Parents’ Child Care Preferences: a Conjoint Analysis

(1)

Master’s Thesis Small Business & Entrepreneurship

Parents’ Child Care Preferences: a Conjoint Analysis

January 2012

Kalle Smit, s1413880

Supervisor: C. Lutz

Faculty of Economics & Business University of Groningen

Abstract

(2)

(3)

Acknowledgement

(4)

(5)

Management summary

Context and research focus

The Dutch child care sector has grown rapidly in the last decade. However, recently growth has diminished and in the near future the market will be saturated. Therefore, it will be harder for child care companies to obtain new customers. Thus, knowing the parents’ child care preferences becomes even more important. It is widely believed that child care is a price inelastic service and parents are expected to base child care choice on quality attributes and not so much on price. Existing literature does only make clear how parents value single attributes of child care. However, the choice of a child care provider is based on many attributes together. To obtain a better understanding about parents’ child care preferences in the Netherlands, parents were asked to complete a questionnaire about child care attributes. This questionnaire will include questions for a conjoint analysis. This analysis can help determine the relative importance of the attributes for choosing a child care provider.

Research questions:

 Which attributes are key for attracting new customers for a child care company? Sub questions:

 How do customer search for child care companies?

 What are important child care attributes from a customer’s point of view?  How do customers value their currently used child care provider

Desk research

(6)

Field research

The selection of a child care company can be separated into two steps. Firstly, the search for child care and secondly, the choice of child care. First the parents’ search behavior will be explored. In the second step it will be determined what parents find important attributes of child care and therefore which child care provider they have the highest preference for. The field research was performed by gathering data from an online questionnaire. Randomly selected parents were asked to fill out this questionnaire. The parents were asked about:

 Their demographic profile

 How they search for child care providers

 To give importance scores to 16 child care attributes  How they rate their current child care provider

 To give scores to 10 conjoint profiles, which consist of different combinations of 5 child care attributes

The final part of the survey will be a conjoint analysis to determine how much value parents place on the following attributes: quality, location, educational development, opening hours, and price. Results & conclusions

(7)

1 Introduction

Context

Until recently, the Dutch child care market has been subject to high growth numbers. In 2010 more than 500,000 children were looked after by formal child care providers (CBS, 2011c). According to Wegemans and Scholte (2011), the market will saturate in the next 2-3 years. This will result in more fierce competition between child care providers and thus it will become harder to grow for these providers. Therefore, it becomes more important to actively attract customers and deliver them the service they require. To attract new customers and keep them satisfied, it is vital to have detailed knowledge about how parents search and select child care. Currently, there is an abundant amount of literature on parents’ search behavior and selection criteria for child care. However, most research has been performed in the United States and child care in the Netherlands is in several respects different to child care in the United States. Furthermore, most literature focuses on day care and especially guest parent child care is most of the time neglected. Special attention is paid to this mode of childcare, to obtain a better understanding of how this mode is valued in comparison to the other modes of child care. Finally, the most important reason for conduction this research, is that previous research did not give insights what combination of attributes is most influential on parent’s choice of a child care provider. Existing literature only revealed importance scores of single attributes, which cannot tell the attributes’ relative importance and utility. A conjoint analysis can deliver this relative importance and utility. Therefore, a conjoint analysis is performed in this research in order to fill the research gap.

Research Objective

The objective of this paper is to build a greater understanding of parents’ search behavior, preferences and selection criteria of child care.

Research questions:

 Which factors are key for attracting new customers for a child care company

Sub questions

 How do customer search for child care companies

(9)

Outline of the thesis

In chapter 2 the desk research will be reported. The desk research will start with a short explanation of the Dutch child care sector. Thereafter the desk research continues with a search for literature on parents’ search behavior and child care preferences. The outcomes and limitations of the literature will be discussed. Finally, the most important attributes of child care will be selected and put into a conceptual model.

The conceptual model will constitute the basis of the field research, which can be found in chapter 3. The field research is based on an online questionnaire. In the first part of the questionnaire parents were asked questions about their demographical profile. In the second part of the survey they were asked to value child care attributes (in general and of their currently used child care provider) and their search behavior. The final part of the questionnaire consists of a conjoint analysis, where the parents were asked to score 10 hypothetical child care providers. In chapter 3 the results will be discussed as well.

Chapter 4 provides the conclusion of this paper.

In chapter 5 and 6, respectively, the limitations of this paper and suggestions for future research will be discussed.

2 Desk research

2.1 Introduction

(10)

2.2 Literature review

2.2.1 Parents’ child care preferences Child care modes

The first separation within the Dutch child care sector is between formal and informal child care. To be categorized as formal child care, the mode of child care should be covered by the Dutch child care law. In this law the government demands several restrictions for the formal child care providers. Informal child care is defined as every type of child care that is not mentioned in this law and therefore parents cannot get grants for its use. Several researches have shown that grants are a powerful incentive for higher use of child nursery and that they encourage labor market activity (Gustafsson & Stafford, 1992).

Informal child care

Babysitting is sometimes on a voluntary basis but most of the times there is some reward for

babysitting. The sitter is most of the time a grandparent, family member, or a friend but can also be another person. The purpose of a playgroup originally was not to look after children, and therefore it is not considered to be a mode of child care (Van Egten & de Hoog, 2007). However, sometimes it is used for this purpose and hence it will be elaborated upon here briefly. Foremost reason why children go to a playgroup is to be with other children and have the possibility to play with them. The development (both educational en social) of the children is important. Unlike the other informal modes of child care, the playgroup has opening hours. But because it is not a formal type of child care, it neither has to meet the governmental demands, nor are parents able to obtain a grant for the usage of a playgroup. An au pair is someone who helps parents looking after their children and sometimes does some housework. The au pair temporally lives at the home of the family. Normally au pairs get some allowance for their work.

