What is the Conjoint

Share

Background

I recently have had many chances to ask many different people out of my company and its contractors, whether they heard something about the conjoint analysis. By now, each time the answer was exact or similar to “Not, at all”. Moreover, none of the recipients intended to even ask what is it about.  Seeing that I started to feel very strange, since everyday in my job the conjoint is the most basic and important topic. The problem is how to explain, what it is, briefly and simultaneously accurately. Einstein might have (it is uncertain see this page) said “if you can’t explain it simply you don’t understand it well enough”, so let me try to do it.

Introduction

Conjoint analysis is a market research technique and nothing more.

No matter which approach is utilized, a typical market research process is consisted of the following steps. Firstly, having request from entity X, marketers define a problem to solve, then basing on that they collect data by involving selected group of respondents in filling specially designed surveys. After survey set is over, marketers analyse outcome using relevant statistical models. Then, actually the most important part,  they need to interpret analysed data correctly in order to finally present to the entity X a solution of the defined problem.        

Typical market research process

A typical market research process (http://ams-centre.com/)

Marketers utilize a conjoint technique to determine what attributes/features a new product/service/sth called concept should have and, mostly in business cases, how it should be priced. They do it by showing to respondents prepared set of tasks (design). There is always a one answer per task. Each task is consisted of one or more concepts composed of multiple attributes (2 or more). Options which can be displayed for an attribute are called levels.  Respondents evaluate concepts by rating or making choices among them. Those evaluations are collected by a researcher.

One may find that the origin of the adjective conjoint is that in study features/attributes of a concept are CONsidered JOINTly.  According to  Sawtooth Software, though, the word “conjoint” derives from the verb “to conjoin,” meaning “joined together.” (see history)

Idea

The idea behind conjoint analysis is, using attributes joined together we are much closer to real life choices/situation than by  surveying with attributes considered separately. For instance

  • Car attribute “driver extra side airbag” considered separately:

The car I would like to buy should have driver extra side airbag:

  1. I agree
  2. I don not agree

Then most of the respondents mark 1)

However..

  • Car attribute “driver extra side airbag” with two possible levels (driver extra side airbag /no driver extra side airbag ) considered jointly:

You are young single, your salary is below 1000 Eur per month, the car I would like to buy is:

conjoint question example

Example survey question with attributes considered jointly

Then the answer is not so obvious, one is forced to make difficult tradeoffs and it is much more real life situation, especially when it comes to buying products. In fact, there is not silver bullet car or any other product/service etc.

Origin

In 1964 mathematical psychologists and statisticians Luce and Tukey released a research  paper on Conjoint Measurment. I am not able judge that it was the first and only article on conjoint but experts from Sawtooth mention it as a significant one (again history). After several years marketing researchers borrowed that theory and henceforth the conjoint analysis started to evolve. In 1974 McFadden released a paper on econometrics related to choice behaviour. Upon this work, a new important type of conjoint analysis i.e. discrete choices was then built (again, according to Sawtooth Software) .

Thus, we can say shortly that the conjoint analysis has been in the market research world since early 70’s of 20th century.

Techniques

As you probably noticed it seeing Origin section, the conjoint analysis has been evolving for years and actually it does not denote only one technique for surveying. To be exact, we can divide types in many different ways. The following classification bases on the presentation (unfortunatley not present online) of experienced conjoint researcher –  Peter Kurz. In fact, It is not the same as in the mentioned slide show, however having much in common I am sure it is well founded.

