Estimating the Willingness-to-Pay for Attributes of General Practitioners: A Travel Cost Approach. g.c.godager@medisin.uio.

Size: px
Start display at page:

Download "Estimating the Willingness-to-Pay for Attributes of General Practitioners: A Travel Cost Approach. E-mail: g.c.godager@medisin.uio."

Transcription

1 Estimating the Willingness-to-Pay for Attributes of General Practitioners: A Travel Cost Approach GEIR GODAGER Institute of Health Management and Health Economics, University of Oslo, P.O. Box 1089 Blindern, 0317 Oslo, Norway g.c.godager@medisin.uio.no Incomplete draft. Not for circulation, citation, or quotation. Version of April 30, 2007 Abstract: The aim of this paper is to estimate the consumers willingness to pay for certain attributes of GPs such as whether or not the physician has fulfilled a post-graduate education programme and achieved status as specialist in general medicine. The individual consumer s choice of general practitioner is modelled within the random utility framework. Data are from the regular general practitioner scheme organized as a list patient system introduced in Norwegian general practice in June Prior to the reform the health authorities asked all the inhabitants to submit a form ranking their three most preferred general practitioners (GPs). Interpreting the most preferred GP as the chosen alternative, choice probabilities are estimated by means of conditional logit regression. Having information on Zip codes of patients residential addresses as well as zip codes of addresses at which practices are located, we are able to study how patients balance travel costs and attributes of GPs. Since travelling when healthy may not be the same as travelling when ill, travel costs associated with different modes of travel are considered, and discussed. At the time our data was collected GPs who are specialists in general medicine received an additional fee of 51 NOK per consultation from the National Insurance Scheme, while the patients out of pocket fee for consulting a specialist in general medicine was the same as for consulting a non-specialist. The results indicate that patients willingness to pay for consulting a specialist is considerably lower than the additional fee specialists receive from the National Insurance Scheme. Keywords: GP services. Discrete choice. Travel cost method. Willingness-to-pay. Health care demand. Norway.

2 1 Introduction An important feature of the market for general practitioners service is that the doctor-patient relationship is often long term and likely to be characterized by repeated transactions. Patients consult their GPs more frequently than they use more specialized health care services. A durable doctor-patient relationship results in two-way information transmission between doctor and patient. From the doctor the patient receives information about health problems and the options available for treatment, and the doctor receives information from the patient about symptoms, social background, and how the patient values alternative health care services. It has been argued that increasing the frequency of transactions between parties reduces the costs of information transmission and may also reduce monitoring costs (Williamson, 1993). Several studies report that patients who see the same GP at each visit both prefer and are more compliant with treatment recommendations (Dietrich and Marton, 1982, Hjortdahl and Laerum, 1992). Several studies provide evidence suggesting that physicians who are trained in communication skills have patients who are more satisfied, more compliant with recommended treatment, and have better health status (Ong et al., 1995, Kaplan et al., 1989, Wartman et al., 1983, Stewart, 1995). In this paper we argue that communication and information transmission between doctors and patients are important input in the GPs production function. The rational consumer and potential patient would thus prefer to be listed in a practice of a GP such that the transmission of information is unstrained. An auxiliary hypothesis is that the transmission of information is easier between physicians and patients that resemble one another on observable characteristics. We model the patient s choice of GP within the framework of the random utility model (RUM), and expect that the consumer s utility associated with a given GP is higher the more similar the GP is to the decision maker. Most earlier studies on matching of GPs and patients consist of analysis of individuals stated preferences with regard to hypothetical GPs. (Examples of such studies are Vick and Scott (1998), Scott and Vick (1999) and Veale, (1995)). In June 2001 a regular GP scheme was introduced in Norwegian general practice, making the GPs responsible for the provision of primary care services to the persons listed at their practice. Prior to the reform the health authorities gathered the 1

3 information needed to assign one GP to each Norwegian inhabitant. All inhabitants were asked to rank their three most preferred GPs in a form, and all GPs were asked to report the maximum number of patients they would like to take care of. The health authorities designed an algorithm merging this information in order to obtain a one-to-one match between inhabitants and GPs. This reform gave opportunities to study the revealed preferences of consumers with regard to choice of GP. Applying data from this natural experiment, Hilde Lurås (2003) studied the consumers ranking of GPs and estimated choice probabilities by means of a Luce model of ranking. Lurås found that individuals preferred GPs that were specialists in general medicine compared to GPs who had not engaged in further education in order to get specialist status. The consumer also preferred a GP with the same gender, and Lurås found that the choice probabilities were declining in the age difference between the GP and the patient. In this paper we utilize a part of the data set applied in Lurås (2003) that was not used in that study, namely observations of choice made by consumers living in Oslo, the capital of Norway. This part of the data set was presumably dropped due to difficulties imposed by vast size of the data set. The large size of the sample necessitates the use of a random sample of the observations in order to estimate choice probabilities by means of available computer resources. The study of the behaviour of inhabitants of Oslo thus comes at cost, but there are also potential benefits of studying the choices made in Oslo. The availability of information on the zip-code of the residential address of consumers as well as zip-codes of the addresses at which practices are located give us the opportunity to study how consumers trade off attributes of the GP with something that we can relate to: travel distance. Further, if it is possible to attach monetary value to travel distances within Oslo, we are able to derive a monetary value of GP attributes, such as GPs engaging in training and education to become a specialist in general medicine.description of difference between being a specialist and not being a specialist to be entered! 2 Random utility and the logit model We denote by U ij the utility consumer i obtains when selecting GP j. The utility is decomposed into a part that is a function of variables that are observable to us, 2

4 and a part that is unobservable. We denote the observable or deterministic part of utility by V ij. The random part of the utility is denoted by ε ij. The logit model is obtained by assuming that ε ij is independently, identically distributed extreme value type 1. The utility consumer i obtains when selecting GP j is stochastic and equal to the sum of the deterministic and stochastic part of utility: U ij = V ij + ε ij In this paper we specify V ij as a linear function of observable variables: V ij = X ij β + Z j γ where X ij and Z j are vectors of explanatory variables and β and γ are vectors of the unknown parameters to be estimated. X ij are explanatory variables interacting characteristics of the alternative j with characteristics of consumer i, and Z j are explanatory variables describing characteristics or attributes of alternative j. In contrast with X ij, Z j does not show any variation between decision makers, or in other words, there are no within alternative variation in Z j. Examples of variables in the X ij vector are dummies indicating travel distance between decision maker and the alternative GP, and dummies indicating whether or not the decision maker and the alternative GP have the same gender. Examples of variables in the Z j vector are dummies indicating whether or not the GP is married, and whether or not the GP is born in Norway. The probability that consumer i choose GP j can be expressed as: P ij = P rob(v ij + ε ij > V ik + ε ik ; j k) = P rob(ε ik < ε ij + V ij V ik +; j k) A property of the logit model is that we get closed form expressions for P ij. It can be shown that the logit probabilities are given by: P ij = ev ij k e V ik We now let y ij = 1 if individual i chose alternative j, and zero otherwise. Assuming that the choice of GP made by consumer i is independent of choices made by other consumers, we may express the likelihood function by: L(β, γ) = i (P ij ) y ij j This is an application of McFaddens (1973, 1974) choice model, and by means of a conditional logistic regression model, we will estimate the parameters β and γ. 3

5 The estimation method applied is the maximum likelihood method available in the software STATA version Matching of GPs and consumers Consumers visit their GP in events of illness. The first of the GP s important tasks is to reveal what is the cause of illness, i.e. what is the diagnosis. A second important task is to recommend an appropriate treatment, and ensure that the patient is compliant with the treatment. Communication between the GP and the patient is an important input in both these processes. If the information transmission is unstrained and the GP and the patient are able to communicate easily and understand each other, than the GP may be more likely to succeed in setting the correct diagnosis than if the converse was true. One may also argue that unstrained communication may cause the treatment to be more effective, since patients may be more likely to be compliant with treatment if the patient receives and understand the information given with regard to the treatment. One might expect that communicative skills vary both between patients and between physicians. Further it seems reasonable that all patients do not communicate equally well with all types of physicians. Correspondingly, a particular GP may communicate better with a patient having certain characteristics. If conversation is an important input in the GPs health production function, then, matching a GP with patients which with whom the GP are able to communicate with ease may improve the GP s productivity compared to the converse case where a GP is matched with patients such that communication is more difficult. In the Norwegian regular general practitioner scheme the consumer and potential patient revealed their preferences for GP by means of returning an entry form ranking their three most preferred GPs in descending order. A large share of the inhabitants only filled in the most preferred GP. 1 In this paper we refer to the most preferred GP as the selected GP, and we assume that the consumer selected the GP that maximized utility. We assume that the consumers when selecting their GP appreciate that communication is an important input in general practice and that the consumer thus preferred GPs with 1 In the next version of the paper we shall state the proportion of inhabitants who only filled in the most preferred GP. 4