Formal child care

In 2009 more than 500,000 children were looked after by a formal child care provider (CBS, 2011c). As mentioned previously, parents, that use formal child care, can claim a grant from the government. The restrictions are: both parents should be either working, studying or should be trying to improve their chances of becoming a member of the Dutch labor force and the child care provider has to be registered in the “national register child care” (LRK). To be able to be registered in the national register child care, the child care provider has to meet some additional criteria, like safety & health regulations and an adequate education of the sitters. The most used mode of formal child care is Day

(11)

number of children per group is 16, with a minimum of two sitters, who need to have proper education. Furthermore, there should be a parent committee. Opening hours are during day time only (Van Egten & de Hoog, 2007). Very similar to day care is the crèche. This type of child care is run by the parents of children. Each of the parents will look after the children alternately. All of the quality demands for day care also apply for crèches except that a crèche does not require a parent committee (Van Egten & de Hoog, 2007). Second most used mode of child care in the Netherlands is

before & after school child care. Before and after school child care aims at children that go to primary

school. Children can go to this mode of child care before or after school (or both); when they have a day off; or sometimes also during the holidays. Following a law in 2007 (Van Egten & de Hoog, 2007), all Dutch primary schools have to offer this mode of child care to the parents, who need their children to be looked after. Finally, the least used mode of child care is Guest parent child care. Guest parent child care is nursery at home. The guest parent can look after the children at the parents’ home or at the home of the guest parent. Therefore, this form of child care has a homey atmosphere. Guest parents can look after the children day and night. A maximum of six children can be looked after, in the range of ages 0 to 12. It is mandatory that a guest parent is mediated by a guest parent agency. When a suitable link is made between the parent and the guest parent, the guest parent agency will take care of administrative tasks, accompaniment and the payment of the guest parents (Van Egten & de Hoog, 2007). Due to regulations, a first aid for children certificate, a certificate of good conduct and a guest parent certificate are mandatory for guest parents.

Percentage selected child care modes

In 2006, 38% of the Dutch households with children (in the ages of 0-12) used some kind of child care. For households with children of the ages 0 to 3, this percentage was 50% and for the households with children between 4 and 12 years old this percentage was 32% (van Oploo & Engelen, 2006). 0-3 years old (n=445) 4-12 years old (n=575) Total Formal childcare 39% 27% 32%

Informal child care 45% 61% 54%

Combination formal/informal 16% 12% 14%

Table 2.1 Relative usage of formal and informal child care of Dutch households with children that use child care (van Oploo & Engelen, 2006)

(12)

Parents’ choice for formal child care is further specified in table 2.2 (CBS, 2011c). Guest parent nursery has grown fast during the last 5 years, except for 2010 when new regulations became mandatory. However, it is still the least frequently used mode of child care. In 2010 one sixth of the children were nursed by a guest parent.

Number of children

(x 1000) Percentage

Guest parent 100 15%

Before and after school care 230 35%

Day care 320 49%

Total 650 100

Table 2.2 formal child care choice of parents in the year 2010 (CBS, 2011c).

The different modes of child care have different characteristics and parents are expected to have different preferences for child care. In the next section an overview of the literature will be given on parents search behavior and preferences regarding child care.

The search behavior of parents

(13)

Number (n = 145) Percent

Friends 71 49

Drove by the center 24 17

Other 19 13 Relatives 10 7 Yellow Pages 8 6 Neighbors 6 4 Licenser 5 3 Advertisements 4 3

Child development Specialists 3 2

Doctors 0 0

Total 145* 100*

Table 2.3 Parents’ first source of information about day care (Bradbard et al. 1994) *the sums exceeds 145 and 100, respectively, since several alternative answers were possible to be given simultaneously.

According to Fuqua & Labensohn (1986), 50% of the surveyed parents used only one source and 34% used only two sources during their child care search. The surveyed parents also mentioned that they had additionally spoken to their friends in 77% of the cases in order to discuss the child care. Kontos et al. (1995) found that 67% of the parents were already familiar with the child care facility before they were using it. They had heard about it via their friends, neighbors, family or colleagues. Parents’ average search time of finding child care is not very long. 74 % of the parents take less than one hour to search for child care (Widdows & Powell, 1990) and on average they only searched information about 2 or 3 child care providers (Bradbard et al., 1994). The parents responded that costs, time and transportation (when visiting the child care provider) were the major reasons for limiting their search time. Parents can also ask for help from referral agencies to find the appropriate child care provider. While often neglected by most authors, Fuqua & Schieck (1989) found that 52% of the parents had used a referral agency. Results indicated that parents, who used a referral agency, used more reliable information about the quality of the child care, took more time to look for child care and visited the child care locations more often. In the case of guest parent nursery in the Netherlands, the guest parent agencies are the referral agencies. They inspect the houses where the children will be looked after and make sure the guest parents have the right certificates. Therefore, quality can be guaranteed more, which is especially important when child care is at home where quality is less visible.

(14)

lower. Nevertheless, the internet will probably not give as much reliable information on quality aspects, as visiting the child care facility, or gathering recommendations from friends and family would do. After gathering information on child care providers, parents sometimes visit these providers. Fuqua & Labensohn (1986) found that 80% of the parents visited the child care provider before their children were attending the child care facility. In more detail, of these parents, 52% visited just one possible placement, 37% visited two or three, 4% visited four or five and 7% visited more than five possible placements. Bradbard et al. (1994) and Pungello & Kurtz-Costes (1999) show similar results about visiting possible placements. In general, parents do not visit many providers and when they visit a child care provider, this child care provider becomes the one the parents will choose in most cases.

The parents’ selection of child care

(15)

Attribute Mean score Safety 4.88 Trained staff 4.25 Activities offered 4.18 Religious/moral standards 4.13 Educational content 4.05 Flexible schedule 3.85 Convenience 3.70 Cost 3.10

Child's friends involved 3.05

Table 2.4 Parents’ rating of child care attributes (Widdows & Powell, 1990).

All studies that showed scores for the importance of the attributes of child care were performed like in the research of Widdows & Powell (1990): the respondents were asked to give scores to the attributes and the average score was calculated. The attribute with the highest score was most important. This only reveals how important each attribute is, not their relative importance. From Widdows & Powell (1990) we know that parents gave cost an importance score of 3.1 and safety a score of 4.88. It is perfectly clear from this that safety is more important than cost to the parents who participated in the research of Widdows & Powell (1990) however it is by no means clear how parents make their decision regarding child care. How much the attribute safety and cost influence the decision cannot be extracted from these scores. It is this shortcoming that gives room for new research opportunities and in this paper it will be tried to fill this research gap. Further, Johansen et al. (1996) conclude that quality of child care is not unidimensional. Each type of child care is in fact referring to a broad range of attributes, in varying proportions (Johansen et al, 1996).