  • First type is a Traditional Conjoint or Card-Sort method. Using it one concept per task is shown to a respondent. An each product/ service is consisted of all attributes that exist in study – such case is called Full profile A respondent is asked to rate a concept for instance in 0-100 scale. Example Card-Sort task:

Using a 100-pt scale where 0 means definitely would NOT and 100 means definitely WOULD…

How likely are you to purchase

card sort conjoint question example

Traditional conjoint question example

  • 2nd example within the Idea section above is a simple example of Choice Based Conjoint (CBC) or called interchangeably Discrete Choice Model (DCM), a type developed later than traditional one. Although CBC seems to be the most popular type in practice (based on my experience) it is important to notice we have scientists, involved in the topic, who have claimed that discrete choice experiments are not conjoint analysis at all (see Louviere Flynn Carson article from 2010). Judging it is surely far beyond the scope of this article and actually it would have required deeper analysis.  Let us assume that DCM is a reliable and valuable conjoint technique.
  • Apart from that we have Adaptive Conjoint Analysis (ACA, ACBC, etc.)  which idea behind is to adjust (during a survey) next tasks upon respondent previous answers in order to get more valuable output from individual or group of people.
  • There are also Hybrid conjoint methods combing described above, as well as, other market research techniques with the conjoint.

Choice based conjoint

Intuitively, discrete choice model seems to be best options in many real life cases, especially business ones like new product development. In DCM we have a target concept and competitors within a question. In many cases the None option, allowing respondent to quit, is also present. An each concept has attributes that describe it, in particular, a price and a brand. Hence, it is the same situation like standing behind the market shelf and trying to choose one of the products.

When we have many attributes in study then naturally not all attributes are displayed within a concept. Such design is called partial profile model. DCM allows marketer to conduct surveys in that way.

Also having many attributes and concepts per task one can use fractional factorial design – not all possible combinations of attributes within a concept are displayed during a study. Such approach is well documented.

Moreover, also number of attributes could vary among concepts since products or services, despite similarity, could have different features. Such designs are called Alternative Specific Designs (ASD).

Design could be static, in other words created manually or semi-manually by researcher, but also random – tasks are generated automatically by dedicated tool. Stochastic approach is particularly useful when we have a fractional factorial survey.  Additionally CBC can be done adaptively, what makes it even better considering collecting relevant output. Important work on that was presented in 2003 by Toubia, Hauser and Simester. Actually, there have been many followers on this article like Saure, Vielma in 2016 (refer to the one ), also popularity of artificial intelligence, machine learning or genetic algorithms might lead to obtaining of more properly adapted surveying in practice.

Sounds it is a perfect model for collecting the data. The question is why conjoint is not a standard and why it is not always utilized in market research … Well

Conjoint is not trivial

The conjoint analysis is an analysis and it is a market research process. As mentioned in Introduction such one is composed of stages. At each stage marketer has to make proper decisions and conduct a particular research according to relevant theories and former practice of conjoints. If you think the analysis is just an analysis, there is no question of an analysis. Actually the conjoint research is not trivial. What you have seen above is more like bird eye view of the analysis. By now they have been nothing on data analysis and Interpretation stages and a very little about technical aspects of data collection.

Problem definition and data collection

There are rules for defining attributes and their levels, as well as, number of them per concept. Then a marketer have to determine a number of tasks in study, a method of data collection and analysis. First of all, attributes have to be independent, an each level within is assumed to be mutually exclusive of the others i.e. respondent can choose only one level at the same time. Levels need to have unambiguous and concrete meaning for instance $2900  instead of very expensive. The usual number of levels per attribute is 3-5, but when it comes to a price or a brand attribute we could have much more… Again, describing all the rules and cases is far beyond the scope of this into article. Important is to see that a lot of topic have to be taken into consideration.

Decisions on a design affect a whole process, obviously. A marketer aims to get valuable data without confusing any respondent. Thus, one have to avoid presenting swarm of tasks or concepts or attributes per concept. On the other hand a researcher cannot end up with not sufficient output when number of tasks, attributes etc. is too small.

A data collection could be conducted by any technique, hence PAPI (Paper and Pencil Personal Interview) survey is possible, however using assistance of computer (CA), thus web surveys and interviewing personally or in closed groups sometimes via phone, seem to be the most appropriate options.