6 attributes such that the communication was expected to be unstrained. In this paper we apply the idea that people that have similar characteristics communicate better than people that are different. We thus apply the old saying that birds of a feather flock together, and state our main hypothesis that patients preferred GPs that resemble them selves on observable characteristics such as gender and age. We construct a dummy variable, SAMEGENDER, equal to one if the GP and patients are of the same gender and zero otherwise. We expect SAMEGENDER to have a positive effect on choice probabilities. Further we compute the absolute value of the age difference between the patient and the GP, AGEDIFF. Since an increase in AGEDIFF implies that that the patient and GP are more unlike, we expect the effect of AGEDIFF on choice probabilities to be negative. We also take account of the possibility that the consumers may not be indifferent to whether the GP is older or younger than oneself by including the GPs age, AGEGP, as an explanatory variable. Since travel is costly, we expect that GPs with practices that are located close to the consumer s residential address are preferred to GPs located further away. The Norwegian Mail Service has given every address in the country a four digit zip code. We have information on the zip code of both the GPs practices, and the residential address of consumers. We have access to a matrix containing travel distance in kilometers and and travel time in hours between all Norwegian zipcodes. We expect the choice probabilities to be decreasing in travel time and travel distance. By applying information on travel times and travel distances we will also calculate different travel costs associated with different modes of travel. 2 The GPs accessibility in terms of opening hours is an important attribute of the GP. We expect that GPs who have many office hours during the week are preferred to GPs that are working part time. Unfortunately we do not have information on the number of weekly business hours. Prior to the implementation of the regular general practitioner scheme the GPs was asked to state the maximum number of patients they would like to take care of. This number was to constitute an upper limit on the number of patients that the health authorities would allocate 2 The company Infomap Norway has collected actual travel distances and travel times associated with travel by means of a light truck on public roads between centers of the zip-code areas. 5

7 to the GPs, and this number has often been interpreted as the GPs preferred list size (Iversen and Lurås 2000, Iversen and Lurås 2002, Iversen 2004). It seems reasonable that GPs who were planning to have a part time practice on average reported a lower preferred list size compared to GPs that were planning to have a full time practice. We thus apply the GPs preferred list size, PREFLIST, as an indicator of the GPs accessibility, and we expect this variable to have a positive effect on the choice probabilities. After receiving a medical licence, GPs may undertake further medical training and become a specialist in general medicine. We expect that GPs who are specialists in general medicine is preferred to GPs who are not specialists, ceteris paribus. We thus include a dummy, SPECIALIST, equal to one if the GP is a specialist in general medicine and zero otherwise. Our data contain information on the birth country of the GP. Most inhabitants of Oslo are born in Norway. By following the reasoning that consumers prefer GPs that resemble themselves on observable characteristics, we expect consumers to prefer GPs that are born in Norway. We introduce a dummy, FOREIGNGP, that equals one if the GP was born outside Norway and zero otherwise. We expect FOREIGNGP to have a negative effect on the choice probabilities. One may argue that persons with unfavorable personal characteristics are less likely to be married. Therefore people with unfavorable characteristics are likely to be underrepresented among people who are married. We include a dummy (MARRIEDGP) equal to one if the GP is married. We expect MARRIEDGP to have a positive effect on the choice probabilities. 4 Data Our data set is provided by the Norwegian Social Science Data Services. The observation unit is the individual inhabitant. All inhabitants in 14 selected Norwegian municipalities are included. The sample is stratified according to Statistics Norway s measure of the centrality of Norwegian municipalities. In this paper we will only use observations from inhabitants and GPs in Oslo. One of the reasons for this is that the data from Oslo render the possibility of including indicators of travel distance to the GP in the analysis by utilizing information on zip codes. 6

8 After taking out inhabitants that is registered to be living in municipalities outside Oslo, our data set has unique observations. Not all inhabitants returned the entry form revealing their preferences for GP, and consequently information on the preferred GPs is missing for some of the inhabitants. We note that inhabitants who refrained from returning the entry form most likely differ from inhabitants who returned the form on both observable and unobservable characteristics. We will not address this issue in this version of the paper. We note however that the representativety of our sample have implications for how we interpret the results. This issue will receive more attention in the next version of this paper. When we take out all inhabitants that are not registered with a first choice GP, observations disappear, and we are left with observations. Not everyone who disappears refrained from returning the entry form observations disappear despite the fact that they actually returned the entry form. Among these persons, have been assigned GPs that have practices outside of the 14 municipalities included in our sample. Most likely the number of inhabitants who preferred a GP outside of these 14 municipalities is even higher. If an inhabitant requested a GP outside of the 14 municipalities, this could explain the fact that information on the first choice GP is missing. When the inhabitants returned the entry form, the inhabitants had the opportunity of refraining from participating in the regular general practitioner scheme all together who disappeared when removing inhabitants without a registered first choice GP have requested not to participate in the regular general practitioner scheme. The rest of the inhabitants that disappear despite the fact that they returned the entry form might be due to problems with registration, such as problems with optic reading of paper forms. We are planning to use the zip code regions in Oslo in our analysis. The zip code starting with 00 are reserved for special addresses such as H.M. The King, the Royal Castle, the National Hospital, The Norwegian Mail Service and the Norwegian Parliament. When we take out observations with zip code codes starting with 00, 34 observations disappear and we are left with observations. We are interested in studying the choice of sovereign consumers. Since parents are likely to choose the GP for their children we exclude observations of consumers younger than 18 years old. When we do this observations disappear, and we are left with observations. Last we take out observations of inhabitants 7

9 who selected a GP with a practice address outside of Oslo. When we do this 95 inhabitants disappear and we are left with a sample of observations. In this large sample there are 437 alternative GPs, meaning that 437 GPs have been ranked as the most preferred GP by at least one inhabitant. This sample is still too big to be used by means of the available computer resources. 3 We thus need to perform our estimation on a sub sample. Since zip code regions play an important role in the analysis we draw our sample from within each of the twelve zip code regions. When stratifying the sample according to zip code regions we also reduce the risk of alternatives falling out of the sample. We ask the program SPSS to draw a random sample of 10 % of the observations from within each of the zip code regions. This sub sample consists of unique observations. We check whether the construction of a sub sample has resulted in some of the alternatives falling out of the sample. It turns out that there are still 437 alternatives in the sub sample. In table 1 we present some descriptive statistics on variables at the patient level and compare the whole sample with sub sample. 3 The most important constraint seems to be the amount of memory (RAM) the computer can provide to the STATA program. The computer used for estimation is able to provide at most 938 MB of memory to the program. Even when variables are compressed to the minimum, the available memory impose a binding constraint on both the size of the estimation sample and the number of explanatory variables that can be included. 8

10 Table 1: Variables describing the consumer. Variable Sample N Mean Std. Dev. Min Max AGE Gross sample FEMALE AGE Sub sample FEMALE We see that the average age of consumers within the sample is 48 years, and that 58 % of the consumers in the sample are females. The high proportion of females might seem a bit surprising. One of the reasons for this is that a higher proportion of females returned the entry form compared to what was the case among male inhabitants. Another possible explanation is that females have higher life expectancy. We have excluded persons younger than 18 from the sample, contributing to inflating the proportion of females. The high proportion of females can also be caused by a higher proportion of females moving from rural areas to the city. In tables 2a and 2b we describe variables where consumer and GP characteristics are interacted. In table 2a we de describe the variables interacting the characteristics of consumers and GPs across all combinations of consumers and alternative GPs. We se that the average age difference between consumer and the selected GP is 16 years. We see that the average traveltime across alternatives is 0.14 hours or about 8 minutes, and that the average travel distance across alternatives is 8.55 kilometers. Table 2A: Variables interacting the characteristics of consumers and GPs. No of obs= Variable Mean Std. Dev. Min Max AGEDIFF SAMEGENDER TRAVELTIME (hrs) TRAVELDISTANCE (km) In table 2B we de describe the variables interacting the characteristics of consumers and the chosen GP. Comparing with table 2A we see that the age difference 9