(16)

care provider, which has become cheaper, instead of going to a higher quality child care provider (which, given the higher grants, would then cost the same as their current child care provider without the grant). Thereby, Hagy (1998) concluded that, parents prefer lower cost over higher quality and thus that cost is more important than quality. Moreover, also Hofferth & Wissoker (1992) reported that price is one of the most important factors when it comes to selecting a mode of child care. Hofferth & Wissoker (1992) acknowledge that the amount of parental use would not significantly change if the quality changes. The opposite is true for change in price. If the price of child care would be lower, for example due to (higher) grants, more parents will use the child care and will also use it more frequently. Hofferth et al. (1996) explored the characteristics of day care, home care and care by relatives, such as price, availability, child/staff ratio and travel time, as well as the characteristics of the decision makers, such as age, education and income. Results showed the child/staff ratio was not a significant factor in the parents’ choice of child care type. On the contrary, travel time was an important (and significant p < 0.001) characteristic that influences child care choice. Therefore, child care that is nearby is more likely to be chosen by the parents. Another outcome of the research of Hofferth et al. (1996) is the price elasticity of the various types of child care. For all types of child care the price elasticity was significant (p < 0.01 for day- and home care and p < 0.05 for relative care). Table 2.5 contains the price elasticity. The price of day care is the most price elastic, followed by home care and care by family, which is the most price inelastic.

Type of child care Price elasticity (n = 1206)

Day care -0.291

Home care -0.181

Relative care -0.029

Table 2.5 price elasticity of child care (Hofferth et al. 1996)

(17)

location, and cost of care do not significantly influence the choice between family day care and center day care.

Van Egten & de Hoog (2007) investigated why Dutch parents choose a specific type of child care. They distinguish between informal child care, guest parent child care and day care nursery. On every type of child care, 10 respondents were asked how important they think 12 attributes of child care are and furthermore, how their currently used child care provider scores on these attributes. The parents were also asked how they believed other modes of child care would score on these attributes. Parents turned out to value the type of child care they used themselves with higher ratings than the child care modes that were unfamiliar to them. Subsequently, Van Egten & de Hoog (2007) concluded that most parents choose the type of child care that they are most familiar with. In most cases, this is informal child care or day care. Guest parent nursery is less well known. As mentioned previously, when parents are looking for child care, they do not search for a long period of time. Parents base their decision most of the times on their own pre-existing knowledge of child care or listen to what friends and family tell them. Finally, van Egten & de Hoog (2007) recommend guest parent agencies to improve the awareness among parents about the possibilities of guest parent nursery. Also, they recommend that parents should be informed with the benefits of guest parent nursery and its quality restrictions demanded by the government.

Ceglowski (2004) assigns the following attributes to child care quality: classroom composition; curriculum and program philosophy; physical environment; staff characteristics; adult-child interactions; and parent-staff communication. Stakeholders (parents; program administrators; teaching educators; and child care providers) were asked how they value the attributes of child care providers and child care programs. Surprisingly, parents showed a different hallmark of the quality of child care providers. 41% of the parents reported that communication with families is the most important. The other stakeholders place more emphasis on professionalism and training as well as caring and stableness of the provider. The attribute of child care programs that the respondents overall mention most frequently, is learning structure and culturally responsiveness. Only the program administrators and teaching educators value group size and ratios as more important. All stakeholders believe that qualitatively good child care programs contribute to the happiness of the children. Ceglowski (2004) is the only author that also investigated other stakeholders besides parents. However, there seem to be not many fundamental differences in child care preferences between the different groups of stakeholders. Only 41 parents were asked to take the questionnaire. However, the differences were significant.

(18)

age of the child was an important determinant in choosing for a specific child care mode. When a child was very young, it was less likely that parents chose a day care center. Further, employed mothers are less sensitive to price changes of day care centers than other modes of child care. Child care price elasticity as researched by Hofferth et al. (1996) acknowledge this result: the choices of non-employed mothers are more sensitive to price changes.

Hofferth & Wissoker (1992) distinguish four modes of child care: day care; care by a sitter; care by the father; and care by a relative. Once more, the major result of this research is that price is a very important factor affecting parents’ child care choice and that parents with a higher income are more likely to use day care. The factor quality, which in their study was defined as the child-staff ratio, is a factor in decision, however not for all modes of child care. Child/staff ratio may not be a factor for day care. Policies that effectively reduce price are likely to have an impact on parental choices (Hofferth & Wissoker, 1992). Policies that have an effect on care quality are not expected to have an impact on parental choices. Since other factors, like warmth and familiarity, have been shown to be more important to parents than the child/staff ratio. Furthermore, location and convenience are also considered to be important. In Chaplin et al. (1996) the results of the article of Hoffert & Wissoker (1992) are revised. This resulted in a higher and significant effect of the child/staff ratio on parental choices and a lower effect of the impact of price on parental choices. However, even following this revision, the price effect remained significant and negative.

(19)

speak. Parents were able to observe the quality more adequately when the mode was family day care or relative care.

2.2.2 Limitations and new research opportunities

As discussed previously, the literature on parents’ search behavior of child care is aged. Although it is not likely that that the selection criteria chance much over time, the way people get familiarized with child care nowadays, is expected to change significantly. Due to the extensive use of the internet, parents are currently expected to use this medium as a first source of information often. Parents will also get their first impression about the child care provider from the website. Furthermore, there are websites and forums on which individuals can give their opinion about companies and rate these companies. These ratings can serve as guidance for parents that look for child care. Recent research from Gregg & Walzcak (2010) shows that websites nowadays are very important for the trustworthiness of a company. Their research was focusing on the trustworthiness of sellers of products online. They concluded that the sellers with websites with a good design were trusted more often, even if these sellers were not recommended. Sellers with a website with a poor design were trusted in fewer cases, even if they were recommended. Therefore it can be concluded that websites of a company are an important tool to give the company a good image. The first part of the questionnaire that was used in this study will consist of questions of the search behavior of the respondents. Results can show if this behavior has changed and what the role of the internet nowadays is. The effect of having a good website is beyond the scope of this paper.

