.
     **Of these indirect costs 19% are due to one time start up expenses which presumably will not reoccur.

In Table 4-7 we present monthly cost data for two periods of time, the initial start up period (until June 30, 1993) and the period since then. It is apparent that there has been some decline in costs over time. If we look at all participants but apply the costs figures from the later period then the average cost per participant falls from $10,077 to $9,643. These savings come from reductions in support payments to the participants.

Table 4-7
Costs Per Participant Month

	All	Enrolled Through 06/30/93	Enrolled Since 07/01/93
Tuition and Fees	98	104	96
Support Payments	240	271	205
Counselors and Occupational Analysts Salaries	138	138	138
Administrative Costs	125	125	125
Total	$601	$638	$564

While in Project QUEST participants frequently receive support from other sources. The most important of these are welfare (AFDC), food stamps, and housing subsidies for people in the San Antonio Housing Authority. In our view these should not be counted as program costs since the participants would, in all likelihood, have received these transfers even if they were not enrolled in Project QUEST. On the other hand, many participants receive Pell Grants which support part of their training. There are no hard data in the AIM system on these grants however Project QUEST counselors estimate that about 65 percent of the participants receive grants which average about $1,000 per person. These costs could reasonably be included as part of program expenses.

Discussion

From any viewpoint Project QUEST is an expensive program. The high costs come from two sources: the program is long and it provides considerable support to its participants. As already noted, the typical JTPA program is far shorter and typically provides very little support.

The correct way to assess Project QUEST's costs is ask about the costs relative to the benefits. From the perspective of an individual, considerable research suggests that each year spent in school increases earnings by between 7 and 10 percent. The data we have presented suggests that Project QUEST's return to an individual is far greater.

Another perspective is to ask whether the program justifies the public investment. The typical Project QUEST participant is in the program for 17 months which is essentially equivalent to two years of public school education. Two year of public school education costs roughly the same as 17 months of Project QUEST. Society generally regards the two years of education as worth the expense because the payoff is high. In the same way we need to ask about the payoff to Project QUEST relative to its expenses.

The costs of that education are the direct costs associated with schooling and the indirect costs of earnings which are foregone while the person is in school. We can apply the same set of costs to evaluating Project QUEST.

If we look purely at program costs and employ the most conservative blended estimates presented earlier, Project QUEST returns 100 percent of its investment in two years. Even if we add earnings foregone (i.e. pre-QUEST annual earnings adjusted by the probability of working) while in training the full payoff occurs in 3 years. Rates of return using our survey are obviously even better. These are rates of return which exceed virtually all other employment and training programs.

Chapter 5: Staying In & Dropping Out

"Being thrust in a full-time college environment is quite a jolt. Being with younger people. Having never been in college before."

"You want to like what you are doing, but I couldn't like it. But I stayed in cause its free."

"I didn't really want to get into medicine, but they told me that was the only area they were hiring in."

"I had to keep going back into remedial and it kept pushing back the time the program took."

(Focus Group Participants)

We have seen that the quality of the outcomes experienced by Project QUEST participants varies considerably depending upon whether they were positive or negative terminators from the program. Of the people who have left Project QUEST 28.9 percent are negative terminators and 71.1 percent terminated positively. A particularly unfortunate aspect of the negative termination process is that the average negative terminator had been in the program for 14.6 months prior to leaving (compared to 17.7 months for positive terminators). In this section we seek to learn more about what distinguishes negative from positive terminators.

In Figure 5-1 we show the status of participants who have left QUEST according to which year they entered the program. It is apparent that there is a downward trend in negative terminations, which is a good sign and which also lends some credence to the view that QUEST was handicapped by an early push to admit some unqualified people.

Figure 5-1
Percentage Positive and Negative Terminations
By Date of Entry to Program
(modified from a graph format to a table for on-line viewing)

	1993	1994
Positive Terminations	68.9	76.9
Negative Terminations	31.1	23

(Source: AIM)

However, a larger attrition rate in the first months of any new start-up program is to be expected and this is an alternative explanation for the difference between cohorts. In addition, it is also worth noting that the difference between the negative termination rates for people who entered in 1993 and 1994 is not that great. Given these two considerations (the start-up nature of the program in 1993 and the small differential) the early push story should not be over-emphasized.

We noted earlier that the average negative terminator had been with the program for some time. In Table 5-1 we show the distribution of months of program participation for people who were negative terminators. It is apparent that the bulk of leaving occurs late in the program experience. One implication is that the problem is not patently bad matches between program and participant which are immediately discovered by one party or the other. Rather it would appear that for most negative terminators something happens late in their program experience which leads them to leave.