11 between consumer and chosen GP is smaller than the average across all alternatives. We see that 59 % of the sample selected a GP with the same gender. We also see that the travel distance measured in both time and kilometers between consumers and the chosen altarnative is considerably shorter. While the average traveltime across all alternatives is 0.14 hours the average traveltime across chosen alternatives is 0.04 hours or 3.4 minutes. We also see that average travel distance across the chosen alternatives is only 2.58 kilometers. Table 2B: Descriptive statistics of variables interacting the characteristics of consumers and the chosen alternative. No of obs=25345 Variable Mean Std. Dev. Min Max AGEDIFF SAMEGENDER TRAVELTIME (hrs) TRAVELDISTANCE (km) The Norwegian Mail Service refers to the two first digits in this code as the zip code region. In Oslo there are 12 different zip code regions 4 : 01, 02, 03,..., 12. For a consumer living in zip code region 2, it is an important attribute of the GP whether the GPs practice is located in zip code region 2 or in an other zip code region such as zip code region 10. In table 3 we present a cross tabulation of the zip code region of the practice address and the zip code region of the consumers residential address. We have highlighted the elements on the main diagonal. We see that the proportion of consumers that have selected a GP located in the same zip code region as the residential address is high across all the zip code regions. We also see that very few of the cells are empty. The fact that few consumers selected a GP located outside of the zip code region is somewhat surprising knowing that many of the zip code regions are within walking distance to each other. 4 if we ignore the zip code region 00 which is reserved for special addresses 10

12 Table 3: Cross tabulation of Zip-codes of residential address of consumer and address of the selected GP. Zip consumers total Zip GPs total In table 4 we describe variables at the level of the GP. We see that 20 % of GPs in Oslo are born in a country other than Norway. Further we see that the average size of the GPs preferred list size is We also see that 38 % of the GPs are females, and that the average age of GPs in Oslo is 47 years. The mean of SPECIALIST indicate that 53% of the GPs in Oslo are specialists in general medicine. We also see that 66 % of the GPs are married. Table 4: Descriptive statistics of variables at the level of the GP. No of obs=437 Variable Mean Std. Dev. Min Max SPECIALIST PREFLIST AGEGEP FEMALEGP FOREIGNGP MARRIEDGP

13 5 Estimation and results In table 5 we present the results from conditional logistic regression where the dependent variable indicate the GP ranked as the consumers first choice of GP. We present the results from two different models, correspondingly with travel indicated by the distance in kilometers and travel indicated by the travel time measured in hours. We see that all the estimated coefficients are statistically significant. If we look at the estimated effect of SPECIALIST in model 1, we see that being a specialist in general medicine has a positive effect on the choice probability. Further we see that GPs that was born in a country other than Norway ceteris paribus have lower choice probabilities compared to GPs that were born in Norway. We see that our results indicate that marital status of the GP also affect the choice probabilities and the effect seems to be of simmilar magnitude as the effect of specialist status. We see that our results also seem to indicate that individuals prefer GPs of the same gender. We see that if the consumer is a female (male), the probability of selecting a GP that is also a female (male) is higher than the probability of selecting a male (female) GP. given that all the other attributes are the same. We see that the coefficient on AGEDIFF is negative. Our results indicate that choice probabilities are declining in the age difference between the consumer and the GP. We see however that our results indicate that choice probabilities, when controlling for the age difference, are increasing in the GPs age. An interpretation of the joint result that choice probabilities are declining in age difference and increasing in the age of the GP is that consumer prefer GPs that are of approximately the same age, but a GP that is older than themselves are preferred to a GP that is younger then themselves. We see that the estimated coefficient on the indicator of accessibility, PREFLIST is positive as expected. Our interpretation is that a longer weekly opening hours as indicated by a longer preferred list, is a positive attribute of the GP. We see that the effect of increasing travel distance by a marginal kilometer is sizable compared to the other coefficients. We see that the consumer values specialist status in general medicine to the equivalent of avoiding 300 meters of travel. Calculating the marginal rate of substitution between travel time and specialist status in general medicine we find that the consumer values a specialist to the equivalent of avoiding approximately twenty seconds of travel time. 12

14 Table 5: Results from conditional logistic regression. No of obs= Distance model Traveltime model CHOICE Coef. Std. Err. Coef. Std. Err. SPECIALIST ** ** PREFLIST ** ** AGEGP ** ** AGEDIFF ** ** SAMEGENDER ** ** FOREIGNGP ** ** MARRIEDGP ** ** TRAVELDISTANCE ** TRAVELTIME ** OSLO ** ** OSLO ** ** OSLO ** ** OSLO ** ** OSLO ** ** OSLO ** ** OSLO ** ** OSLO ** ** OSLO * ** OSLO * ** OSLO ** ** Loglikelihood Pseudo R LR chi2(19) P-value Willingnes to pay for attributes of the GP Since travel distance is an attribute that may be easier to measure in monetary units than other attributes, a lot of studies within the field of environmental economics apply recreation demand models and the so called travel cost method as an alternative to contingent valuation methods when the aim is to get estimates on 13

15 the willingness to pay for for example changes in attributes of public goods. A survey on recreation demand models in environmental economics is given in Phaneuf and Smith (2005) while a survey on applications of travel cost methods is given in Parsons (2003). We are now ready to estimate the willingness-to pay for attributes of the GP. Remembering that the ratio of two coefficients has the interpretation of marginal rate of substitution between the two attributes, all we need in order to get a meassure of the WTP for the various attributes is a measure of the travel costs. There are a lot of difficulties with this procedure. One of the difficulties is that traveling when healthy and traveling when ill may not be the same. It is not obvious whether traveling costs are higher when ill due to the discomfort associated with traveling in the event of illness, or whether it is lower because the marginal utility of leisure is lower when the consumer is on sick-leave. Never the less travel involves both the use of time, as well as travel expenses. In the following the value of the consumers time use is set to 100 NOK per hour. We now suggest two different travel modes corresponding to one low-cost and one high-cost alternative. High cost: Taxi One suggestion on how to measure the cost of travel within an urban area in a way that incorporates both the cost of time use and cost related to distance is to measure travel costs by means of taxi fares. Taxi fares are calculated as a linear combination of time- and distance rates. In this version of the paper we apply the Taxi rate fee schedule from 2005, and discount the rates by 21.2% as indicated by the Taxi cost index provided by the the Norwegian Association of Taxi Drivers. This results in the following travel cost function: Taxi= 28 nok nok*kilometers + [ nok nok]*hours Low cost: Private Car Another suggestion is to apply the compensation rate of 3.20 NOK per kilometers that is paid to public employees when they use their private car in work related activities. This results in the following travel cost function: 14

16 Private car = 3.2 nok*kilometers nok*hours In table 6A and 6B we present a description of the two travelcost variables associated with correspondingly the chosen alternative and all alternatives. Table 6A: Travelcosts associated with the chosen alternative. No of obs=25345 Variable Obs Mean Std. Dev. Min Max PRIVATE CAR TAXI Table 6B: Travelcosts across all alternatives. No of obs= Variable Obs Mean Std. Dev. Min Max PRIVATE CAR TAXI The results from conditional logistic regression with the two suggested travel costs variables are given in table 7. 15

17 Table 7: Results from conditional logistic regression with travel costs. No of obs= Private car model Taxi model choice Coef. Std. Err. Coef. Std. Err. SPECIALIST ** ** PREFLIST ** ** AGEGP ** ** AGEDIFF ** ** SAMEGENDER ** ** FOREIGNGP ** ** MARRIEDGP ** ** PRIVATE CAR ** TAXI ** oslo ** ** oslo ** ** oslo ** ** oslo ** ** oslo ** ** oslo ** ** oslo ** ** oslo ** ** oslo ** ** oslo ** ** oslo ** ** Log likelihood Pseudo R LR chi2(19) P-value From the estimated coefficients in table 7 we are able to calculate the representative consumer s willingnes-to-pay for partial changes in the attributes of GPs. The results from these calculations are given in table 8. 16