Below example CAWI (Computer assistance web interview) CBC task when a web browser is used to display questions, Each time only one task is displayed on a computer/tablet/phone screen, one choice feasible only:

CBC CAWI task

Example CBC CAWI task displayed in a web browser

Having data stored in hard disks of a computer it can be analysed then with help of dedicated statistics and market simulation software tools. Important decision here is to determine how many respondents need to take part in a survey in order to have sufficient data at output.

Although CA surveys seem to be the most convenient option, there are also problems. One can have broken connections, incorrect layout displays, or respondents who quits survey before it ends and therefore incomplete output and what is worse angry participants.

Data analysis and interpretation

After data is collected marketer chooses a statistical model in order to, first and foremost, know how valuable are attributes and their levels. Measures called utilities or part-worths or part-worth utilities tell about that. Depending on the project type and basing on a researcher experience and knowledge one can use one of many models. The Counts seems to be easiest and the most intuitive one, when proportions of wins for each level are calculated, based on how many times a concept including that level is chosen, divided by the number of times a concept including that level appeared in the task. When a researcher has a CBC study, then more complex, thus sophisticated, models like Latent Classes when segmentation is important, or popular Hierarchical Bayes (HB) estimation that operates on each individual data, but also Logit and others, can be utilized.

After that, using utilities a researcher can do further analysis, where simulators are helpful to get useful interpretation factors for instance preference shares, price sensitivity etc..

Again, many decisions have to be made by a researcher and deep statistical insight is required or at least very good command on estimation and simulation tools. Not only results but also designs, formerly displayed within a data collection stage, have to be analysed. For the example having similar concepts in a study, one should keep in mind the so called Red-Bus/Blue-Bus problem because otherwise, by choosing improper estimation model, they can improperly either decrease or increase an utility of a particular level. You can read more about it on the Internet for instance here

To summary it, yes it does. The conjoint is not trivial, since many mistakes can be made until a marketer finds a solution for a defined problem. Even the best market research tools do not do the whole job without researchers or perhaps in future complex intelligent machines.

Problem solution

Having respondent answers, utilities, and outcome from simulations marketer can present a suggested meaningful solution to a client. Then one can say that a conjoint is over.

Human factor

Last but not least, a human factor. Real respondents are not machines they are humans (at least assumed for today) and thus not only raw data is important also psychological aspects matter. Basic question is whether selecting a concept is truly the same situation as choosing from supermarket shelf, because actually one do not spend real money during a survey… Another thing is that depending on current own mood, a  respondent can make different, more spontaneous or on the opposite more conservative decisions etc. than usually. Those aspects are hard to measure correctly even if segmentation, extra questions are applied and correct decisions are made.

Unfortunately, as always, even the greatest conjoint outcome could be manipulated or presented selectively by a client or other third parties to meet their goals.

Summary

The conjoint analysis is a market research technique which is relevant to support making decisions regarding new product/service development. It could be applied in many areas both business or any others successfully. A respondent evaluates concepts consisted of attributes, displayed individually or among others. Although it seems to emulate real consumers decisions perfectly the conjoint is arduous and mistake sensitive process. The conjoint requires researches with great insight into statistics and market research in general, otherwise one can get wrong outcome. Hopefully, the development artificial intelligence and thus adaptive models can make the technique more adequate, reduces human factor troubles and number of failures possible during the process.

Share

Number plates in Greater Poland

Share

Introduction

Number plates that are currently in use in Poland firstly appeared in 2000. A new form of vehicle identification has been introduced due to a reform from 1999 regarding a new administrative division of the country. The regulations entered into force on 31st March 2000. Since then each new or newly registered vehicle has plates with a black type on a white reflective background. There are also temporary, special or vintage ones with different regulation on colours and patterns, however in this article I would like to focus on those the most common. All the things visible on each number plate, including type, colours, shapes, measures etc., are strictly defined within the mentioned act of law (unfortunately I couldn’t find the full act in english thus only original polish text in is linked here).

Content

Blue area

Starting from the left side of  plates we have blue field covering country code – “PL” and a flag of European Union or Polish one (second option just for those vehicle labels issued before 1st May 2006).