(20)

In the United States the criteria in order to receive child care funds are:  The child must be between 0-12 years old

 The income of the child’s family must not exceed 85% of the median income of a family in the United States.

 The parent or parents need to either have a job, or have to be actively searching for a job or have to be studying.

These criteria are similar to the ones used in the Netherlands, however in the Netherlands there is no income restriction, although the grant will be lower in the case of a higher income. In the United States the care provider also faces some restrictions. The restrictions are aimed at guaranteeing the children’s safety, health en educational development (Greenberg et al., 2003). In the Netherlands similar restrictions for child care providers exist. From the comparison between the United Stated and the Netherlands it can be concluded that both countries have fairly similar rules and regulations regarding child care. However, the government’s spending on child care grants per child is a bit higher in the Netherlands. And in the United States to be eligible for a child care grant, one's income has to be lower than average. Since many regulations in the United States are similar to the ones in the Netherlands, it is not likely that these small differences change the way parents choose child care. Thus the outcomes of the research performed in the United States can be used as a guideline how Dutch parents choose their child care.

(21)

percentages of the people who participated in the survey, where no tests were performed to obtain the reliability of the sample. Most analyses were carried out by comparing the modes of child care with a logit analysis. Although this analysis gives good insight in how the preferences for a mode of child care differs and which attributes parents find most important, it does not show what the relative importance of the attributes is. Some papers also showed the importance of the attributes by asking parents to give scores to the attributes. For example, Powell & Widdows (1990) ranked child care attributes according to their mean score they obtained from the interviews. Although Bradbard et al. (1994) also ranked attributes of child care, they did not provide a score or value of each of the attributes. Nevertheless, both studies give insights in what parents find important attributes and which characteristics are less important but again both papers do not give insights how the importance of the attributes relate to each other. This shortcoming, that attributes are studied as independent factors, can be found in all the existing literature. As Pungello & Kurtz-Costes (1999) mentioned, all the authors let the respondents give scores to child care attributes separately. Subsequently, how exactly parents make their child care choice is not yet clear. And especially this can be useful for a child care provider: to know exactly what combination of attributes is most influential on parent’s choice. To obtain the relative importance and utility of each of the child care attributes, a conjoint analysis can be performed. In this case the utility is how much satisfaction parents obtain from a child care attribute. The utility of the attributes can be easily compared because an attribute that has a utility that is twice as high as another attribute is thus twice as important for the total satisfaction of the parent. In chapter 3 of this paper the utility and relative importance of 5 attributes will be determined by performing a conjoint analysis. Yet first, an overview of the child care attributes that were found in the literature will be given. Since not all attributes can be used in the questionnaire, especially for the conjoint analysis, a selection is made of which attributes to focus on.

2.3 Conceptual model

(22)

(23)

Child care’s utility

Price

Safety and hygiene

Offered activities

Similar to home situation

Small scale

Child care when child is sick

Same sitter every day

Quality control

Cosiness environment

Playing with other children

Child’s development

Flexible opening hours

Distance

Sitter’s caring Sitter’s Professionalism Personal attention +

-+ + + + + + + +

-+ + + + +

Figure 2.1 the conceptual model.

3 Field research

3.1 Research methodology

3.1.1 Introduction

(24)

3.1.2. Questionnaire – closed questions

Existing literature can be used as a guideline of what parents find important child care attributes. However, a new questionnaire is made to know if the results of that literature also apply for the Netherlands and to obtain knowledge whether nowadays parents’ search behavior and child care preferences have changed. The questionnaire will be online and parents can visit a website were the questionnaire will be available. The advantage of an online questionnaire is that many people can be reached, who live spread around the Netherlands, at low cost. However, there are disadvantages of using an online questionnaire. First of all, the participation rate is low. Most people who receive an email in which they are asked to fill out a questionnaire will not participate. Furthermore, an online questionnaire is only available for a certain group of people and therefore there will be a selection bias: only people who signed up to complete questionnaires on a voluntary basis, will be able to participate. Therefore, their demographic profile is likely to be different than the average parent in the Netherlands. Finally, the respondent will only answer the questions which are in the questionnaire. No further explanations and additional questions can be asked. Therefore, clear questions, which contain the subjects of research, should be provoked. The questions of the questionnaire should at least cover the conceptual variables from the conceptual model (Emans, 2007). In this case the conceptual variables are the attributes that influence the utility of the child care provider. Most of the questions will be closed questions. Closed questions normally require less time to answer and process. The closed questions used in this questionnaire have two different forms. The first one is a closed question with the 7 point Likert scale. For these questions the respondent is asked to indicate how much he/she agrees with the question or statement. The second form of closed questions are multiple choice questions. Besides questions about the child care attributes, questions about the search behavior of the parents and some general questions about the respondent are asked. The latter can be important to know if some demographic group of parents values the attributes differently. In total around 310 parents answered all questions. This amount can change per question because some parents did not fill out every question.

3.1.2 Questionaire - conjoint analysis

(25)

(26)

has two or three levels. The overview of the attributes with their levels that will be used by the conjoint analysis can be found in table 3.1.

Attribute Level

Child care location (child care mode and distance)

At parent’s home

At sitter’s home (15 min.) At Day care (15 min.)

Quality Sitter has child care certificates and first and

degree and the location is inspected

Sitter does not have child care certificates and a first aid degree and no location inspection has taken place

Child’s (educational) development There are many toys and educational materials There are not many toys and educational materials

Price Price per child per hour is 2 euro

Price per child per hour is 4 euro

Opening hours Only during office hours

24/7

Table 3.1 attributes and their levels for the conjoint analysis

(27)

But again it is hard formulate different levels of such an attribute. The amount of books and toys on the contrary can be distinguished. In this case there are two levels. The first level is that there are many toys and educational materials. The second level is that there are not many toys and educational materials. Price is the only attribute that can be formulated into numbers. There are two levels: 2 and 4. Child care without a grant from the government will cost around 4 euro per child, per hour. With a grant from the government it will cost around 2 euro per child, per hour. The latter is an average, because the price, after subtraction of the government grant, depends greatly on the income of the parents. The attributes opening hours also has two levels: 24 hours per day, 7 days a week versus only during office hours.

