Comparative methods of scaling - Marketing research

7.3. Comparative scaling methods

We examined the types of scales used in marketing. The next problem is how to get the information necessary for the scaling from the respondents? From this point of view, the methods of scaling are divided into two groups: comparative and incomparable.

With comparative scaling, respondents compare the proposed objects with each other. For example, what drink for the respondent is preferable: "Coca-Cola" or Pepsi-Cola & quot ;, kvass or Coca-Cola & quot ;, Pepsi-Cola or kvass?

The advantages of comparative methods are as follows .

First, they can detect even small differences in preferences.

Secondly, the respondent makes a comparison, based on criteria known only to him. Therefore, the rules of comparative scaling are easy to explain to the respondent, and the results are easy to apply.

Thirdly, because the researcher does not impose his criteria on respondents, the possibility of transferring estimates made by one criterion to estimates made by another criterion is also reduced.

Comparative scaling methods have disadvantages .

First, they make it possible to work only with an ordinal scale.

Secondly, they allow you to receive estimates of only those objects that were included in the list in the study. If a new object has appeared (for example, dairy products Large mug "), it is not enough to evaluate it only, but you need to re-conduct the entire study.

Third, and most importantly, the fact that the respondent prefers brand A to brand B does not mean that he is good with brand A. For example, you might prefer some sort of chocolate to another, but not love (and therefore do not buy) chocolate.

In view of these shortcomings in marketing, non-comparable methods of scaling are more often used. It is assumed that they provide information about the position of objects on an interval or proportional scale. With non-equal scaling, the respondent is asked to rate each object separately, for example, on a six-point scale.

Let's start with comparative methods. These include methods: pairwise comparison, ordering, scaling with a constant sum, Q-sorting and some others.

The method of pairwise comparisons. The respondent is presented with two objects and is asked to choose the most preferable ones according to his own criteria. Often they give him something to actually look at, touch, sniff, try, etc. To study the respondent's attitude to N objects, it is necessary for the respondent to do pairwise comparisons. After this, the tournament table is compiled, the number of times respondents choose brand A for brand B. The objects are placed on the scale in accordance with the number of entries in this tournament points.

Sometimes after this, procedures are applied that make it possible to move from the ordinal scale to some analogue of the interval scale. Without going into details, we note that the idea of ​​obtaining an interval scale is akin to determining the rating of chess players or tennis players. When calculating this rating, not only the number of player wins is taken into account, but also the strength of the rivals defeated by it. The calculation is performed iteratively. First, every player for each victory one point is awarded. Then the calculation is repeated, and for each victory the player is credited with the amount proportional to the number of points awarded to the opponent he defeated in the previous step of the algorithm. The procedure continues iteratively until the next iteration almost does not change the players' ratings [13; 31].

One of the authors developed special polling procedures and a data processing program that implements the procedure described above. With her help, the results of the survey commissioned by the company, which is one of the largest producers of sweets in the world, were processed. In this survey, each respondent was given three candies according to a special plan and was asked to compare them - first in appearance and then in taste. As a result, the ratings sweets both among all those surveyed and in individual socio-demographic groups.

Using the same procedure, the data of several other specially organized representative polls of the country's population conducted by the Foundation Public Opinion were processed.

The pairwise comparison method has modifications. So, sometimes the respondent is allowed to say that, from his point of view, some two objects are equivalent or that it is difficult to compare them. And sometimes the respondent is asked to assess the degree of his preference: for example, how much more he would be willing to pay for a brand, given his advantage over another brand and the cost of the latter.

The pairwise comparison method has two serious drawbacks:

First, the method works pure If the respondents' opinions are transitive, i.e. if brand A is rated higher than brand B, and brand B is higher than brand C, then brand A must necessarily be rated higher than brand C. If there is no transitivity, we can get a distorted view of the "power ratio" between brands, not noticing partial preferences.

Secondly, more importantly, the pairwise comparison situation itself is far from life. So, usually buyers choose not from two, but at once from several available goods. Therefore, not always each of them chooses exactly the goods that he would prefer to all the rest, he conducts thoughtful pairwise comparisons. In addition, you can prefer some brand of goods to its competitors, but do not like and do not buy all the goods of this type. Failure to understand this lack of paired comparisons has led one of the United States research firms to a serious error in predicting the outcome of parliamentary elections. It turned out to be very significant that there are actually two parties in the voting ballot, and several, and that the voter can refuse to participate in the elections or vote against all.

Ordering. Respondents are offered to order not a pair, but many objects at once: assign the value 1 to the most preferable object, the next to 2, and so on. However, one drawback of the method of pairwise comparisons is inherent in this method: one can not love and accordingly do not buy the most preferable object. On the other hand, ranking has advantages over a pairwise comparison.

First, the process of ordering a set of brands is more like a real multivariate choice.

Secondly, when ordering there is no danger of nontransitive responses appearing and, therefore, it is always possible to get a fully ordered scale of estimates.

Using the ordering method, as well as the method of pairwise comparisons, there are procedures that allow you to move from the ordinal scale to the interval counter.

The distribution of the constant amount. Respondents are asked to give each of the objects in the list of points in such a way that their amount is equal to a certain number, for example, 100 points. In one study in the US, representatives of three different customer segments distributed 100 points between eight different criteria for selecting soap on average as follows (Table 7.1).

Table 7.1. Importance of criteria for selecting soap for representatives of three consumer segments, points

Soap selection criteria

Segment 1

Segment 2

Segment 3

Softness

8

2

4

Foaminess

2

4

17

Fast drying

3

9

7

Price

53

17

9

Fragrance

9

0

19

Appearance of the package

7

5

9

Moisturizing capacity

5

3

20

Washing power

13

60

15

Total points

100

100

100

The method has important advantages.

First, these data are closer to the interval scale: here the respondents themselves give information about the magnitude of the differences they perceive between the evaluated objects.

Secondly, it is easier for a respondent to distribute 100 units than to arrange objects.

Now about the drawbacks.

First, the estimates of the objects included in the study become obsolete as soon as an important object appears on the market. This object can not be evaluated separately from others; you need to repeat the whole study.

Secondly, for respondents, research by this method is still a hard and tedious work.

Thirdly, respondents often make mistakes, receiving amounts, say, 108 or 94 units. Therefore, the answers of each respondent have to be further transformed so that their sum is equal to 100 points.

Q-Sort (Q-Sort) The method is as follows. The respondent is given cards with statements about something and they are asked to put them into piles: for example, 60-140 (but preferably not more than 90) cards - for 11 or other odd number of stacks. Stacks are commonly referred to as: most agree & quot ;, also highly agree and so on. The approximate number of cards in each pile is pre-determined. It is established, being guided by the form of normal distribution.

thematic pictures

Also We Can Offer!

Other services that we offer

If you don’t see the necessary subject, paper type, or topic in our list of available services and examples, don’t worry! We have a number of other academic disciplines to suit the needs of anyone who visits this website looking for help.

How to ...

We made your life easier with putting together a big number of articles and guidelines on how to plan and write different types of assignments (Essay, Research Paper, Dissertation etc)