Place indicator

Then, on a number plate, we have either two or three capital letters that denote an administrative district of where the machine has been licensed. Two chars indicate urban, three rural one, however there are few exceptions for that rule (details later in this article). First among them always label a province of Poland. Greater Poland (wielkopolskie voivodeship) is denoted by “P” since it is a first letter of a capital and simultaneously the  biggest city of the province – Poznan. People usually register their vehicles in a department that is closest to their place of residence, thus briefly we could say that first part of a number plate shows from where a driver is from, however it’s not always the truth. It is so, as one can currently live in a different district than he used to. Formally, every change in a residence address should be registered at office, but often this is only a theory. We could also have company and other let’s say third-party cars, then a particular business or a lessor (i.e. not an user) licences vehicle in a place of its headquarters. Hence you could meet for instance a lot of vehicles with place indicator denoting Warsaw (the capital city of Poland) like “WD”, “WF”, “WX”, while driver lives and works in Greater Poland.

Remaining part

After that we have a space with a hologram and then vehicle indicator that is a permutation with repetition consisted of four or five digits and/or letters. A permutation can cover any digit and almost any capital letter from a latin alphabet – all apart from B, D,I ,0 ,Z to avoid ambiguities with digits 8, 0, 1, 0, 2 respectively. I must add here that a digit 0 is also not allowed for some positions. Actually, it is worth to mention here that rules for exact number of chars (either 4 or 5), letter-allowed positions and order of releasing within number plates are strictly defined.

Place indicator along with the remaining part is a number plate string that denotes a particular vehicle.

Number plate urban district

Sample draft of a number plate denoting urban district in Greater Poland voivodeship (the proportions of content are generally maintained but the width and height of characters modified intentionally)

Number plate rural district

Sample draft of a number plate denoting rural district in Greater Poland voivodeship (the proportions of content are somehow maintained, the width and height of characters modified intentionally)

Exceptions

As mentioned above, in Greater Poland, there are exceptions when it comes to number plates place indicators. Here there are:

  • PZ – two instead of three chars indicate rural area. It is so due to a large number of residents in that area, as it is around the city of Poznan. Formerly we have three letters indicator there – POZ – but we would have quickly run out of possible permutations consisted of just 4 chars in a remaining part
  • PP – two chars denote both the biggest city in this (north) part of Greater Poland and the area around it. Although city of Piła ((pronounce Peeua) could have been urban district separately, since 1999 it is formally rural one along with villages and small towns nearby.
  • PKO – three instead of two chars indicate urban district – a city of Konin, but strictly speaking they used to, initially. Currently, each new vehicle licensed therein has on its number plates two letters denoting the area  – PN

Common mistakes

  • PZ – denotes the city of Poznań – Not – only (but surely large and still growing though) area around the city.
  • PP – indicates only the city of PIła – Not – also small villages and towns around the city.
  • PSR – denotes town of Śrem with area around – Not – the town of Środa Wielkopolska along with area nearby
  • PKO – indicates Koło town with area around – Not – it used to the city of Konin (formerly, currently it’s PN)
  • PGO – denotes town of Gostyń with area around – Not – the town of Grodzisk Wielkopolski along with area around

Complete list of place indicators (in alphabetical order):

Urban districts:

PK – city of Kalisz

PKO, PN – city of Konin

PL – city of Leszno

PO, PY – city of Poznan (PY currently only very small group for instance motorbikes but other vehicles in future would have it as well)

Rural districts:

PCH – Chodzież (pronounce Hodjesh) town alongside small towns and villages around

PCT – towns  of Czarnków (pronounce Charnkoov) and Trzcianka (pronounce Tshcianka)  along with area around