First step in the design of the conjoint analysis is the insertion of the attributes into SPSS. The software will then generate a representative subset of profiles: the orthogonal array. This orthogonal array, or orthogonal design, will be the profiles that will be presented to the respondents. In this case a number of 10 profiles are chosen. This was also the minimum for the amount of attributes. Two of these profiles will be the holdout profiles. These holdout profiles will not be included when the utility and importance ratings will be calculated. The purpose of including holdout profiles is to provide a proximal indication of validity, measured by the utilities’ ability to predict choices not used in their estimation. (Johnson & Orme, 2010). Thus, when the scores that parents gave to the holdout cases is similar to the predicted score (utility), based on the scores of the other profiles, the ability to predict the utility is high.

3.2 Results

3.2.1 Parents’ search behavior

(28)

were working. First if all the correlation between some of these variables is checked. As can be seen from table 3.2 most correlations are weak and thus the variables are independent. The only two variables that have a modest correlation are education and income and are therefore not completely independent. Education Income Number of children Age of youngest child Education 1 Income 0,401578483 1 Amount of children 0,001775372 0,087933522 1

Age youngest child -0,139173956 -0,040474842 0,055667099 1

Table 3.2 correlation between variables of the parents who participated in the questionnaire.

The parents were randomly chosen from a database and asked to take the survey. Therefore the sample is as representative as possible for the whole population of parents in the Netherlands. To check for bias, the sample is compared with properties of parents that have child care. First of all it will be checked whether the parents in the sample chose the same proportion of modes of child care as all the parents in the Netherlands that use child care.

Mode of child care N Sample Population

Guest parent n=34 12% 15%

Before and after school care n=85 30% 35%

Day care n=167 58% 50%

Table 3.3 proportion of the modes of child care of the sample and population (CBS, 2011c)

(29)

Income group N Sample Population

0 - 20.000 euro n = 5 2% 1%

20.000 - 40.000 euro n = 41 16% 14%

40.000 - 60.000 euro n = 78 29% 22%

60.000 - 80.000 euro n = 70 27% 38%

more than 80.000 euro n = 70 26% 25%

Table 3.4 income of the respondents and of the population (Jongen, 2010)

Finally, the education level of the respondents and of the Dutch workforce with children can be found in table 3.5. Again there exists some differences between the sample and the population, and according to the chi-squared test these differences are highly significant as well.

Level N Sample Population

1 = primary school n=0 0% 2% 2 = Mavo n=3 1% 11% 3 = Havo n=9 3% 7% 4 = Vwo n=5 2% 5 = MBO n=61 18% 36% 6 = HBO n=157 48% 27% 7 = University n=95 29% 15% Total n=330 100% 98%*

Table 3.5 education levels of respondents as well as the education levels of parents in the Netherlands (CBS, 2011b) * the sum is not equal to 100 percent because the education level of some people is not known.

The education level that appeared most frequent among the respondents was HBO (48%) In the population this percentage is 27% and there MBO is most frequently found education level. Also the percentage of university graduates is much higher in the sample. On the contrary there are relatively less people with a low education in the sample. Overall it can be concluded that the respondents have a higher education level than the average Dutch parent. However, the higher education does not result in a higher household’s income than average levels of Dutch households with children (even though there is a modest correlation between these two characteristics). The distribution of the modes of child care differs a bit as well, but they are ranked the same. In a later paragraph the difference in demographics will be checked on differences in preferences and behavior.

(30)

obtained the information about child care providers. This information can be their friends’ experience with the child care provider, information about prices, subsidies, if there is a sitter available or some general information about the child care provider. In conclusion, all information that causes the parent to know the child care provider. The results of how the parents obtain information about the child care provider can be found in figure 3.1. There are three sources most parents use to find a child care provider. First of all, the internet. The internet has been used by 2/3 of the respondents. This information source has not been included in the existing literature. Also friends and family are frequently used. Finally, many people also responded they start looking for child care in their neighborhood. All other sources of information are used much less frequent.

Figure 3.1 sources of information of child care providers (n=306, the x-axis represents the type of information source, the y-axis represents the number of people who used this source )

Next the parents were asked from how many child care providers they requested information. In figure 3.2 the amount of information request are shown. Most parents request information from 1 or 2 child care facilities (together more than 60% of the respondents). Only few parents request information of more than 5 child care providers. This result is in accordance to the literature (Bradbard et al., 1994), which also found that most parent only search for information about one or two child care providers.

(31)

Figure 3.2 Number of information request (n= 306, the x-axis represents parents’ number of information request, y-axis represents amount of people who had that number of requests)

Furthermore parents were asked from what modes of child care they requested information. Results can be found in figure 3.3. Parents, who ultimately had chosen a certain mode of child care, also requested more information about this mode of child care. Therefore, it is expected that people already have a mode in mind and will search information of this mode more often. Especially parents who use guest parent child care had searched for information of this mode more often.

Figure 3.3 Percentage of the mode of child care (n=306, the x-axis represents the percentage from which mode parents requested information from, the y-axis represents the child care mode parents chose ultimately) 0 20 40 60 80 100 120 1 2 3 4 5 6 7 8 >9 0% 20% 40% 60% 80% 100% guest parent (n=26)

before and after school (n=82) day care (n=164) sitter care (n=34)

before and after school day care

(32)

Search time

Literature shows that people do not search for a long time (Widdows & Powell, 1990). The questionnaire shows similar results (figure 3.4). The mode of the search time is 1 hour. Most people do not search for more than 5 hours. Average search time is around 5 hours because some people responded that they search for a long time (20 hours or more)

Figure 3.4 search time for child care providers (n = 317, the x-axis represents total search time in hours, the y-axis represents the amount of people who searched for this amount of time)

In figure 3.5 the amount of information requests and the search time per child care mode can be found. The graph shows that parents who chose formal child care on average take more time to find the appropriate child care provider. The amount of information requested of child care providers is less different for each mode of child care, but still the parents who use formal child care modes have higher average number of information requests.