18 Table 8: Estimates of willingness-to-pay for partial changes in GP attributes (Norwegian kroner). Travel costs Low (Private car) High (Taxi) SPECIALIST PREFLIST AGEGP AGEDIFF SAMEGENDER FOREIGNGP MARRIEDGP If the suggested travel costs are representative for a representative consumer travelling to the GP, the willingness to pay for consulting a specialist is in the range between 1.52nok to 4.84NOK per visit. We note that when the general practitioner scheme was implemented the patients paid the same out of pocket fee when consulting a specialist in general medicine as when consulting a GP that is not a specialist. From July 2005 the patients needed to pay 25 kroner extra per consultation when seeing a GP with specialization compared to a GP without the specialization. Our results indicate that a representative individual s willingness-to-pay for consulting a GP with specialist status might be considerably lower than the new out of pocket fee introduced. We see that an estimate of the willingness to pay for not having to go to a non-norwegian GP is within the range 2.85nok to 9.02NOK. We see that the consumer has a higher willingness to pay for consulting a GP that has the same gender as the consumer as compared to the willingness to pay for consulting a specialist in general practice. We see that the estimated willingness to pay for consulting a a GP with the same gender is within the range 3.53nok to 11.19NOK. 6 Discussion and conclusion The value or importance that decision makers place on each attribute of the alternatives will in general show variation over decision makers. A weakness of the 17

19 logit model is that it can not handle random taste variation. The logit model can however represent taste variation of decision makers as long as tastes variation is related to observable characteristics (Train, 2003). In this paper we have handled some taste variation, by taking account of the possibility that attributes of a GP such as the GPs gender does not affect a male decision maker in the same way as a female decision maker. Most likely however there is a lot of taste variation that is not taken account of in this paper. For instance the taste for GPs that are specialists in general medicine is likely to vary across decision makers with different income levels and different levels of education. In the data set applied in this paper, we are fortunate to have information on the decision makers income, wealth and education available. Unfortunately, these variables can not be compressed as much as dummy variables, and in this version of the paper we had to delete these variables due to limits on the available computer resources. In the next version of the paper, heterogeneity in taste that relates to income, wealth and education of decision makers will be given more attention. In this paper we have devoted little attention to the fact that it was voluntary to return the entry form ranking the three most preferred GPs. If the set of decision makers who returned the entry form differs systematically from those who refrained from returning the entry form, we must be careful with interpreting the effect on the choice probabilities as effects on market demand. If there are heterogeneity in tastes such that the inhabitants who refrained from returning the entry form are indifferent to for instance the GPs birth country, our approach would exaggerate the effect of birth country on the market demand facing the GP. In the next version of the paper, the aim is to also estimate the effect of inhabitants characteristics on the probability of submitting the entry form and thus the probability of entering the estimation sample applied in this version of the paper. Our results confirm all the results by Lurås (2003). Lurås found that choice probabilities are higher when the GP is a specialist in general medicine and when the GP and patients have the same gender. Lurås also found a positive effect of the GPs preferred list size and of the GPs age, while the effect of age difference between patient and GP was found to be negative. It seems reasonable to conclude that consumer s choice of GP was not random. Further there is evidence in the data suggesting that consumers prefer GPs that 18

20 resemble themselves on observable characteristics. Our somewhat rough calculation of the willingness to pay for consulting a specialist in general medicine seems to indicate that the willingness to pay is lower than the extra fee introduced in From welfare theory we know that if the marginal income of the suppliers exceeds the consumers willingness to pay, we are in a sub optimal equilibrium with excess supply. An implication of the results in this study may thus be that the supply of specialists in general medicine now and in the future is likely to be higher than what is socially optimal. 19

21 References Dietrich, A.J.and K.I. Marton (1982): Does continuous care from a physician make a difference? Journal of Family Practice 15: Hjortdahl, P., and E. Laerum (1992): Continuity of care in general practice: Effect on patient satisfaction. British Medical Journal 304: Iversen, T., and H. Lurås (2000): Economic Motives and Professional Norms: The Case of General Medical Practice. Journal of Economic Behavior and Organization 43, Iversen, T., and H. Lurås (2002):Legemangelen som ble til pasientmangel: Variasjoner i listensker og pasientknapphet ved innfring av fastlegeordning. Økonomisk Forum (8) (In Norwegian) Iversen, T. (2004): The Effects of a Patient Shortage on General Practitioners Future Income and List of Patients. Journal of Health Economics 23, Kaplan, S.H., S. Greenfield and J.E. Ware (1989): Impact of the doctor-patient relationship on the outcomes of chronic disease. in: M. Stewart and D. Roter, eds., Communicating With Medical Patients, Sage Publications, Newbury Park, CA Lurås, H., (2003): Individuals preferences for GPs Choice analysis from the establishment of a list patient system in Norway. HERO Working Paper 2003: 5 McFadden, D., (1973): Conditional Logit Analysis of Qualitative Choice Behavior. In P. Zarembka (Ed.), Frontiers in Econometrics, New York, Academic McFadden, D., (1974): The Measurement of Urban Travel Demand. Journal of Public Economics 3: Ong, L.M., J.C. Haes, A.M. Hoos and F.B. Lammes (1995): Doctor-patient communication: A review of the Literature. Social Science and Medicine 40: Parsons, G.R. (2003): The travel cost model. In: Boyle, K., Peterson, G. (Eds.), A Primer on Non-Market Valuation. Kluwer Academic Publishers, Dordrecht. Phaneuf, D.J., and V.K. Smith (2005): Recreation Demand Models. In K-G Mler and J.R. Vincent (Eds) Handbook of Environmental Economics, Volume 2, Elsevier Science Scott, A. (2000): Economics of general practice. In A.J. Culyer and J.P. Newhouse (Eds.) Handbook of Health Economics, Volume 1, Elsevier Science Scott, A. and S. Vick (1998): Agency in health care. Examining patients preferences for attributes of the doctor-patient relationship. Journal of Health Economics 17:

22 Scott, A. and S. Vick (1999): Patients, doctors and contracts: An application of principalagent theory to the doctor-patient relationship. Scottish Journal of Political Economy 46: Stewart, M.A. (1995): Effective physician communication and health outcomes: A review. Canadian Medical Association Journal 152: Train, K. E., (2003): Discrete Choice Methods with Simulation. Cambridge university press Veale et al. (1995): Consumer use of multiple general practitioners: An Australian epidemiological study. Family Practice 12: Wartman, S.A., L.L. Morlock, EE. Malitz et al. (1983): Patient understanding and satisfaction as predictors of compliance. Medical Care 21: Williamson, O.E. (1993): Opportunism and its critics. Managerial and Decision Economics 14:

Multinomial and Ordinal Logistic Regression

Multinomial and Ordinal Logistic Regression Multinomial and Ordinal Logistic Regression ME104: Linear Regression Analysis Kenneth Benoit August 22, 2012 Regression with categorical dependent variables When the dependent variable is categorical,

More information

is paramount in advancing any economy. For developed countries such as

is paramount in advancing any economy. For developed countries such as Introduction The provision of appropriate incentives to attract workers to the health industry is paramount in advancing any economy. For developed countries such as Australia, the increasing demand for

More information

Poisson Models for Count Data

Poisson Models for Count Data Chapter 4 Poisson Models for Count Data In this chapter we study log-linear models for count data under the assumption of a Poisson error structure. These models have many applications, not only to the

More information

UNIVERSITY OF OSLO HEALTH ECONOMICS RESEARCH PROGRAMME

UNIVERSITY OF OSLO HEALTH ECONOMICS RESEARCH PROGRAMME UNIVERSITY OF OSLO HEALTH ECONOMICS RESEARCH PROGRAMME The impact of accessibility on the use of specialist health care in Norway Tor Iversen & Gry Stine Kopperud Center for Health Administration, University

More information

Chapter Seven. Multiple regression An introduction to multiple regression Performing a multiple regression on SPSS

Chapter Seven. Multiple regression An introduction to multiple regression Performing a multiple regression on SPSS Chapter Seven Multiple regression An introduction to multiple regression Performing a multiple regression on SPSS Section : An introduction to multiple regression WHAT IS MULTIPLE REGRESSION? Multiple

More information

Yew May Martin Maureen Maclachlan Tom Karmel Higher Education Division, Department of Education, Training and Youth Affairs.