Table 5-1
Months In Program for Negative Terminators

Less than 3	6.6%
Between 3 and 6	13.7
More than 6 and Less than 12	24.2
12 or More	55.5

(Source: AIM)

As a first step in understanding this problem, in our survey we directly asked negative terminators why they had left. Table 5-2 gives their responses. It is apparent that personal and financial problems are the most important reason they gave, followed by dislike of the occupation they were trained for and attractive outside opportunities (a job or education). Although it is always risky to rely on self-assessments there is some objective verification of these responses: people who said they left because of a job offer in fact have wages higher than those of other negative terminators. Those who left because of education or training are indeed much more likely than others to be in an education or training program.[1]

Table 5-2
Reasons for Leaving the Program

	Very Important	Somewhat Important
Financial Problems	44.7%	13.8%
Personal Problems	39.4	10.5
Didn't Like Occupation	27.8	5.3
Found A Good Job	26.2	9.6
Wanted Different Educational/Training Program	22.3	6.0
Didn't Get Along With QUEST Staff	11.8	9.2
Program Lasted Too Long	9.9	9.9
Program Too Difficult	9.2	9.8
Family Wanted Me to Leave	6.0	3.3
Didn't Get Along With Trainers	2.6	3.9

(Source: Survey)

The central story, however, to be learned from this Table is that personal and financial problems are at the heart of participants' explanations of why they left early.

We explored in depth two other sets of explanations about negative termination. First, we examined the statistical relationship between personal characteristics and the probability of negative termination. Then we looked at the impact of program characteristics and the participants' prior relationship to COPS/Metro Alliance.

To examine the role of personal characteristics we used the AIM database to estimate a statistical relationship [2] between the probability of dropping out and a set of personal characteristics:

• sex
• college attendance
• arrest record
• military service prior to QUEST
• whether the person entered QUEST via ACCD
• welfare receipiency prior to QUEST
• handicap status
• race
• single parent status prior to QUEST
• number of dependents
• how many years before QUEST the person had last worked
• The wage of the person in the job held prior to QUEST [3]

In this multi-variable statistical analysis the only considerations which proved statistically significant were whether the person had attended any college, whether the person came to QUEST through ACCD, whether the person had received welfare prior to entering QUEST and the ethnicity of the participant. The mean values for each of these variables is shown in Table 5-3. This Table shows the means for positive terminators, negative terminators, and (as a matter of interest) for people currently enrolled in the program.

Table 5-3
Distribution of Statistically Significant Personal Characteristics
By Termination Status

	Still Enrolled	Positive Termination	Negative Termination
Black	.13	.09	.14
Hispanic	.61	.71	.73
ACCD	.11	.13	.09
Welfare	.61	.47	.57
College	.54	.45	.37
Pre-Quest Wage [4]	$5.61	$S.45	$4.90

(Source: AIM)

Finally, we used some of the questions in our survey to examine what aspects of the program seem associated with termination status. There is a danger in this analysis: people are reporting about the program and there may be a bias in the answers. That is, negative terminator who is unhappy about this outcome may seek to justify it by being critical of some aspect of the program even though by an objective measure the program was doing well on that dimension. Nonetheless, we believe that there are insights to be gained by using these data.

Once again, we estimated a statistical relationship between a series of variables and termination status. The variables we included in the analysis were:

• The significant variables from the earlier analysis
• whether the person had been active in COPS prior to QUEST (i.e. had been very familiar with COPS and/or had participated in activities)
• the same question with respect to Metro Alliance
• how often the person saw his/her counselor (at least once a week versus less than this)
• whether the person had held a job just prior to entering QUEST
• how satisfied the person was with the occupational field they were assigned
• whether the person had a personal or financial problem during their QUEST experience
• if they had such a problem whether the counselor had been helpful
• whether the person worked while in QUEST
• whether the person was satisfied with the length of the program
• whether the person was satisfied with the pace of the program
• whether the person was satisfied with the quality of the training
• whether the person was satisfied with the quality of the counseling

The variables which proved statistically significant were satisfaction with training field, the frequency of seeing a counselor, whether the person had a personal or financial problem during QUEST, if the counselor helped resolve that problem, and overall satisfaction with counseling. We show the means of these variables in Table 5-4.

Table 5-4
Assessments of Program Experience
By Termination Status: Statistically Significant Predictors

	Still Enrolled	Positive Termination	Negative Termination
Satisfied with Field	.89	.83	.59
Saw Counselor	.89	.87	.77
Had Problems	.63	.65	.73
Counselor Helped With Problems	.53	.45	.34
Overall Satisfaction with Counseling	.92	.87	.67
Training Quality	.92	.92	.80

It is very clear from these results that the counselors and the quality of the counseling play a critical role in the process. This is consistent with our interviews, observations, and focus groups. In addition, it seems to be important to assure as much congruency in occupational choice as possible (given available opening and people's abilities). It is also worth noting here that QUEST seems to be improving ova time along the dimensions cited in this Table.