PGN – Gniezno town and area around

PGS – town of Gostyń alongside small towns and villages around

PGO – Grodzisk Wielkopolski town along with area around

PJA – town of Jarocin and area around

PKA – area around the city of Kalisz

PKE – Kępno town along with small towns and villages around

PKL – town of Koło and area nearby

PKN – area around the city of Konin

PKS – Kościan town and area around

PKR – town of Krotoszyn alongside area around

PLE – area around the city of Leszno

PMI – Międzychód town and area around

PNT – town of Nowy Tomyśl along with area around

POB – Oborniki Wielkopolskie town alongside area around

POS – Ostrów Wielkopolski town and area neaarby

POT – town of Ostrzeszów alongside small towns and villages around

POZ, PZ – area around the city of Poznan

PP – city of Piła (pronounce Peeua) along with area around

PPL – Pleszew town and area around

PRA – town of Rawicz alongside area around

PSL – Słupca town and small towns and villages around

PSZ – town of Szamotuły and area around

PSR – Środa Wielkopolska town and area around

PSE – Śrem town along with area around

PTU – town of Turek  and area around

PWA – Wągrowiec town along with small towns and villages around

PWL – town of Wolsztyn (pronounce Volshtyn) and area around

PWR – Września town alongside area around

PZL – town of Złotów (pronounce Zwotoov) with area around

Share

Web 2.0 sophie

Share

Philosophising about various important issues is commonly ascribed to people that read a lot, are intelligent, well educated and experienced. Well, it seems to be truth, but are those human beings always wise?  Somehow, it is that a person who reflects some significant topics is worth of taking care, being listened, even in order to make a recipient of the philosophy willing to consider their life, restate views and own place in this world.  It is OK, however what if a target group get those topics to reflect while it hardly seem to require them at a particular moment?

Is not philosophising characteristic for people who have free time, though?  Even if  there is a person who works everyday as philosopher or one do it out of pure passion, still it is a matter of time. Although sacrifices have to be made or author has to give up their other life activities, surely the time frame on an axis with consecutively escaping seconds is needed, in order to introduce philosophy to mind. Even this post, I am writing now, will appear here directly because I stole some time from usually busy hours and definitely, it is better to me to release it today before I again reach hectic time, what inevitably will be happening more and more often.  Well, brilliant, please take no offence, I do not want to criticize, dismiss, insult anybody who philosophizes. Actually, now, I am doing nothing more than philosophy instead of getting to work, thus if so, it means that, first of all, I criticize myself.  Honestly, I am not even ironical now, on the contrary, since I partially mentioned it in a first part of this article, I think a moment of reflection, analysis or synthesis, looking from a different point of view, for many people, including me, is very helpful and moreover necessary from time to time, no matter which form of such routine break one chooses. A man is not a machine though. My aim here is rather to put forward a thesis that people who can afford to focus on philosophising are those who have some minutes free. Thus, it is a factor that definitely do not promote they so called men of action, who do not do the math, nor judge, nor reflect for hours, they just simply act as there is so much to do.

Now it is 2016, while they are many ways to access the global network, it is so easy to make somebody’s thoughts public. Mere one mouse click, one touchpad tap, one screen touch on smart phone tablet are truly sufficient. No matter if you are in Poland, Canada or in the north part of Ethiopia, sitting and sipping a cheap wine or trying on a brand new suit (undoubtedly, fashionable navy).  Unconditionally, without queues, without special preparation, right now everybody, who wants to, can be like Hume, Kant or Heller. Despite the obvious fact that such universality and speed have disadvantages and downsides, it is a really great opportunity. People share their thoughts with close relatives, at family parties, at work, at school, with more formal or informal groups and communities they attend to, but also on the Internet. Would any of such currently unknown thinkers issue a book with philosophies at famous publishing house X or an philosophical article at bestselling newspaper Y, if it was possible? Simultaneously, would the one agree on having that hypothetical book on the same shelf as Nietzsche’s “Antichrist” or on publishing, taken out of context snippets of their thoughts, in an infamous tabloid? Questions of that kind, decisions to be made, possible consequences to analysis would definitely appear then. Unfortunately, since one uploads philosophy solely on-line and on their own, a machine called the Internet, i.e. people that co-create it, may not put them. Consequently the publication might start living its own, even a brand new, life.