Yew May Martin Maureen Maclachlan Tom Karmel Higher Education Division, Department of Education, Training and Youth Affairs. How is Australia s Higher Education Performing? An analysis of completion rates of a cohort of Australian Post Graduate Research Students in the 1990s. Yew May Martin Maureen Maclachlan Tom Karmel Higher

More information

Standard errors of marginal effects in the heteroskedastic probit model

Standard errors of marginal effects in the heteroskedastic probit model Standard errors of marginal effects in the heteroskedastic probit model Thomas Cornelißen Discussion Paper No. 320 August 2005 ISSN: 0949 9962 Abstract In non-linear regression models, such as the heteroskedastic

More information

Food Demand Survey (FooDS) Technical Information on Survey Questions and Methods. May 22, 2013. Jayson L. Lusk

Food Demand Survey (FooDS) Technical Information on Survey Questions and Methods. May 22, 2013. Jayson L. Lusk Food Demand Survey (FooDS) Technical Information on Survey Questions and Methods May 22, 2013 Jayson L. Lusk The purpose of FooDS is to track consumer preferences and sentiments on the safety, quality,

More information

The Wage Return to Education: What Hides Behind the Least Squares Bias?

The Wage Return to Education: What Hides Behind the Least Squares Bias? DISCUSSION PAPER SERIES IZA DP No. 8855 The Wage Return to Education: What Hides Behind the Least Squares Bias? Corrado Andini February 2015 Forschungsinstitut zur Zukunft der Arbeit Institute for the

More information

LOGISTIC REGRESSION ANALYSIS

LOGISTIC REGRESSION ANALYSIS LOGISTIC REGRESSION ANALYSIS C. Mitchell Dayton Department of Measurement, Statistics & Evaluation Room 1230D Benjamin Building University of Maryland September 1992 1. Introduction and Model Logistic

More information

UNIVERSITY OF WAIKATO. Hamilton New Zealand

UNIVERSITY OF WAIKATO. Hamilton New Zealand UNIVERSITY OF WAIKATO Hamilton New Zealand Can We Trust Cluster-Corrected Standard Errors? An Application of Spatial Autocorrelation with Exact Locations Known John Gibson University of Waikato Bonggeun

More information

I L L I N O I S UNIVERSITY OF ILLINOIS AT URBANA-CHAMPAIGN

I L L I N O I S UNIVERSITY OF ILLINOIS AT URBANA-CHAMPAIGN Beckman HLM Reading Group: Questions, Answers and Examples Carolyn J. Anderson Department of Educational Psychology I L L I N O I S UNIVERSITY OF ILLINOIS AT URBANA-CHAMPAIGN Linear Algebra Slide 1 of

More information

Wooldridge, Introductory Econometrics, 4th ed. Chapter 7: Multiple regression analysis with qualitative information: Binary (or dummy) variables

Wooldridge, Introductory Econometrics, 4th ed. Chapter 7: Multiple regression analysis with qualitative information: Binary (or dummy) variables Wooldridge, Introductory Econometrics, 4th ed. Chapter 7: Multiple regression analysis with qualitative information: Binary (or dummy) variables We often consider relationships between observed outcomes

More information

Mobility Tool Ownership - A Review of the Recessionary Report

Mobility Tool Ownership - A Review of the Recessionary Report Hazard rate 0.20 0.18 0.16 0.14 0.12 0.10 0.08 0.06 Residence Education Employment Education and employment Car: always available Car: partially available National annual ticket ownership Regional annual

More information

Multiple Choice Models II

Multiple Choice Models II Multiple Choice Models II Laura Magazzini University of Verona laura.magazzini@univr.it http://dse.univr.it/magazzini Laura Magazzini (@univr.it) Multiple Choice Models II 1 / 28 Categorical data Categorical

More information

Aileen Murphy, Department of Economics, UCC, Ireland. WORKING PAPER SERIES 07-10

Aileen Murphy, Department of Economics, UCC, Ireland. WORKING PAPER SERIES 07-10 AN ECONOMETRIC ANALYSIS OF SMOKING BEHAVIOUR IN IRELAND Aileen Murphy, Department of Economics, UCC, Ireland. DEPARTMENT OF ECONOMICS WORKING PAPER SERIES 07-10 1 AN ECONOMETRIC ANALYSIS OF SMOKING BEHAVIOUR

More information

A Study to Predict No Show Probability for a Scheduled Appointment at Free Health Clinic

A Study to Predict No Show Probability for a Scheduled Appointment at Free Health Clinic A Study to Predict No Show Probability for a Scheduled Appointment at Free Health Clinic Report prepared for Brandon Slama Department of Health Management and Informatics University of Missouri, Columbia

More information

How to set the main menu of STATA to default factory settings standards

How to set the main menu of STATA to default factory settings standards University of Pretoria Data analysis for evaluation studies Examples in STATA version 11 List of data sets b1.dta (To be created by students in class) fp1.xls (To be provided to students) fp1.txt (To be

More information

PERFORMANCE MANAGEMENT AND COST-EFFECTIVENESS OF PUBLIC SERVICES:

PERFORMANCE MANAGEMENT AND COST-EFFECTIVENESS OF PUBLIC SERVICES: PERFORMANCE MANAGEMENT AND COST-EFFECTIVENESS OF PUBLIC SERVICES: EMPIRICAL EVIDENCE FROM DUTCH MUNICIPALITIES Hans de Groot (Innovation and Governance Studies, University of Twente, The Netherlands, h.degroot@utwente.nl)

More information

The Impact of the Medicare Rural Hospital Flexibility Program on Patient Choice

The Impact of the Medicare Rural Hospital Flexibility Program on Patient Choice The Impact of the Medicare Rural Hospital Flexibility Program on Patient Choice Gautam Gowrisankaran Claudio Lucarelli Philipp Schmidt-Dengler Robert Town January 24, 2011 Abstract This paper seeks to

More information

1. What is the critical value for this 95% confidence interval? CV = z.025 = invnorm(0.025) = 1.96

1. What is the critical value for this 95% confidence interval? CV = z.025 = invnorm(0.025) = 1.96 1 Final Review 2 Review 2.1 CI 1-propZint Scenario 1 A TV manufacturer claims in its warranty brochure that in the past not more than 10 percent of its TV sets needed any repair during the first two years

More information

Association Between Variables

Association Between Variables Contents 11 Association Between Variables 767 11.1 Introduction............................ 767 11.1.1 Measure of Association................. 768 11.1.2 Chapter Summary.................... 769 11.2 Chi

More information

How Sensitive are Low Income Families to Health Plan Prices?

How Sensitive are Low Income Families to Health Plan Prices? American Economic Review: Papers & Proceedings 100 (May 2010): 292 296 http://www.aeaweb.org/articles.php?doi=10.1257/aer.100.2.292 The Massachusetts Health Insurance Experiment: Early Experiences How

More information

Logistic Regression. Jia Li. Department of Statistics The Pennsylvania State University. Logistic Regression

Logistic Regression. Jia Li. Department of Statistics The Pennsylvania State University. Logistic Regression Logistic Regression Department of Statistics The Pennsylvania State University Email: jiali@stat.psu.edu Logistic Regression Preserve linear classification boundaries. By the Bayes rule: Ĝ(x) = arg max

More information

Social Security Eligibility and the Labor Supply of Elderly Immigrants. George J. Borjas Harvard University and National Bureau of Economic Research

Social Security Eligibility and the Labor Supply of Elderly Immigrants. George J. Borjas Harvard University and National Bureau of Economic Research Social Security Eligibility and the Labor Supply of Elderly Immigrants George J. Borjas Harvard University and National Bureau of Economic Research Updated for the 9th Annual Joint Conference of the Retirement

More information

Institut für Soziologie Eberhard Karls Universität Tübingen www.maartenbuis.nl

Institut für Soziologie Eberhard Karls Universität Tübingen www.maartenbuis.nl from Indirect Extracting from Institut für Soziologie Eberhard Karls Universität Tübingen www.maartenbuis.nl from Indirect What is the effect of x on y? Which effect do I choose: average marginal or marginal

More information

The Cheap-talk Protocol and the Estimation of the Benefits of Wind Power

The Cheap-talk Protocol and the Estimation of the Benefits of Wind Power The Cheap-talk Protocol and the Estimation of the Benefits of Wind Power Todd L. Cherry and John Whitehead Department of Economics Appalachian State University August 2004 1 I. Introduction The contingent

More information

"Statistical methods are objective methods by which group trends are abstracted from observations on many separate individuals." 1

Statistical methods are objective methods by which group trends are abstracted from observations on many separate individuals. 1 BASIC STATISTICAL THEORY / 3 CHAPTER ONE BASIC STATISTICAL THEORY "Statistical methods are objective methods by which group trends are abstracted from observations on many separate individuals." 1 Medicine