Finally, the rate of negative termination varies by training area. In Table 5-5 we show (for training fields with 10 or more leavers) the percentage of leavers who were negative laminators. This may reflect some combination of the quality of the training, the career prospects of the field, or the character of the people placed in the training area.

Table 5-5

	Percent of Leavers Who Were Negative Terminators
Aircraft Technician	46
Office Systems Technician	45
Customer Service Representative	44
Health Unit Coordinator	43
Accounting Technician	40
Electronic Technician	36
Diesel Mechanic	32
Medical Technician	29
Registered Nurse	25
Licensed Vocational Nurse	21
Facilities Administrator	20
Plumbing	20
Financial Services Representative	16

(Source: AIM)

Endnotes

Chapter 4

[1] We used the Bureau of Labor Statistics Consumer Price Index for Dallas, Texas since we were unable to obtain a CPI series for San Antonio.

[2] This does not mean that they never had worked, rather the question is about their situation just prior to entering Quest. According to the AIM database, 85 participants had never worked and another 3 persons reporting earning less than $2 per hour. These 88 people were not included in our estimate of the pre-Quest wage.

[3] Our survey is also likely to give higher figures because it was taken after some graduates had been in the labor force for some time whereas the AIM data is for the initial placement wage.

[4] For example, in our survey negative terminators earned $7.08 an hour post-Quest while positive terminators earned $8.65. The ratio of these figures is .82 and we multiply this times the positive terminators in the AIM database ($7.82) to arrive at an estimated average post-Quest wage for negative terminators of $6.41. The overall average post-Quest wage is then the weighted average (the weights being the numbers of positive and negative terminators) using the positive terminator AIM wage for positive terminators and the "adjusted" AIM wage for negative terminators. We refer to this as the "blended" wage. We use a similar procedure for hours, concluding that for negative terminators the average hours post-Quest is 32.4.

[5] There are 88 people who, according to AIM, never worked and they are ignored in calculating the pre-Quest wage. If they are included and assigned a zero wage then the pre-Quest (inflation adjusted) average wage falls to $5.37 and the gains are even more impressive.

[6] It is important to know whether these gains are sustained over time. We have no long term data on the participants and hence cannot answer this question. However, there is thirty, sixty, and ninety day (from placement) wage data available and these show the wage patterns holding steady.

[7] Recall, these gains are in real dollars: inflation is not a factor. It is important, however, to repeat the qualifications noted earlier in the report. We do not know how these people would have done in the absence of Project Quest because we lack a control group. Also, people typically enter training programs when their circumstances are at a temporary low point. However, as our earlier discussion of barriers should have demonstrated, the Project Quest participants are a group who, on the face of it, would have had substantial problems doing well on their own in the labor market.

[8] US Department of Labor, What's Working (And What's Not), Government Printing Office, January,1995, p. 29

[9] The data include all employees covered by Unemployment Insurance. These figures do not control for inflation.

[10] There are in fact a number of different medical technician fields-surgical tech., respiratory tech., etc.- which we group together.

[11] We estimated a regression in which the dependent variable was the post-Quest wage from the AIM database and the independent variables were the pre-Quest wage, age when entered Quest, sex, how many years since the person had last worked, whether the person had any hours of college credit, whether the person had handicapped status, if the person had been in the military, if the person had an arrest record, the person's ethnic status (broken down by white, black, and Hispanic), if the person had entered Quest through ACCD, the number of very young dependents the person had, and the number of older dependents the person had.

[12] We use our survey estimates because they provide pre and post data. However, for pre-Quest welfare receipiency AIM reports 47.1% ehich is essentially identical to our survey.

[13] 56% of positive terminators reported that they definitely use their Quest skills in their jobs, and 22% said "yes, somewhat."

Chapter 5

[1] One motivation which we could not directly check is that some people temporarily remained in the program simply because it was free and even provided some support The focus groups indicate that this was not uncommon.

[2] We estimated a logit probability model

[3]We set this variable equal to zero if the person had never worked before

[4] This is inflation adjusted. However, because of the adjustment noted in the prior footnote, it is not consistent with other pre-Quest wages we have cited.

Index

Chapter 1: Introduction and Summary of Results
Chapter 2: History and Structure of Project QUEST
Chapter 3: Who Are The QUESTers?
Chapter 4: The Benefits and Costs of Project QUEST
Chapter 5: Staying In and Dropping Out
Chapter 6: Institutional Change
Chapter 7: Choices and Issues
Appendicies & References