(33)

Figure 3.5 average amount of search time and child care information requests. (n=306, x-axis represents the average number of information requests as well as the average search time in hours, the y-axis represents the child care mode)

Finally, the parents were asked if they have visited the child care provider before their child was looked after at that facility. It turned out that most of the parents visited the child care location only once (65%). 27% of the parents went to the child care location more than once and 8% never went to the child care provider. In figure 3.6 the results are shown divided into the 4 modes of child care. Again differences between formal and informal child care can be distinguished: parents who chose the formal modes of child care visit the provider more often. Parents who use sitter nursery do not visit the location of child care in 20% of the cases. Sitter care and guest parent child care can also be at the parents’ home and therefore visiting the child care location is not needed, which would explain why many parents that use sitter care do not visit the location. However, parents who use guest parent child care respond that they always visit the child care location at least ones. The results of Fuqua & Labensohn (1986) show similar results, most parents just go to one child care provider.

0 2 4 6 8 10 12 14 Guestparent (n=26)

before and after school child care (n=82)

Day care (n=164) Sitter care (n=34)

(34)

Figure 3.6 Amount of visits to the child care provider per child care mode (n=306, the x-axis represents the percentage of parents who visited the provider never, one time or more than one time, the y-axis represents the mode of child care)

3.2.2 Parents’ child care preferences

(35)

Figure 3.7 Parent’s average importance rating of child care attribute (n=321, the x-axis represents the score on a 7 point Likert scale, the y-axis represents the child care attributes).

Clearly, parents find quality attributes most important (quality of sitter and environment) followed by the educational development of the child. Later these results will be compared with the outcomes of the conjoint analysis. These initial results are in accordance to most literature that quality is indeed more important than price. However, from these results it is hard to tell which attribute is decisive for the choice of child care provider.

Attribute Importance per child care mode

Different child care modes have different characteristics. For example: the possibility that a child can play with other children is higher at the day care than for sitter and guest parent care, since there are more children at the day care. It is expected that parents who chose a certain mode of child care place higher value on the attributes that are specific for that child care mode. Therefore, the importance ratings were split into different child care modes to see if parents who make use of the different modes give different importance ratings to the attributes of child care. The results are shown in figure 3.8. Surprisingly, the graph shows that parents who use different child care gave very similar importance ratings to almost all the child care attributes. Nevertheless, there are some

(36)

(37)

Figure 3.8 attribute’s importance rating per child care mode (n=321, the x-axis represents the score on a 7 point Likert scale, the y-axis represents the child care attributes)

Parents’ rating of their child care mode

Parents were also asked to rate their currently, or most recently, used child care provider. The results can be found in figure 3.9. Before and after school nursery and day care scored somewhat similar. Sitter nursery and guest parent nursery scored on average higher. Especially guest parent nursery scored higher on almost all attributes of child care and thus parents are most satisfied with this mode of child care. Somewhat surprisingly, also sitter care scored very high on many attributes. Also the attributes that are not distinctive for this mode of child care scored high. For example the professionalism of the sitter scored high for sitter care. However, no education is required for a

0 1 2 3 4 5 6 7 price Personal attention Professionalism sitter's caring closeness of location flexible opening hours child's development playing with other children cosiness environment quality control same sitter every day child care when child is sick small scale similar to home situation offered activities safety and hygiene

sitter care (n=35) day care (n=173)

(38)

sitter. Therefore from these scores it is expected that parents are most satisfied when the child care provider is at home and of small scale. At a later stage, when we will introduce the conjoint analysis, the utility parents obtain from child care that is at home will be tested.

Figure 3.9 Attribute’s scores of parent’s currently used mode of child care (n=321, the x-axis represents the score on a 7 point Likert scale, the y-axis represents the child care attributes).

The differences are substantial, but the amount of people that use guest parent child care (n=27) and sitter care (n=35) is not very high. Therefore the results need to be validated. To test the validity of the scores, the scores of guest parent nursery were tested on significance against the scores of the other modes of child care. The result from this test can be found in table 3.6. The test concludes that almost all scores of guest parent nursery are significant higher than the scores of before and after school nursery and day care at the 99.9% level. Sitter care and guest parent care have more similar results, here the significance levels are 95% and 90% and also on just 4 attributes.

0 1 2 3 4 5 6 7 price Personal attention Professionalism sitter's caring closeness of location flexible opening hours child's development playing with other children cosiness environment quality control same sitter every day child care when child is sick small scale similar to home situation offered activities safety and hygiene

sitter (n=35) day care (n=173)

pre and after school child care (n=86)

(39)

Attribute

pre and after school

child care day care sitter care

price 0.001 0.005 -

Personal attention 0.001 0.001 -

Professionalism 0.001 0.05 0.1

sitter's caring 0.001 0.05 -

closeness of location - - -

flexible opening hours 0.001 0.001 -

child's development 0.001 0.001 -

playing with other children - - -

coziness environment 0.001 0.001 -

quality control - - -

same sitter every day 0.001 0.001 0.05

child care when child is sick 0.001 0.001 -

small scale 0.001 0.001 0.05

similar to home situation 0.001 0.001 0.1

offered activities 0.01 - -

safety and hygiene - - -

Table 3.6 Significance of the difference of the attribute’s scores of guest parent nursery compared to the other modes of child care.

(40)

Figure 3.10 Total satisfaction of parents’ currently used child care, which is the sum of the scores from figure 3.9 (n=321, the x-axis represents the mode of child care, the y-axis represents the sum of the attribute scores)

3.2.3 Conjoint analysis

The conjoint analysis was based on a sample of 294 respondents who gave scores to all the 10 profiles. The conjoint’s importance rating of all the attributes taken together is 100 and an attribute with an importance score of 20 will be twice as important as an attribute with a score of 10. Thus the real differences in importance can be extracted from the results. The conjoint analysis includes fewer attributes, as discussed before. There are 5 attributes used: location; opening hours; quality; education and price. Results show, as can be found in table 3.7, that quality is the most important and price is the least important attribute. Which is in accordance to most found literature and also with the results as discussed in the previous paragraphs. However, this conjoint analysis also makes clear that quality is indeed the most important factor for choosing a child care provider. First parents look at the quality but after that, and almost as important, they look at what kind of child care mode it is and what the distance to the child care provider is. The price is of much lower importance for choosing a child care provider.