More information

Discussion Section 4 ECON 139/239 2010 Summer Term II

Discussion Section 4 ECON 139/239 2010 Summer Term II Discussion Section 4 ECON 139/239 2010 Summer Term II 1. Let s use the CollegeDistance.csv data again. (a) An education advocacy group argues that, on average, a person s educational attainment would increase

More information

11. Analysis of Case-control Studies Logistic Regression

11. Analysis of Case-control Studies Logistic Regression Research methods II 113 11. Analysis of Case-control Studies Logistic Regression This chapter builds upon and further develops the concepts and strategies described in Ch.6 of Mother and Child Health:

More information

Documents. Torbjørn Hægeland and Lars J. Kirkebøen

Documents. Torbjørn Hægeland and Lars J. Kirkebøen 2008/8 Documents Torbjørn Hægeland and Lars J. Kirkebøen Documents School performance and valueadded indicators - what is the effect of controlling for socioeconomic background? A simple empirical illustration

More information

ASSIGNMENT 4 PREDICTIVE MODELING AND GAINS CHARTS

ASSIGNMENT 4 PREDICTIVE MODELING AND GAINS CHARTS DATABASE MARKETING Fall 2015, max 24 credits Dead line 15.10. ASSIGNMENT 4 PREDICTIVE MODELING AND GAINS CHARTS PART A Gains chart with excel Prepare a gains chart from the data in \\work\courses\e\27\e20100\ass4b.xls.

More information

From the help desk: hurdle models

From the help desk: hurdle models The Stata Journal (2003) 3, Number 2, pp. 178 184 From the help desk: hurdle models Allen McDowell Stata Corporation Abstract. This article demonstrates that, although there is no command in Stata for

More information

Submission to the National Health and Hospitals Reform Commission (nhhrc).

Submission to the National Health and Hospitals Reform Commission (nhhrc). Submission to the National Health and Hospitals Reform Commission (nhhrc). A New Health Savings Based System for Australia. A new health savings based system is proposed based on the best aspects of the

More information

Statistical modelling with missing data using multiple imputation. Session 4: Sensitivity Analysis after Multiple Imputation

Statistical modelling with missing data using multiple imputation. Session 4: Sensitivity Analysis after Multiple Imputation Statistical modelling with missing data using multiple imputation Session 4: Sensitivity Analysis after Multiple Imputation James Carpenter London School of Hygiene & Tropical Medicine Email: james.carpenter@lshtm.ac.uk

More information

Elisa Iezzi* Matteo Lippi Bruni** Cristina Ugolini**

Elisa Iezzi* Matteo Lippi Bruni** Cristina Ugolini** 24-25 June 2010 Elisa Iezzi* Matteo Lippi Bruni** Cristina Ugolini** elisa.iezzi@unibo.it matteo.lippibruni2@unibo.it cristina.ugolini@unibo.it * Department of Statistics, University of Bologna **Department

More information

The Loss in Efficiency from Using Grouped Data to Estimate Coefficients of Group Level Variables. Kathleen M. Lang* Boston College.

The Loss in Efficiency from Using Grouped Data to Estimate Coefficients of Group Level Variables. Kathleen M. Lang* Boston College. The Loss in Efficiency from Using Grouped Data to Estimate Coefficients of Group Level Variables Kathleen M. Lang* Boston College and Peter Gottschalk Boston College Abstract We derive the efficiency loss

More information

Longitudinal Meta-analysis

Longitudinal Meta-analysis Quality & Quantity 38: 381 389, 2004. 2004 Kluwer Academic Publishers. Printed in the Netherlands. 381 Longitudinal Meta-analysis CORA J. M. MAAS, JOOP J. HOX and GERTY J. L. M. LENSVELT-MULDERS Department

More information

Market Efficient Public Transport? An analysis of developments in Oslo, Bergen, Trondheim, Kristiansand, and Tromsø

Market Efficient Public Transport? An analysis of developments in Oslo, Bergen, Trondheim, Kristiansand, and Tromsø TØI report 428/1999 Authors: Bård Norheim and Erik Carlquist Oslo 1999, 63 pages Norwegian language Summary: Market Efficient Public Transport? An analysis of developments in Oslo, Bergen, Trondheim, Kristiansand,

More information

Keep It Simple: Easy Ways To Estimate Choice Models For Single Consumers

Keep It Simple: Easy Ways To Estimate Choice Models For Single Consumers Keep It Simple: Easy Ways To Estimate Choice Models For Single Consumers Christine Ebling, University of Technology Sydney, christine.ebling@uts.edu.au Bart Frischknecht, University of Technology Sydney,

More information

BMJcareers. Informing Choices

BMJcareers. Informing Choices : The Need for Career Advice in Medical Training How should the support provided to doctors and medical students to help them make career decisions during their training be improved? Experience elsewhere

More information

SAMPLE DESIGN RESEARCH FOR THE NATIONAL NURSING HOME SURVEY

SAMPLE DESIGN RESEARCH FOR THE NATIONAL NURSING HOME SURVEY SAMPLE DESIGN RESEARCH FOR THE NATIONAL NURSING HOME SURVEY Karen E. Davis National Center for Health Statistics, 6525 Belcrest Road, Room 915, Hyattsville, MD 20782 KEY WORDS: Sample survey, cost model

More information

A Basic Introduction to Missing Data

A Basic Introduction to Missing Data John Fox Sociology 740 Winter 2014 Outline Why Missing Data Arise Why Missing Data Arise Global or unit non-response. In a survey, certain respondents may be unreachable or may refuse to participate. Item

More information

Testing Market Efficiency in a Fixed Odds Betting Market

Testing Market Efficiency in a Fixed Odds Betting Market WORKING PAPER SERIES WORKING PAPER NO 2, 2007 ESI Testing Market Efficiency in a Fixed Odds Betting Market Robin Jakobsson Department of Statistics Örebro University robin.akobsson@esi.oru.se By Niklas

More information

The Probit Link Function in Generalized Linear Models for Data Mining Applications

The Probit Link Function in Generalized Linear Models for Data Mining Applications Journal of Modern Applied Statistical Methods Copyright 2013 JMASM, Inc. May 2013, Vol. 12, No. 1, 164-169 1538 9472/13/$95.00 The Probit Link Function in Generalized Linear Models for Data Mining Applications

More information

Sensitivity Analysis 3.1 AN EXAMPLE FOR ANALYSIS

Sensitivity Analysis 3.1 AN EXAMPLE FOR ANALYSIS Sensitivity Analysis 3 We have already been introduced to sensitivity analysis in Chapter via the geometry of a simple example. We saw that the values of the decision variables and those of the slack and

More information

CHAPTER 2 Estimating Probabilities

CHAPTER 2 Estimating Probabilities CHAPTER 2 Estimating Probabilities Machine Learning Copyright c 2016. Tom M. Mitchell. All rights reserved. *DRAFT OF January 24, 2016* *PLEASE DO NOT DISTRIBUTE WITHOUT AUTHOR S PERMISSION* This is a

More information

From this it is not clear what sort of variable that insure is so list the first 10 observations.

From this it is not clear what sort of variable that insure is so list the first 10 observations. MNL in Stata We have data on the type of health insurance available to 616 psychologically depressed subjects in the United States (Tarlov et al. 1989, JAMA; Wells et al. 1989, JAMA). The insurance is

More information

Calculating the Probability of Returning a Loan with Binary Probability Models

Calculating the Probability of Returning a Loan with Binary Probability Models Calculating the Probability of Returning a Loan with Binary Probability Models Associate Professor PhD Julian VASILEV (e-mail: vasilev@ue-varna.bg) Varna University of Economics, Bulgaria ABSTRACT The

More information

Simple linear regression

Simple linear regression Simple linear regression Introduction Simple linear regression is a statistical method for obtaining a formula to predict values of one variable from another where there is a causal relationship between

More information

Marginal Effects for Continuous Variables Richard Williams, University of Notre Dame, http://www3.nd.edu/~rwilliam/ Last revised February 21, 2015

Marginal Effects for Continuous Variables Richard Williams, University of Notre Dame, http://www3.nd.edu/~rwilliam/ Last revised February 21, 2015 Marginal Effects for Continuous Variables Richard Williams, University of Notre Dame, http://www3.nd.edu/~rwilliam/ Last revised February 21, 2015 References: Long 1997, Long and Freese 2003 & 2006 & 2014,

More information

INDIRECT INFERENCE (prepared for: The New Palgrave Dictionary of Economics, Second Edition)

