Updated 20.3.2000

Quality description

The quality of the material is studied in the following description. The sections concerning data collection and target population, unit and item non-response and the generalisability of the results are written by Matti Simpanen and the section on the correction of non-response by Ismo Teikari who has also designed the regression imputation model presented.

Data collection and target population

In addition to the target population, purposive selection was made of some (8) multi-activity enterprises with some other principal activity than training but with one or more significant training activities.

The names and addresses of the enterprises were gathered from Statistics Finland's Register of Enterprisesin spring 1997. The number of educational enterprises was 1,407. A postal inquiry was sent to them and to the eight purposively selected multi-activity enterprises mentioned above in May 1997.

The possible data collection methods were primarily telephone interviewing and a postal inquiry. The postal inquiry was clearly the least expensive method, although it was known that non-response would be higher with a postal inquiry than with telephone interviewing.

Until now, information has not been available on the training activities of educational enterprises in Finland. The main object of this survey was to describe the educational enterprises and the training activities they pursue. In spite of a higher risk of non-response, data collection was decided to be made by a postal inquiry because the material was considered sufficient for describing the basic activities of educational enterprises. The inquiry was to be timed so that the financial statements of the previous year had been completed and a new financial year was about to begin.

This quality examination concerns the target population. The above-mentioned multi-activity enterprises having a small proportion (eight enterprises) are not included in the examination of non-response.

Some of the educational enterprises to whom the postal inquiry was sent reported that they had had no training activities in 1996 (the enterprise had changed the sector, did not organise training or had ceased to operate). Some respondents informed that the enterprise was an educational institution of the educational system or a part of it.

Statistics Finland's Register of Enterprises has been updated since the selection of the addresses for the postal inquiry. After the register had been updated, it was found out that some of the enterprises had not (any more) been engaged in training activities in 1996.

There were thus 133 such enterprises not belonging to the target population. Hence, 1,274 enterprises formed the target population of educational enterprises in Finland.
 

1. Gross frame and target population, unit non-response and accepted responses
 

	Enterprises1)
Postal inquiry sent	1407
Not belonging to the target population (the enterprise was not engaged in educational activities or not in operation)	133
Target population	1274
Unit non-response:
* Did not return the questionnaire	668
* Returned a blank questionnaire or refused to reply	42
* Questionnaire data cannot be considered reliable/response only oral	19
* Address unknown	4
Accepted responses	541
Response rate of target population	42.5%

1) Multi-activity enterprises were not included. There were eight of them, three of which did not return the questionnaire. An acceptable response was obtained from five multi-activity enterprises.

A total of 541 accepted responses were returned, the response rate being thus 43 (a total of 684 enterprises out of 1,407 returned the questionnaire. Calculated from the target population, 48 per cent, or 608 enterprises returned the questionnaire.)

Unit non-response

Non-response is studied for the target population, for which reason the multi-activity enterprises are not included. There were clear differences between sectors in the response rates (cf. Table 2). For example, 57 per cent of the enterprises belonging to the sector private music and art schools responded, whereas an acceptable response was obtained from 40 per cent of driving schools (driver training).

Enterprises operating on the principles of business economics were more often non-respondents than the other enterprises. Their non-response rate was 59 per cent, while it was 43 per cent for non-profit enterprises.

There were no great differences in responding with respect to turnover. Most responses were received from enterprises with conceptually no turnover. In comparison by the size of enterprise, small enterprises appeared more reluctant to respond than larger ones.
 

2. Respondents in the target population of educational enterprises and non-response according to some background variables
 

Background variables	Accepted responses		Non-response		Target population
Sectors	N	%	N	%	N	%
Driving schools	203	40.0	305	60.0	508	100.0
Language schools and centres	60	44.4	75	55.6	135	100.0
Private music and art schools	50	54.3	42	45.7	92	100.0
In-service training centres	30	50.8	29	49.2	59	100.0
Private vocational and professional education institutions	2	33.3	4	66.7	6	100.0
Correspondence schools	2	66.7	1	33.3	3	100.0
Other educational institutions and units	194	41.2	277	58.8	471	100.0
Purpose of activity
Business enterprises	490	41.4	694	58.6	1184	100.0
Non-profit institutions	51	56.7	39	43.3	90	100.0
Size of enterprise
0-4 persons	465	41.2	665	58.8	1130	100.0
5-9 persons	38	48.7	40	51.3	78	100.0
More than 9 persons	18	52.9	16	47.1	34	100.0
Not known	20	62.5	12	37.5	32	100.0
Turnover (FIM)
Less than 50,000	44	44.0	56	56.0	100	100.0
50,000-299,000	113	38.8	178	61.2	291	100.0
300,000-699,000	132	43.6	171	56.4	303	100.0
More than 700,000	100	40.2	149	59.8	249	100.0
Not known / No conceptual turnover	152	45.9	179	54.1	331	100.0
Division of area
Province of Southern Finland	291	45.8	344	54.2	635	100.0
Province of Western Finland	154	37.7	255	62.3	409	100.0
Province of Eastern Finland	53	49.5	54	50.5	107	100.0
Province of Oulu	30	36.6	52	63.4	82	100.0
Province of Lapland	9	28.1	23	71.9	32	100.0
Province of Åland	4	44.4	5	55.6	9	100.0
Total	541	42.5	733	57.5	1274	100.0