Location 26,175

Opening Hours 12,098

Quality 33,486

Education 17,959

Price 10,282

Table 3.7 Averaged Importance Score of the conjoint analysis (n=294)

To be able to make a better comparison between the previous results and the results of the conjoint analysis, the attributes as used in the first part of the survey will be combined into the 5 attributes as used with the conjoint analysis (see appendix B). The results of the combined attributes can be found

75 80 85 90 95 100 105

Guest parent (n=27) Before and after Child care (n=86)

Day care (n=173) Sitter care (n=35)

(41)

in figure 3.11. Note that, since all attributes scored at least somewhat important, the scores are all decreased by 4 (4 is the ’neutral’ score) and therefore the differences between the attributes become more visible. When we compare the conjoint result with the results of the first part of the questionnaire we can see many similarities. First of all quality is the most important attribute. Secondly, educational development is also important. However in the conjoint analysis location (the mode of child care and distance) is found very important while in the results of the first part of the questionnaire distance and the attributes that are distinctive for a certain mode of child care are less important. This can probably be explained because the distinctive child care mode attributes from the questionnaire include similarity with home situation and small scale. These attributes are the ones that received the lowest importance scores. A second explanation is that people believe distance is not very important, but when parents actually have to choose between child care providers, the first one is close and the other distant, distance becomes a very important attribute. Price is valued least and opening hours as quite important in both cases.

Figure 3.11 importance’s rating of all attributes as in figure 3.7, but were the 15 attributes are put together into 5 attributes for better comparison with the conjoint analysis (n=321, the x-axis represents the attribute, the y-axis represents the average importance scores minus the neutral score of 4)

The first part of the questionnaire already gave us a good overview of how parents value the 16 child care attributes. However the questionnaire was not able to determine the utility of each of the attributes. In table 3.8 the utilities of each of the attributes and their levels is shown. Most surprisingly is that the utility of price 4 is higher than the utility of price is 2. Some respondents commented that they do not trust a child care provider that costs only 2 euro and also has all the

0 0,2 0,4 0,6 0,8 1 1,2 1,4 1,6 1,8

(42)

quality attributes (certificates and location inspection). These parents also responded that it is just not possible to have a high quality child care for just 2 euro.

Utility Estimate Std. Error

Location At Home 7,461 2,075

At Sitter's Place -2,479 2,433

At Day Care -4,982 2,433

Opening Hours 24/7 3,904 1,556

Office Hours -3,904 1,556

Quality Quality Control 13,911 1,556

No Quality Control -13,911 1,556

Education Educational Materials 6,549 1,556

No Educational Materials -6,549 1,556

Price 2 0,341 3,113

4 0,682 6,225

(Constant) 47,364 4,949

Table 3.8 utilities of the attributes and their levels of the total sample (n=294)

When we look at the utility of the individual respondents as shown in appendix E we can see that most of the time they have a similar utility for the attributes. While the magnitude of the utility changes, whether the utility is positive or negative was most of the time the same. Therefore the utility is clearly not random, parents have a distinct preference for a level of the attribute. The only utility that does not have a similar magnitude and sign among the respondents is the price. Prices, two and four, have both positive as well as negative utility values. And therefore it is not really clear what the meaning of this utility is. As discussed before some people do not trust child care that costs 2 euro, while other prefer child care of 2 euro because it is cheaper (then child care is a normal good: more will be demanded when it is cheaper and a lower price is preferred over a higher price). The result of this utility score of price is a high standard error. Therefore, it is hard to say what utility parents obtain from the price of child care. However, this does not change the importance score of price: parents consider price not important for choosing child care.

A measure for the validity of the conjoint analysis is the Kendall’s tau. This test compares the respondents’ rank of the scores and the rank of the estimated scores. When the Kendall’s tau is 1, both rankings are the same and when the Kendall’s tau is -1 the ranking is perfectly inverted. The Kendall’s tau is 0.929 and thus the utility can be predicted well (see table 3.9).

Value Sig.

Pearson’s R 0,996 0

Kendall’s tau 0,929 0,001

Kendall's tau for Holdouts -1

(43)

The respondents had to give scores to 10 profiles however just 8 of them were used to predict the importance and utility scores. The other two, the holdout cases or holdout profiles, can be used to provide validity. Surprisingly the Kendall’s tau for the holdout is minus one. However since there are just two holdout cases there are also just 2 possibilities. The ranking of Case A > Case B or the ranking of Case A < Case B. From table 3.10 can be seen that the first holdout (profile nr. 9) received on average a lower score than the second holdout (profile nr. 10) were it was predicted that the first holdout profile would be higher valued and thus the ranking is inverted. As a consequence the Kendall’s tau is -1. Since there are just two holdout cases, the ranking is of not much value. Especially since the predicted values are close to each other (see table 3.10). It is therefore more useful to see if observed scores and estimated scores are close to each other. From table 3.10 we know that the first holdout case has a score that is very close to the estimated score. Also the estimated ranking is the same as the observed ranking (ranking all 10 cases). The second holdout case shows a slightly different predicted than actual score and also the ranking is not the same. However the predicted score as well as the predicted ranking comes close to the actual score and ranking. From the ranking and scores of the holdout cases one can conclude that the utility scores indeed can be predicted quite well and that therefore the utility scores seems to be valid.

Table 3.10, actual and estimated scores and ranking.

With the respondents’ scores already given it is possible to change the holdout cases. For example, to make the holdout cases profile nr. 5 and profile nr. 7, respectively the lowest and highest scores. When these two profiles are not used for determining the utility scores, but as a holdout case the importance and utility scores are similar. This results consolidates the believe that the utility can be estimated well. Furthermore, since the observed score of profile 7 is higher than the score of profile 5 (and thus the ranking profile 5 < profile 7) the Kendall’s tau for the holdouts would be 1 as a result. To test for different child care preferences when parents’ demographical profile changes the sample is divided into different demographical groups.

(44)

Results by income

First of all the sample will be divided into a high and low income group. The first group will have the income of 0-40.000 euro per year, per household and the second group of more than 40,0000 euro per year, per household. It was expected that people with a higher income find price a less important attribute than parents with a lower income. However, from table 3.11 it can be seen that the high income group find price more important than the low income group. From appendix C can be seen that utility is not the same between the two income groups as well. The high income group has a higher utility for child care that cost 4 euro whether the low income group has a higher utility when the price is just 2 euro. This implies that low cost is indeed more important to the low income group. However, the standard error of the utility scores are high, therefore the validity of the results can be questioned. Overall the attributes scored similar.