INDIRECT INFERENCE (prepared for: The New Palgrave Dictionary of Economics, Second Edition) INDIRECT INFERENCE (prepared for: The New Palgrave Dictionary of Economics, Second Edition) Abstract Indirect inference is a simulation-based method for estimating the parameters of economic models. Its

More information

ESTIMATING AVERAGE TREATMENT EFFECTS: IV AND CONTROL FUNCTIONS, II Jeff Wooldridge Michigan State University BGSE/IZA Course in Microeconometrics

ESTIMATING AVERAGE TREATMENT EFFECTS: IV AND CONTROL FUNCTIONS, II Jeff Wooldridge Michigan State University BGSE/IZA Course in Microeconometrics ESTIMATING AVERAGE TREATMENT EFFECTS: IV AND CONTROL FUNCTIONS, II Jeff Wooldridge Michigan State University BGSE/IZA Course in Microeconometrics July 2009 1. Quantile Treatment Effects 2. Control Functions

More information

SAS Software to Fit the Generalized Linear Model

SAS Software to Fit the Generalized Linear Model SAS Software to Fit the Generalized Linear Model Gordon Johnston, SAS Institute Inc., Cary, NC Abstract In recent years, the class of generalized linear models has gained popularity as a statistical modeling

More information

Modelling the Scores of Premier League Football Matches

Modelling the Scores of Premier League Football Matches Modelling the Scores of Premier League Football Matches by: Daan van Gemert The aim of this thesis is to develop a model for estimating the probabilities of premier league football outcomes, with the potential

More information

Generalized Linear Models

Generalized Linear Models Generalized Linear Models We have previously worked with regression models where the response variable is quantitative and normally distributed. Now we turn our attention to two types of models where the

More information

Multiple logistic regression analysis of cigarette use among high school students

Multiple logistic regression analysis of cigarette use among high school students Multiple logistic regression analysis of cigarette use among high school students ABSTRACT Joseph Adwere-Boamah Alliant International University A binary logistic regression analysis was performed to predict

More information

MULTIPLE REGRESSION WITH CATEGORICAL DATA

MULTIPLE REGRESSION WITH CATEGORICAL DATA DEPARTMENT OF POLITICAL SCIENCE AND INTERNATIONAL RELATIONS Posc/Uapp 86 MULTIPLE REGRESSION WITH CATEGORICAL DATA I. AGENDA: A. Multiple regression with categorical variables. Coding schemes. Interpreting

More information

Demand for Life Insurance in Malaysia

Demand for Life Insurance in Malaysia Demand for Life Insurance in Malaysia Yiing Jia Loke 1+ and Yi Yuern Goh 2 1 School of Social Sciences, Universiti Sains Malaysia 2 HSBC Bank, Penang. Abstract. The insurance sector in Malaysia has shown

More information

Agency in Health-Care: Are Medical Care-Givers Perfect Agents?

Agency in Health-Care: Are Medical Care-Givers Perfect Agents? DISCUSSION PAPER SERIES IZA DP No. 2727 Agency in Health-Care: Are Medical Care-Givers Perfect Agents? Einat Neuman Shoshana Neuman April 2007 Forschungsinstitut zur Zukunft der Arbeit Institute for the

More information

CHAPTER 14 ORDINAL MEASURES OF CORRELATION: SPEARMAN'S RHO AND GAMMA

CHAPTER 14 ORDINAL MEASURES OF CORRELATION: SPEARMAN'S RHO AND GAMMA CHAPTER 14 ORDINAL MEASURES OF CORRELATION: SPEARMAN'S RHO AND GAMMA Chapter 13 introduced the concept of correlation statistics and explained the use of Pearson's Correlation Coefficient when working

More information

Ordinal Regression. Chapter

Ordinal Regression. Chapter Ordinal Regression Chapter 4 Many variables of interest are ordinal. That is, you can rank the values, but the real distance between categories is unknown. Diseases are graded on scales from least severe

More information

Health Economics Demand for health capital Gerald J. Pruckner University of Linz & Lecture Notes, Summer Term 2010 Demand for health capital 1 / 31

Health Economics Demand for health capital Gerald J. Pruckner University of Linz & Lecture Notes, Summer Term 2010 Demand for health capital 1 / 31 Health Economics Demand for health capital University of Linz & Gerald J. Pruckner Lecture Notes, Summer Term 2010 Demand for health capital 1 / 31 An individual s production of health The Grossman model:

More information

Failure to take the sampling scheme into account can lead to inaccurate point estimates and/or flawed estimates of the standard errors.

Failure to take the sampling scheme into account can lead to inaccurate point estimates and/or flawed estimates of the standard errors. Analyzing Complex Survey Data: Some key issues to be aware of Richard Williams, University of Notre Dame, http://www3.nd.edu/~rwilliam/ Last revised January 24, 2015 Rather than repeat material that is

More information

Overview of Violations of the Basic Assumptions in the Classical Normal Linear Regression Model

Overview of Violations of the Basic Assumptions in the Classical Normal Linear Regression Model Overview of Violations of the Basic Assumptions in the Classical Normal Linear Regression Model 1 September 004 A. Introduction and assumptions The classical normal linear regression model can be written

More information

Statistical Machine Learning

Statistical Machine Learning Statistical Machine Learning UoC Stats 37700, Winter quarter Lecture 4: classical linear and quadratic discriminants. 1 / 25 Linear separation For two classes in R d : simple idea: separate the classes

More information

Logit Models for Binary Data

Logit Models for Binary Data Chapter 3 Logit Models for Binary Data We now turn our attention to regression models for dichotomous data, including logistic regression and probit analysis. These models are appropriate when the response

More information

Using Stata for Categorical Data Analysis

Using Stata for Categorical Data Analysis Using Stata for Categorical Data Analysis NOTE: These problems make extensive use of Nick Cox s tab_chi, which is actually a collection of routines, and Adrian Mander s ipf command. From within Stata,

More information

An assessment of consumer willingness to pay for Renewable Energy Sources use in Italy: a payment card approach.

An assessment of consumer willingness to pay for Renewable Energy Sources use in Italy: a payment card approach. An assessment of consumer willingness to pay for Renewable Energy Sources use in Italy: a payment card approach. -First findings- University of Perugia Department of Economics, Finance and Statistics 1

More information

Introduction to General and Generalized Linear Models

Introduction to General and Generalized Linear Models Introduction to General and Generalized Linear Models General Linear Models - part I Henrik Madsen Poul Thyregod Informatics and Mathematical Modelling Technical University of Denmark DK-2800 Kgs. Lyngby

More information

LOGISTIC REGRESSION. Nitin R Patel. where the dependent variable, y, is binary (for convenience we often code these values as

LOGISTIC REGRESSION. Nitin R Patel. where the dependent variable, y, is binary (for convenience we often code these values as LOGISTIC REGRESSION Nitin R Patel Logistic regression extends the ideas of multiple linear regression to the situation where the dependent variable, y, is binary (for convenience we often code these values

More information

STATISTICA Formula Guide: Logistic Regression. Table of Contents

STATISTICA Formula Guide: Logistic Regression. Table of Contents : Table of Contents... 1 Overview of Model... 1 Dispersion... 2 Parameterization... 3 Sigma-Restricted Model... 3 Overparameterized Model... 4 Reference Coding... 4 Model Summary (Summary Tab)... 5 Summary

More information

Michigan Department of Community Health

Michigan Department of Community Health Michigan Department of Community Health January 2007 INTRODUCTION The Michigan Department of Community Health (MDCH) asked Public Sector Consultants Inc. (PSC) to conduct a survey of licensed dental hygienists

More information

How To Compare Social Preferences Of The General Public With The National Health Service

How To Compare Social Preferences Of The General Public With The National Health Service Social Science & Medicine 64 (2007) 2499 2503 www.elsevier.com/locate/socscimed Do NHS clinicians and members of the public share the same views about reducing inequalities in health? Aki Tsuchiya a,,

More information

On Marginal Effects in Semiparametric Censored Regression Models

On Marginal Effects in Semiparametric Censored Regression Models On Marginal Effects in Semiparametric Censored Regression Models Bo E. Honoré September 3, 2008 Introduction It is often argued that estimation of semiparametric censored regression models such as the

More information

The Importance of Community College Honors Programs

The Importance of Community College Honors Programs 6 This chapter examines relationships between the presence of honors programs at community colleges and institutional, curricular, and student body characteristics. Furthermore, the author relates his