Item non-response

The proportion of item non-response is rather small in the material. The proportion of missing data is largest when estimating the number of female participants (6 per cent). In-service training centres left the number of women unreported more often than others (23 per cent of respondent in-service training centres did not give the number of women).

The proportion of missing data in responses:
 

- Number of women missing	6,3 %	(34 enterprises)
- Number of courses missing	5,0 %	(27 enterprises)
- Number of participants missing	3,7 %	(20 enterprises)
- Number of hours taught missing	3,7 %	(20 enterprises)

Language schools and centres left out some information more often than others. One fifth of respondent language schools and centres did not report the number of courses. The proportion of women was missing from 14 per cent of the responses. Incomplete numbers of participants were given by 12 per cent of respondent language schools and centres.

Four per cent of respondent enterprises did not report the number of hours taught. Driving schools, for example, failed to do this more often than others. Some of the driving schools only gave the number of their theoretical lessons, partly leaving out the number of driving lessons.

Correction of non-response

When correcting unit and item non-response, the material was divided with respect to each variable to be imputed into two sets: the material of accepted responses and the material of missing responses. Logically empty cells were set to zero.

The regression model was applied to the material of accepted responses using auxiliary variables from Statistics Finland's Register of Enterprises, on which data was available for the whole target population. The estimates obtained from the model for the coefficients were placed into a model made for each unit and each missing value was estimated with the model.

Before non-response was corrected, 18 enterprises were removed from the material of accepted responses as they skewed the results of imputation. All eight enterprises with other activities than training skewed the results, because in-house training was more significant in them than external training. In addition, ten units were also omitted, some for the same reason and some because they differed too much in size (too large) or the responses were illogical.

The determining variables selected for the regression model were the sector, the legal form and the size of the enterprise (personnel). They proved to be the best determining variables for most of the variables to be imputed. The coefficients of determination in the model ranged between 20 to 80 per cent, which, although the coefficient of determination is not that significant here, indicates that considerably more variance can be gained by imputation than by replacing each missing value by an average.

Generalisability of the results

The reliability of questionnaire data is increased by imputation. Despite that, the numbers (participants, hours taught, the mutual proportions of subject contents) relating to the training activities of educational enterprises should be taken with some reservations. The survey does not give exact numbers.

The proportions between different sectors should not be compared too closely. It would be better to examine the data inside each sector. Enterprises from varying sectors differed in manners of responding. For example, the non-response rates of private music and art schools and in-service training centres were clearly smaller than those of driving schools. Therefore the results are more reliable in respect of them than for driving schools.

One reason for the growth in the non-response rate in some sectors is that the questionnaire was not necessarily suitable for describing the operation of those enterprises. This concerns driving schools, for example. When designing the questionnaire, the object was to ensure comparability as far as possible with Statistics Finland's surveys made on the educational institutions within the education system.

The description of the target population of educational enterprises (their number, turnover, size and location) is based on the data from Statistics Finland's Register of Enterprises. The material covers enterprises operating outside the regular educational system in the sector of education and training. The data concerning the target population are reliable and accurate in that respect.

It should, however, be borne in mind that relatively great changes take place in the field of businesses even within one year. The changes cannot always be explained unambiguously. An enterprise may change its location, sector, legal form, owner or employees. It can also be merged into another enterprise or divided into several enterprises.

The majority of educational enterprises are small businesses employing one or two persons. Some of these small enterprises may have different activities each year or even each month. Depending on the economic trends, for example, the principal activity may be training at one time, and consulting or training planning at another. And at times, the operation of an enterprise that started in the educational sector can be focused on a completely different sector.

Enterprise surveys for which information is collected retrospectively from enterprises are always faced with some changes in businesses that have not yet been recorded in the Register of Enterprises. These changes have an effect on surveys such as this. Although some of the outdated data can be checked by updating the register data, some may still remain in the material. This appears primarily as a factor increasing non-response. Enterprises that did not return the questionnaire and no longer operate in the educational sector raise the unit non-response rate (the proportion of those who did not return the questionnaire). These enterprises have to, however, be considered when correcting the data.

The questionnaire and instructions for responding in English are available from Statistics Finland.

See also Education and training in educational enterprises 1996:

Education and training in educational enterprises 1996: Summary