Low income (n=115) High income(n=174) Sample average

Location 26,762 25.049 26,468

Price 8.985 10.445 10.282

Opening hours 11.977 11.972 12.098

Quality 32.570 34.132 33,486

Educational development 17.052 18.402 17.969

Table 3.11 importance of the 5 attributes by income group.

Results by child care mode

(45)

of opening hours is similar for most the parents, only parents who use guest parent child care think opening hours are less important.

Guest parent (n=23)

Before and after school care (n=73) Day Care (n=147) Sitter Care (n=30) Sample average Location 29,852 26,295 24,686 29,470 26,175 Opening hours 10,344 11,238 12,012 11,747 12,098 Quality 31,409 33,501 34,911 27,557 33,486 Education 16,481 17,377 19,392 19,583 17,959 Price 11,914 11,588 8,999 11,644 10,282

Table 3.12 the conjoint’s importance score per child care mode.

When the importance rating is high, normally the utility of the levels is high (positive or negative) as well. For most attributes it is clear which level has the high and which level has the low utility (for example quality control has high a utility and no quality control has a low utility, see also appendix E). For the different child care locations it is less clear which one has the highest utility. Therefore these utilities will be further analyzed. In table 3.13 the utilities of the separate child care locations is shown for the whole sample and for each mode of child care. First of all it becomes clear that people have highest utility when child care is at home (a familiar environment and no travel time), at the sitter’s home scored second best (only parents who use day care have higher preference for day care) and the lowest utility parents obtain from day care.

Guest parent (n=23) Before and after school care (n=73) Day Care (n=147) Sitter Care (n=30) Sample average At home 7,196 6,450 7,221 11,056 7,461 At sitter’s home -,598 -1,629 -4,258 ,870 -2,479 At day care -6,598 -4,821 -2,964 -11,926 -4,982

Table 3.13 Parents’ utility of the location

(46)

chose different modes of child care is that parents who use before and after school child care or day care obtain higher utility from the mode “day care” and sitter and guest parent users obtain much lower utility for this location, all parents have highest preference for child care at home.

Results by Education

It is expected that people with a higher education demand from their child care that they put greater emphasis on the educational development of their children. From the table 3.14 it can be seen that that indeed parents with a higher education demand more educational development from their child care provider. On the contrary, parents with a lower education find quality more important. Price is a bit more important for the high education group. However, they have a higher utility for the 4 euro child care than the 2 euro child care. The low education group has a higher utility for a price of 2 euro than a price of 4 euro. Since earlier a modest correlation between education and income was discovered this result can maybe be explained because this correlation as well.

Attribute High education (n=214) Low education (n=61) Sample average

Location 25,919 26,788 26,175

Opening Hours 11,849 13,221 12,098

Quality 33,109 36,615 33,486

Education 18,907 14,705 17,959

Price 10,216 8,672 10,282

Table 3.14 importance rating for parent with a high and low education

Search time, amount of children & age of youngest child

(47)

Conjoint summary.

From the conjoint analysis it can be concluded that price is indeed the least important attribute. On the contrary, parents find quality attributes most important. This is in accordance to previous outcomes of this research and most of the literature. However, the conjoint analysis also revealed that the mode of child care and distance is very important for the choice of child care. Furthermore, the parents responded that educational development and opening hours, respectively, are third and fourth most important. When analyzing the different demographical groups it can be concluded that parents with a higher income do not think a high price is a reason to reject a child care provider. At least some parents in the sample even think that child care with a price of 2 euro is too low and therefore cannot be trusted. Therefore their utility of a price of 4 euro is higher than the utility of a child care provider with a price of 2 euro. On the contrary, the low-income group does prefer a price of 2 euro compared to the price of 4 euro. However, the standard error of the utility of price was high. Therefore, much remains uncertain. Parents with a high education think that educational development is more important than parents with a lower education. However, parents with a high education put less emphasis on quality than the low education group. Apart from previous two groups, the importance scores and utilities were similar, and thus on average parents do not have much different child care preferences. The ideal child care provider is a professional sitter at the home of the parents. Here the sitter has to pay attention to educational development by playing with the child as well as educate the child. Furthermore the opening hours of the ideal child care provider are 24 hours a day and 7 days a week. There is no consensus on the price, low income groups prefer a lower price, were higher income group do not trust a child care that has a low very price. Of the 4 child care modes that were taken into account in this research, guest parent nursery comes closest to the child care that is ideal, according to the results of the conjoint analysis. The results of how satisfied parents are with their currently used child care provider confirms this. From that result, guest parent child care was significantly better valued than the other modes (sitter care, before and after school child care and day care) of child care.

4 Conclusion

(48)

parents’ home by a professional sitter. Preferred opening hours are 24 hours a day, 7 days a week. Parents believe that price is not very important, all the results of this research confirm this. Remarkably, the conjoint analysis revealed that parents (with a higher income) obtain a higher utility from a higher price and thus have a higher preference for a higher price.

Search behavior has changed significantly since the publication of the existing literature about child care. The internet is widely available and used by most of the parents to look for an appropriate child care provider. Because of the internet, searching for child care providers has become easier and less time demanding. However, the amount information requests of child care providers did not change much. Neither did parents visit a provider more often. Parents also still obtain information from friends and family and look in their neighborhood for child care providers often.

Because parents mainly base their choice of provider on quality, it is important for a child care provider to show the quality of the child care for attracting new customers. The location (at home or at day care) and distance are also very important but it is not something a child care provider can change easily. The location “at the parents home” is highest valued by parents. Therefore, especially guest parent agencies should make parents more aware what the benefits of guest parent child care are. This child care is at home by a professional sitter and there is quality control. Thus guest parent child care fits the profile of the ideal child care provider best. Parents already have the highest preference for this mode of child care. Guest parent child care scored significantly higher than the other modes of child care on most attributes. But still parents use this mode the least, even though almost all the parents responded they knew guest parent nursery as a mode of formal child care. The child care providers which have the mode day care or before and after school care should mainly focus on their strengths: quality and development of the child. Moreover, distance has become a major constraint for day care and before and after school care use. Therefore more and smaller providers can be beneficial for both child and parent since a smaller scale means more personal attention from the sitter and more providers a lower travel time.

Parents’ Child Care Preferences: a Conjoint Analysis