More information

ECON 142 SKETCH OF SOLUTIONS FOR APPLIED EXERCISE #2

ECON 142 SKETCH OF SOLUTIONS FOR APPLIED EXERCISE #2 University of California, Berkeley Prof. Ken Chay Department of Economics Fall Semester, 005 ECON 14 SKETCH OF SOLUTIONS FOR APPLIED EXERCISE # Question 1: a. Below are the scatter plots of hourly wages

More information

The performance of immigrants in the Norwegian labor market

The performance of immigrants in the Norwegian labor market J Popul Econ (1998) 11:293 303 Springer-Verlag 1998 The performance of immigrants in the Norwegian labor market John E. Hayfron Department of Economics, University of Bergen, Fosswinckelsgt. 6, N-5007

More information

VI. Real Business Cycles Models

VI. Real Business Cycles Models VI. Real Business Cycles Models Introduction Business cycle research studies the causes and consequences of the recurrent expansions and contractions in aggregate economic activity that occur in most industrialized

More information

Department of Economics Session 2012/2013. EC352 Econometric Methods. Solutions to Exercises from Week 10 + 0.0077 (0.052)

Department of Economics Session 2012/2013. EC352 Econometric Methods. Solutions to Exercises from Week 10 + 0.0077 (0.052) Department of Economics Session 2012/2013 University of Essex Spring Term Dr Gordon Kemp EC352 Econometric Methods Solutions to Exercises from Week 10 1 Problem 13.7 This exercise refers back to Equation

More information

Module 14: Missing Data Stata Practical

Module 14: Missing Data Stata Practical Module 14: Missing Data Stata Practical Jonathan Bartlett & James Carpenter London School of Hygiene & Tropical Medicine www.missingdata.org.uk Supported by ESRC grant RES 189-25-0103 and MRC grant G0900724

More information

The Effects of Start Prices on the Performance of the Certainty Equivalent Pricing Policy

The Effects of Start Prices on the Performance of the Certainty Equivalent Pricing Policy BMI Paper The Effects of Start Prices on the Performance of the Certainty Equivalent Pricing Policy Faculty of Sciences VU University Amsterdam De Boelelaan 1081 1081 HV Amsterdam Netherlands Author: R.D.R.

More information

Motorist valuation of traffic information Results of a stated-preference pilot study

Motorist valuation of traffic information Results of a stated-preference pilot study ITE report 537/2001 Authors: Marit Killi Hanne Samstad Kjartan Sælensminde Oslo 2001, 77 pages Summary: Motorist valuation of traffic information Results of a stated-preference pilot study Introduction

More information

Regression Analysis of the Relationship between Income and Work Hours

Regression Analysis of the Relationship between Income and Work Hours Regression Analysis of the Relationship between Income and Work Hours Sina Mehdikarimi Samuel Norris Charles Stalzer Georgia Institute of Technology Econometric Analysis (ECON 3161) Dr. Shatakshee Dhongde

More information

HURDLE AND SELECTION MODELS Jeff Wooldridge Michigan State University BGSE/IZA Course in Microeconometrics July 2009

HURDLE AND SELECTION MODELS Jeff Wooldridge Michigan State University BGSE/IZA Course in Microeconometrics July 2009 HURDLE AND SELECTION MODELS Jeff Wooldridge Michigan State University BGSE/IZA Course in Microeconometrics July 2009 1. Introduction 2. A General Formulation 3. Truncated Normal Hurdle Model 4. Lognormal

More information

Statistics in Retail Finance. Chapter 2: Statistical models of default

Statistics in Retail Finance. Chapter 2: Statistical models of default Statistics in Retail Finance 1 Overview > We consider how to build statistical models of default, or delinquency, and how such models are traditionally used for credit application scoring and decision

More information

CALCULATIONS & STATISTICS

CALCULATIONS & STATISTICS CALCULATIONS & STATISTICS CALCULATION OF SCORES Conversion of 1-5 scale to 0-100 scores When you look at your report, you will notice that the scores are reported on a 0-100 scale, even though respondents

More information

MISSING DATA TECHNIQUES WITH SAS. IDRE Statistical Consulting Group

MISSING DATA TECHNIQUES WITH SAS. IDRE Statistical Consulting Group MISSING DATA TECHNIQUES WITH SAS IDRE Statistical Consulting Group ROAD MAP FOR TODAY To discuss: 1. Commonly used techniques for handling missing data, focusing on multiple imputation 2. Issues that could

More information

Statistics 104 Final Project A Culture of Debt: A Study of Credit Card Spending in America TF: Kevin Rader Anonymous Students: LD, MH, IW, MY

Statistics 104 Final Project A Culture of Debt: A Study of Credit Card Spending in America TF: Kevin Rader Anonymous Students: LD, MH, IW, MY Statistics 104 Final Project A Culture of Debt: A Study of Credit Card Spending in America TF: Kevin Rader Anonymous Students: LD, MH, IW, MY ABSTRACT: This project attempted to determine the relationship

More information

Deterministic and Stochastic Modeling of Insulin Sensitivity

Deterministic and Stochastic Modeling of Insulin Sensitivity Deterministic and Stochastic Modeling of Insulin Sensitivity Master s Thesis in Engineering Mathematics and Computational Science ELÍN ÖSP VILHJÁLMSDÓTTIR Department of Mathematical Science Chalmers University

More information

The frequency of visiting a doctor: is the decision to go independent of the frequency?

The frequency of visiting a doctor: is the decision to go independent of the frequency? Discussion Paper: 2009/04 The frequency of visiting a doctor: is the decision to go independent of the frequency? Hans van Ophem www.feb.uva.nl/ke/uva-econometrics Amsterdam School of Economics Department

More information

Financial framework and development of the public transport sector in six Norwegian cities

Financial framework and development of the public transport sector in six Norwegian cities Summary: TØI report 752/2004 Author: Bård Norheim Oslo 2004, 88 pages Norwegian language Financial framework and development of the public transport sector in six Norwegian cities The Norwegian Ministry

More information

1 The total values reported in the tables and

1 The total values reported in the tables and 1 Recruiting is increasingly social and Adecco wants to know how it works. An international survey, that involved over 17.272 candidates and 1.502 Human Resources managers between March 18 and June 2,

More information

Stata Walkthrough 4: Regression, Prediction, and Forecasting

Stata Walkthrough 4: Regression, Prediction, and Forecasting Stata Walkthrough 4: Regression, Prediction, and Forecasting Over drinks the other evening, my neighbor told me about his 25-year-old nephew, who is dating a 35-year-old woman. God, I can t see them getting

More information

Lecture 19: Conditional Logistic Regression

Lecture 19: Conditional Logistic Regression Lecture 19: Conditional Logistic Regression Dipankar Bandyopadhyay, Ph.D. BMTRY 711: Analysis of Categorical Data Spring 2011 Division of Biostatistics and Epidemiology Medical University of South Carolina

More information

Statistics 2014 Scoring Guidelines

Statistics 2014 Scoring Guidelines AP Statistics 2014 Scoring Guidelines College Board, Advanced Placement Program, AP, AP Central, and the acorn logo are registered trademarks of the College Board. AP Central is the official online home

More information

Handling missing data in Stata a whirlwind tour

Handling missing data in Stata a whirlwind tour Handling missing data in Stata a whirlwind tour 2012 Italian Stata Users Group Meeting Jonathan Bartlett www.missingdata.org.uk 20th September 2012 1/55 Outline The problem of missing data and a principled

More information

Binary Logistic Regression

Binary Logistic Regression Binary Logistic Regression Main Effects Model Logistic regression will accept quantitative, binary or categorical predictors and will code the latter two in various ways. Here s a simple model including

More information

An Analysis of the Health Insurance Coverage of Young Adults

An Analysis of the Health Insurance Coverage of Young Adults Gius, International Journal of Applied Economics, 7(1), March 2010, 1-17 1 An Analysis of the Health Insurance Coverage of Young Adults Mark P. Gius Quinnipiac University Abstract The purpose of the present

More information

PARALLEL LINES ASSUMPTION IN ORDINAL LOGISTIC REGRESSION AND ANALYSIS APPROACHES

PARALLEL LINES ASSUMPTION IN ORDINAL LOGISTIC REGRESSION AND ANALYSIS APPROACHES International Interdisciplinary Journal of Scientific Research ISSN: 2200-9833 www.iijsr.org PARALLEL LINES ASSUMPTION IN ORDINAL LOGISTIC REGRESSION AND ANALYSIS APPROACHES Erkan ARI 1 and Zeki YILDIZ

More information