For a creator misunderstanding, or bad opinion, hate, seem to be the worst repercussions after thoughts are issued. Some of the authors include in their publications statements like “This is only my humble opinion” or even “I do not care what other people say about it, at all”. Yes, but the question is how much humility and thick skin, correspondingly, it represents actually or rather the statements appear in the context because these philosophers will is to protect themselves against unwanted reactions of readers or they aim to show they are superior to all potential haters? The question is if it is possible to influence on all the possible future reactions and opinions? Furthermore, referring to the saying that has been known since early ages, although silence is golden, speech is still silver, therefore one can ask if in fact the worst thing for a creator is not wrong reaction of audience but whether there is not even a particle of this proverbial silver within philosophy – timeless, just small value?

Share

Visit Bługowo

Share

Polish Bługowo (pronounce Bwoogovo w as in word, g as in group) is a small village located in north part of Greater Poland. Driving along its main road apart from picturesque landscapes and then after the old Cross of Christ and several both old and modern houses you will see an old fashioned green metal fence with a gate in the middle of it and concrete stairs in a front of the entrance. Don’t be discouraged by this wire fencing just pass it if a gate is open, because a building you would see is a pearl of architecture definitely. The old red brick church is more visible from the road since autumn until early spring, while in late spring and during the summer is a little hidden by widely spreading trees in a front.

Bługowo church

The church in late spring visible behind the wire fencing.

Saint James Apostle’s (son of Zebeedee) the church is also located (exactly 53°15′ N 17°07′ E) very close to the cross-shaped Sławianowskie lake (pronounce Swavianovskie w as in word, i as in big). Moreover, the lake is also called Wielkie (pronounce Vielkie) – what means a great one, as it’s one of the biggest in this part of Poland. Briefly the location is indeed attractive for visitors.

History

The church in its current form was built in 1864-1869 A.D in Neo-Romanesque style. In 1953 both the interior was redecorated substantially and electricity was installed. Then in 2000 the whole interior was refreshed and renovated. Although the building was raised in 19th century, in chronicles, one can find mention even from 1349 suggesting that a parish and also from 1445 that a church were present in the village.

blugowo_church_interior_view

A view on the main nave of the chuch,

Nowdays

The church is a parish for two villages Bługowo and Kunowo (pronounce Koonovo). I have had a chance to be in the described place many times but the one on 26th May 2016 was actually a special one for me. There had been a special Holy Mass and after it we went outside, then through the village, decorated with religious symbols and small birches,  we attended religious procession. Such processions take place every year on Thursday – exactly 60 days after Easter all over the Poland. Corpus Christi, day, as it’s called, has strong catholic tradition in this country thus is a bank holiday.

I encourage you to visit Bługowo, the church and surrounding area any time, however especially on Corpus Christi day, to feel the taste of Polish village. Surely, that experience would be unique and to remember.

Bługowo church

A view on the main altar of the church.

blugowo_church_main_entrance

A main entrance of the church.

blugowo_church_side_entrance

A side entrance and a back wall of the church

Share

Hello world!

Share

Briefly or comprehensively?

How to start this webpage briefly? I haven’t made any holistic statistic research on short forms of content but it wouldn’t be an exaggeration to claim  that currently  short forms are very popular and one video spot, one picture can say more than thousand of words.  Just take a look around – Facebook, Twitter, Youtube, banner ads, catchy titles on information services. To be simple and effective says one of the advertisement I have now on my desk. I don’t want to make here psychological or social analysis on that tendency as there are many other wiser people who have already successfully done concise research considering such topic. I just think that despite we commonly accept and utilize short forms we’re honestly already fed up with them. Unfortunately we have no choice, as there is too much articles to read, too much videos to watch, too much duties. Hence I predict that we’re about to start a new chapter of communication forms and artificial intelligence will be a key player in background…

OK, but we’re here we’re now, we have own mission. Moreover, our influence on political or big company decisions is very limited, that’s why I eventually started my small web place today and I hope that you dear visitor will find here something that is somehow valuable for you or at least you enjoy it.

Share